研究者詳細 - 小田　昌宏

Classification of fundus autofluorescence images based on macular function in retinitis pigmentosa using convolutional neural networks Open Access

Semantic segmentation dataset authoring with simplified labels 査読有り Open Access

Leo Uramoto, Yuichiro Hayashi, Masahiro Oda, Takayuki Kitasaka, Kensaku Mori

International Journal of Computer Assisted Radiology and Surgery 2025年4月

Development of a YOLOv3-Based Model for Automated Detection of Thoracic Ossification of the Posterior Longitudinal Ligament and the Ligamentum Flavum on Plain Radiographs 査読有り Open Access

Sadayuki Ito, Hiroaki Nakashima, Naoki Segi, Jun Ouchida, Ippei Yamauchi, Takashi Hirai, Masahiro Oda, Kensaku Mori, Masashi Yamazaki, Toshitaka Yoshii, Shiro Imagama

Journal of Clinical Medicine 14 巻 ( 7 ) 2025年3月

Bidirectional teaching between lightweight multi-view networks for intestine segmentation from CT volume 査読有り Open Access

Qin An, Hirohisa Oda, Yuichiro Hayashi, Takayuki Kitasaka, Aitaro Takimoto, Akinari Hinoki, Hiroo Uchida, Kojiro Suzuki, Masahiro Oda, Kensaku Mori

Journal of Medical Imaging 12 巻 ( 2 ) 2025年3月

Enhanced self-supervised monocular depth estimation with self-attention and joint depth-pose loss for laparoscopic images 査読有り Open Access

Wenda Li, Yuichiro Hayashi, Masahiro Oda, Takayuki Kitasaka, Kazunari Misawa, Kensaku Mori

International Journal of Computer Assisted Radiology and Surgery 2025年2月

Multi-dimensional consistency learning between 2D Swin U-Net and 3D U-Net for intestine segmentation from CT volume 査読有り Open Access

Qin An, Hirohisa Oda, Yuichiro Hayashi, Takayuki Kitasaka, Hiroo Uchida, Akinari Hinoki, Kojiro Suzuki, Aitaro Takimoto, Masahiro Oda, Kensaku Mori

International Journal of Computer Assisted Radiology and Surgery 2025年2月

Kominami, T; Ueno, S; Ota, J; Inooka, T; Oda, M; Mori, K; Nishiguchi, KM

JAPANESE JOURNAL OF OPHTHALMOLOGY 2025年2月

記述言語：英語出版者・発行元：Japanese Journal of Ophthalmology

Purpose: To determine whether convolutional neural networks (CNN) can classify the severity of central vision loss using fundus autofluorescence (FAF) images and color fundus images of retinitis pigmentosa (RP), and to evaluate the utility of those images for severity classification. Study design: Retrospective observational study. Methods: Medical charts of patients with RP who visited Nagoya University Hospital were reviewed. Eyes with atypical RP or previous surgery were excluded. The mild group was comprised of patients with a mean deviation value of > − 10 decibels, and the severe group of < − 20 decibels, in the Humphrey field analyzer 10-2 program. CNN models were created by transfer learning of VGG16 pretrained with ImageNet to classify patients as either mild or severe, using FAF images or color fundus images. Results: Overall, 165 patients were included in this study; 80 patients were classified into the severe and 85 into the mild group. The test data comprised 40 patients in each group, and the images of the remaining patients were used as training data, with data augmentation by rotation and flipping. The highest accuracies of the CNN models when using color fundus and FAF images were 63.75% and 87.50%, respectively. Conclusion: Using FAF images may enable the accurate assessment of central vision function in RP. FAF images may include more parameters than color fundus images that can evaluate central visual function.

DOI： 10.1007/s10384-025-01163-w

Artificial intelligence support improves diagnosis accuracy in anterior segment eye diseases 招待有り査読有り

Hiroki Maehara, Yuta Ueno, Takefumi Yamaguchi, Yoshiyuki Kitaguchi, Dai Miyazaki, Ryohei Nejima, Takenori Inomata, Naoko Kato, Tai-ichiro Chikama, Jun Ominato, Tatsuya Yunoki, Kinya Tsubota, Masahiro Oda, Manabu Suzutani, Tetsuju Sekiryu, Tetsuro Oshika

Scientific Reports 15 巻 2025年2月

Double-mix pseudo-label framework: enhancing semi-supervised segmentation on category-imbalanced CT volumes Open Access

Zhang, LY; Hayashi, Y; Oda, M; Mori, K

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY 2025年2月

記述言語：英語出版者・発行元：International Journal of Computer Assisted Radiology and Surgery

Purpose: Deep-learning-based supervised CT segmentation relies on fully and densely labeled data, the labeling process of which is time-consuming. In this study, our proposed method aims to improve segmentation performance on CT volumes with limited annotated data by considering category-wise difficulties and distribution. Methods: We propose a novel confidence-difficulty weight (CDifW) allocation method that considers confidence levels, balancing the training across different categories, influencing the loss function and volume-mixing process for pseudo-label generation. Additionally, we introduce a novel Double-Mix Pseudo-label Framework (DMPF), which strategically selects categories for image blending based on the distribution of voxel-counts per category and the weight of segmentation difficulty. DMPF is designed to enhance the segmentation performance of categories that are challenging to segment. Result: Our approach was tested on two commonly used datasets: a Congenital Heart Disease (CHD) dataset and a Beyond-the-Cranial-Vault (BTCV) Abdomen dataset. Compared to the SOTA methods, our approach achieved an improvement of 5.1% and 7.0% in Dice score for the segmentation of difficult-to-segment categories on 5% of the labeled data in CHD and 40% of the labeled data in BTCV, respectively. Conclusion: Our method improves segmentation performance in difficult categories within CT volumes by category-wise weights and weight-based mixture augmentation. Our method was validated across multiple datasets and is significant for advancing semi-supervised segmentation tasks in health care. The code is available at https://github.com/MoriLabNU/Double-Mix.

DOI： 10.1007/s11548-024-03281-1

A deep learning-based prediction model for prognosis of cervical spine injury: a Japanese multicenter survey 査読有り

Sadayuki Ito, Hiroaki Nakashima, Naoki Segi, Noriaki Yokogawa, Takeshi Sasagawa, Toru Funayama, Fumihiko Eto, Akihiro Yamaji, Kota Watanabe, Satoshi Nori, Kazuki Takeda, Takeo Furuya, Atsushi Yunde, Hideaki Nakajima, Tomohiro Yamada, Tomohiko Hasegawa, Yoshinori Terashima, Ryosuke Hirota, Hidenori Suzuki, Yasuaki Imajo, Shota Ikegami, Masashi Uehara, Hitoshi Tonomura, Munehiro Sakata, Ko Hashimoto, Yoshito Onoda, Kenichi Kawaguchi, Yohei Haruta, Nobuyuki Suzuki, Kenji Kato, Hiroshi Uei, Hirokatsu Sawada, Kazuo Nakanishi, Kosuke Misaki, Hidetomi Terai, Koji Tamai, Akiyoshi Kuroda, Gen Inoue, Kenichiro Kakutani, Yuji Kakiuchi, Katsuhito Kiyasu, Hiroyuki Tominaga, Hiroto Tokumoto, Yoichi Iizuka, Eiji Takasawa, Koji Akeda, Norihiko Takegami, Haruki Funao, Yasushi Oshima, Takashi Kaito, Daisuke Sakai, Toshitaka Yoshii, Tetsuro Ohba, Bungo Otsuki, Shoji Seki, Masashi Miyazaki, Masayuki Ishihara, Masahiro Oda, Seiji Okada, Shiro Imagama, Satoshi Kato

European Spine Journal 2025年2月

The importance of clinical experience in AI-assisted corneal diagnosis: verification using intentional AI misleading 査読有り

Scientific Reports 15 巻 ( 1462 ) 2025年1月

Content-consistent adaptation network for substantia nigra segmentation in cross-modalities and -datasets scenarios without retraining

Hu T., Itoh H., Oda M., Saiki S., Hattori N., Kamagata K., Sako W., Ishikawa K.I., Aoki S., Mori K.

Progress in Biomedical Optics and Imaging - Proceedings of SPIE 13406 巻 2025年

出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

Fully automated segmentation of substantia nigra (SN) is crucial for developing the interpretable computer-aided diagnosis system for Parkinson's disease (PD). In real-world applications, the segmentation models may be trained with high quality medical images, while the testing dataset may be scanned with an older machine and is of relatively lower quality. To address this domain shift problem, conventional domain adaptation methods propose various strategies to adapt the segmentation model to the testing data. These conventional adaptation methods typically include a retraining stage in the target domain, either for the segmentation model or a translation model. This process makes the predictions in real multi-modalities and -datasets scenarios highly time-consuming. To enable rapid SN segmentation in cross-modalities and -datasets tasks, this paper proposes a content-consistent adaptation network using only source domain data to synthesize training data with the consistent contents of the target domain. The proposed method requires no retraining during the testing stage, thus enabling rapid SN segmentation across modalities and datasets. The results indicated the competitive performances of our proposed method in both the cross-modalities and -datasets tasks.

DOI： 10.1117/12.3047729

CLSMI2T3: 3D CLSM Vasculature Volume Reconstruction from A Single 2D Slice by Off-Focal Plane Signal Using Synthetic Data

Wu Y., Menten M.J., Kreitner L., Kawamura S., Oda M., Hayashi Y., Takebe T., Rueckert D., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 15371 LNCS 巻頁： 108 - 118 2025年

出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

In this paper, we propose a novel framework for reconstructing 3D Confocal Laser Scanning Microscopy (CLSM) vasculature volumes from single 2D slices using synthetic and real CLSM data. CLSM is an imaging technique commonly used for in-vivo 3D imaging. However, the utility of 3D CLSM imaging is limited due to long acquisition times and phototoxicity caused by prolonged light exposure. We propose a novel solution to reconstruct a 3D volume from a 2D slice by off-focal plane signal using deep learning (DL). However, the limited availability of real 3D data has to be overcome to train such a 3D reconstruction model. Therefore, we propose a novel framework to effectively address the lack of 3D data for training a robust and accurate 3D reconstruction model by simulating and synthesizing CLSM volumes. Our framework consists of (1) a 3D CLSM volume simulation and synthesizing for training data and (2) a 3D CLSM volume reconstruction part. Experiments show that we can effectively reconstruct 3D CLSM vasculature volumes from single slices, even only using synthetic data. Furthermore, we will also release our CLSM Image of Vascular Organoid for REConstruction (CLSMI-VO-REC) dataset and code at https://github.com/MoriLabNU/3D_CLSM_Volume_Reconstruction.

DOI： 10.1007/978-3-031-77786-8_11

A Deployable Microscopic Image Segmentation Look-Up Table Based on A Dilated CNN

Wu Y., Pan J., Kawamura S., Oda M., Hayashi Y., Takebe T., Rueckert D., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 15371 LNCS 巻頁： 3 - 13 2025年

出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Microscopic image segmentation is indispensable biomedical research. However, accurately segmenting using traditional automatic segmentation methods remains challenging due to fluorescence presenting high intra- and inter-image variability, hindering biomedical analysis. Deep learning (DL)-based methods show remarkable potential in microscopic image segmentation. However, DL models typically require GPU resources, specialized programming skills, and a non-trivial environment setup, making it difficult for medical researchers to use them currently. Therefore, we propose a turnkey solution for microscopic image segmentation by Look-up Table (LUT) which is a data structure used for storing input values and pre-calculated results. Our solution can inherit the segmentation accuracy of DL-based methods and eliminates the need for extensive computational resources, specialized programming, and a complex implementation environment. Specifically, (1) By transferring all potential inputs and their DL model-computed outputs into a LUT, segmentation is achieved through direct output look-up for given pixel values. (2) Additionally, we develop a dilated CNN with adaptive edge padding, enhancing feature learning while meeting LUT building requirements. (3) Moreover, we publicly release a confocal microscopic image dataset and develop a LUT-based plugin (SEG-LUT plugin) for Fiji ImageJ. The experimental results show our framework outperforms other automatic segmentation methods in the Fiji ImageJ. Our solution offers a pathway for easily utilizing DL-based methods for microscopic image segmentation. Our codes, plugin, and dataset are available at https://github.com/MoriLabNU/Microscopic_Image_Segment_LUT.

DOI： 10.1007/978-3-031-77786-8_1

Automated Skill Evaluation Method for Surgical Images using Optical Flow and Attention Mechanism

Feng Y., Hayashi Y., Oda M., Kitasaka T., Yasui A., Shirota C., Uchida H., Mori K.

Progress in Biomedical Optics and Imaging Proceedings of SPIE 13408 巻 2025年

出版者・発行元：Progress in Biomedical Optics and Imaging Proceedings of SPIE

This paper describes an automated skill evaluation method for surgical images based on optical flow and self-attention mechanisms. This method uses a segment of a surgical video as input to a deep learning network and outputs scores. Previous methods primarily analyzed the kinematic data of surgical robots to assess surgical skills; however, kinematic data is often unavailable in non-robotic surgeries. Therefore, we propose a regression model that uses surgical video as input data to evaluate surgical skills through optical flow and self-attention mechanisms. We use optical flow to extract motion information from surgical images and employ the ViViT (Video Vision Transformer) model to extract features from both image information and motion information. Finally, we concatenate the two feature vectors and output six scores corresponding to the GRS (Global Rating Scale) evaluation criteria. To validate the evaluation performance of the model, we calculated the RMSE (Root Mean Squared Error) between the model's output scores of surgical videos and the ground truth scores in the surgical skill assessment dataset JIGSAWS. A low RMSE of 0.79 demonstrated that our method can reliably estimate scores, highlighting its potential for surgical skill assessment.

DOI： 10.1117/12.3049068

Enhanced Sequential Picking Strategy with Hyperparameter Optimization for Efficient Surgical Phase Recognition

Yao G., Hayashi Y., Oda M., Misawa K., Mori K.

Progress in Biomedical Optics and Imaging Proceedings of SPIE 13408 巻 2025年

出版者・発行元：Progress in Biomedical Optics and Imaging Proceedings of SPIE

Accurate surgical phase recognition from surgical videos is critical for surgical training and evaluation. However, traditional models like Transformers face limitations in memory and computational efficiency, particularly as input sequence length increases. We introduce an enhanced Sequential Picking Strategy with hyperparameter optimization within the SPT-SVNet model, which integrates spatial and temporal embeddings. Through hyperparameter search, we apply different sequential picking strategies by optimizing these hyperparameters, enabling the model to selectively extract crucial features from temporal embeddings and reduce computational demands to varying degrees. Optimizing these hyperparameters led to significant performance improvements, reducing memory usage by 8.6% and training time by 16.2%, while maintaining or even enhancing model accuracy. We evaluated our approach using the SPT-SVNet model. Evaluations on the Cholec80 dataset demonstrated remarkable results, with our method achieving an accuracy of 90.5% and a precision of 92.4%, surpassing existing models.

DOI： 10.1117/12.3047731

Estimation of hidden blood vessel locations in laparoscopic videos by combining preoperative and intraoperative information

Enomoto K., Hayashi Y., Misawa K., Kitasaka T., Oda M., Mori K.

Progress in Biomedical Optics and Imaging Proceedings of SPIE 13408 巻 2025年

出版者・発行元：Progress in Biomedical Optics and Imaging Proceedings of SPIE

This paper presents a method for estimating hidden blood vessel locations in laparoscopic videos by combining preoperative and intraoperative information. In many laparoscopic surgeries, surgeons need to cut blood vessels. However, because the blood vessels are often hidden by adipose tissues and other anatomical structures, direct observation may not be possible. A computer-aided surgery system that reveals the locations of these concealed blood vessels could significantly assist surgeons during laparoscopic surgery. Therefore, we develop a method for estimating the hidden blood vessels in laparoscopic videos. The proposed method estimates the location of the hidden blood vessels by utilizing both the preoperative information and the intraoperative laparoscopic videos. We apply fully convolutional networks (FCNs) to each frame of the laparoscopic videos to estimate these locations. Our FCN model utilizes attention mechanisms to incorporate the preoperative information. The preoperative information for the first frame is derived from preoperative CT volumes. In the following frames, the preoperative information is deformed using the optical flow among successive two frames. We tested the proposed method on seven videos recorded during robot-assisted laparoscopic gastrectomies for gastric cancer. Experimental results demonstrated that the proposed method successfully estimated the locations of hidden blood vessels in laparoscopic video. The results also highlight the benefits of integrating preoperative and intraoperative data.

DOI： 10.1117/12.3047525

Prior knowledge-based blood vessel reconstruction and thrombus visualization using confocal laser scanning microscopic images

Wang, C; Hayashi, Y; Oda, M; Kawamura, S; Takebe, T; Mori, K

MEDICAL IMAGING 2025: CLINICAL AND BIOMEDICAL IMAGING 13410 巻 2025年

出版者・発行元：Progress in Biomedical Optics and Imaging Proceedings of SPIE

This paper describes an algorithm for the reconstruction of 3D blood vessel (BV) structure and thrombus. Biologists use a confocal laser scanning microscope (CLSM) to observe the brain of a mouse. However, the CLSM images obtained are in 2D, which limits the understanding of the 3D structure. In a previous method, the 3D BV structure was reconstructed from the centerline of the BV extracted from CLSM images. However, the reconstructed 3D BV structure is not realistic due to the influence of image noise. Therefore, we apply a smooth operation to the centerline of BV to make the reconstructed 3D BV structure more realistic. Meanwhile, we segment the thrombus from the CLSM images and visualize it within the reconstructed 3D BV. The 3D BV structure reconstruction method consists of three steps: (1) BV region segmentation, (2) centerline calculation and 3D mapping, and (3) 3D BV structure reconstruction. The 3D thrombus is reconstructed based on the reconstructed BV shape. Experimental results showed that the BVs obtained by the proposed method were smoother and more realistic compared to the previous method. In addition, the proposed method displayed the reconstructed thrombus in 3D space together with the BV, allowing 3D viewing from any viewpoint.

DOI： 10.1117/12.3048339

SemiOrth: A Novel Orthogonal Dual Network Architecture for Enhanced Semi-Supervised Medical Image Segmentation

Zhang, LY; Oda, M; Mori, K

MEDICAL IMAGING 2025: IMAGE PROCESSING 13406 巻 2025年

出版者・発行元：Progress in Biomedical Optics and Imaging Proceedings of SPIE

Medical image segmentation is crucial for Computer-Aided Diagnosis in healthcare. However, the confidentiality of medical data and the difficulty of creating segmentation annotations require training in high-precision models with limited annotated data. This study introduces a novel co-model training framework that uses an orthogonality loss function to enhance the heterogeneity between two models, enabling them to focus on different target features more effectively. Additionally, we employ a hybrid approach to refine pseudo-labels by evaluating predictions from the same image under varying augmentations, improving their accuracy. Tested on the Automated Cardiac Diagnosis Challenge dataset, our methodology achieved competitive performance with only 8% labeled data, underscoring its efficacy in optimizing medical image segmentation with minimal annotations.

DOI： 10.1117/12.3047175

Uncertainty-Aware Semi-Supervised Learning for Enhanced Multi-Organ Segmentation in CT Volumes

Dong, G; Oda, M; Hayashi, Y; Misawa, K; Mori, K

MEDICAL IMAGING 2025: IMAGE PROCESSING 13406 巻 2025年

出版者・発行元：Progress in Biomedical Optics and Imaging Proceedings of SPIE

We propose an uncertainty-aware semi-supervised learning (SSL) strategy for multi-organ segmentation in CT volumes. CT volume segmentation is crucial for enhancing diagnostic accuracy, facilitating quantitative analysis, and enabling automation and assisted diagnosis. Previous SSL strategies have shown great performance on CT volumes by ensuring consistency between different outputs of the model. However, these methods do not address the noise and uncertainty within individual outputs, which can impact the results. Our method integrates uncertainty estimation into the SSL framework, dynamically generating masks based on the uncertainty of individual predictions. This approach not only maintains consistency between multiple outputs but also mitigates the impact of noise and uncertainty within each prediction. Under the 10% labeled data setting, comparative experiments demonstrated that our method improved segmentation performance across effectively segmented organs. Specifically, compared to the baseline, the Dice scores increased by 2.50% for the vein, 1.99% for the liver, 5.16% for the spleen, and 6.09% for the stomach, indicating the potential benefits of incorporating uncertainty estimation into the SSL framework.

DOI： 10.1117/12.3047155

WPDM: Enhancing Subarachnoid Hemorrhage CT Detection with Worley-Perlin Noise Diffusion on Imbalanced Data

Lu, ZY; Hu, T; Oda, M; Hayashi, Y; Watadani, T; Mori, K

MEDICAL IMAGING 2025: COMPUTER-AIDED DIAGNOSIS 13407 巻 2025年

出版者・発行元：Progress in Biomedical Optics and Imaging Proceedings of SPIE

We propose a diffusion model based on Worley-Perlin noise, which provides higher robustness for image synthesis problems with striated textures, e.g. subarachnoid hemorrhage (SAH) CT slices. Unlike Gaussian noise, which uniformly perturbs images and causes the generated samples to be overly homogeneous, or Simplex noise, which excessively disrupts images and leads to distortion, our method accounts for the morphological structure of textured images. It ensures effective perturbation while avoiding image distortion and generates high-quality samples for downstream tasks. To validate the effectiveness of our proposed generative model in improving SAH CT detection performance on imbalanced datasets, we incorporated the generated SAH samples into the training process and compared the results with those of models trained on the original dataset. The experimental results demonstrated that our synthesized data can help improve the classification accuracy in an imbalanced data situation, with the F1-score improving by nearly 6.4%.

DOI： 10.1117/12.3047558

【放射線科と人工知能】脊椎・脊髄疾患の画像診断におけるAI利用の現状

Preliminary study of substantia nigra analysis by tensorial feature extraction 査読有り

Itoh, H; Oda, M; Saiki, S; Kamagata, K; Sako, W; Ishikawa, KI; Hattori, N; Aoki, S; Mori, K

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY 19 巻 ( 11 ) 頁： 2133 - 2142 2024年11月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：International Journal of Computer Assisted Radiology and Surgery

Purpose: Parkinson disease (PD) is a common progressive neurodegenerative disorder in our ageing society. Early-stage PD biomarkers are desired for timely clinical intervention and understanding of pathophysiology. Since one of the characteristics of PD is the progressive loss of dopaminergic neurons in the substantia nigra pars compacta, we propose a feature extraction method for analysing the differences in the substantia nigra between PD and non-PD patients. Method: We propose a feature-extraction method for volumetric images based on a rank-1 tensor decomposition. Furthermore, we apply a feature selection method that excludes common features between PD and non-PD. We collect neuromelanin images of 263 patients: 124 PD and 139 non-PD patients and divide them into training and testing datasets for experiments. We then experimentally evaluate the classification accuracy of the substantia nigra between PD and non-PD patients using the proposed feature extraction method and linear discriminant analysis. Results: The proposed method achieves a sensitivity of 0.72 and a specificity of 0.64 for our testing dataset of 66 non-PD and 42 PD patients. Furthermore, we visualise the important patterns in the substantia nigra by a linear combination of rank-1 tensors with selected features. The visualised patterns include the ventrolateral tier, where the severe loss of neurons can be observed in PD. Conclusions: We develop a new feature-extraction method for the analysis of the substantia nigra towards PD diagnosis. In the experiments, even though the classification accuracy with the proposed feature extraction method and linear discriminant analysis is lower than that of expert physicians, the results suggest the potential of tensorial feature extraction.

DOI： 10.1007/s11548-024-03175-2

その他リンク： https://link.springer.com/article/10.1007/s11548-024-03175-2/fulltext.html

伊藤定之, 中島宏彰, 世木直喜, 大内田隼, 小田昌宏, 山内一平, 大出幸史, 長谷康弘, 岡田裕也, 鏡味佑志朗, 森健策, 今釜史郎

臨床放射線 69 巻 ( 6 ) 頁： 787 - 792 2024年11月

記述言語：日本語出版者・発行元：金原出版(株)

＜文献概要＞脊椎・脊髄疾患は,高齢化社会においてますます注目される分野であり,特に脊髄腫瘍や脊髄圧迫症などの重大な病態が含まれる。画像診断はこれらの疾患の検出と診断に不可欠であり,近年,人工知能(AI)技術が医療において急速に進展し,画像診断の効率と精度を向上させる手段として期待されている。本稿では,脊椎・脊髄疾患におけるAIの現状とその応用について,特に画像診断分野における具体的な利用例や技術的な進展を中心に紹介する。

Skeleton-guided 3D convolutional neural network for tubular structure segmentation

Zhu, RY; Oda, M; Hayashi, Y; Kitasaka, T; Misawa, K; Fujiwara, M; Mori, K

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY 2024年9月

記述言語：英語出版者・発行元：International Journal of Computer Assisted Radiology and Surgery

Purpose: Accurate segmentation of tubular structures is crucial for clinical diagnosis and treatment but is challenging due to their complex branching structures and volume imbalance. The purpose of this study is to propose a 3D deep learning network that incorporates skeleton information to enhance segmentation accuracy in these tubular structures. Methods: Our approach employs a 3D convolutional network to extract 3D tubular structures from medical images such as CT volumetric images. We introduce a skeleton-guided module that operates on extracted features to capture and preserve the skeleton information in the segmentation results. Additionally, to effectively train our deep model in leveraging skeleton information, we propose a sigmoid-adaptive Tversky loss function which is specifically designed for skeleton segmentation. Results: We conducted experiments on two distinct 3D medical image datasets. The first dataset consisted of 90 cases of chest CT volumetric images, while the second dataset comprised 35 cases of abdominal CT volumetric images. Comparative analysis with previous segmentation approaches demonstrated the superior performance of our method. For the airway segmentation task, our method achieved an average tree length rate of 93.0%, a branch detection rate of 91.5%, and a precision rate of 90.0%. In the case of abdominal artery segmentation, our method attained an average precision rate of 97.7%, a recall rate of 91.7%, and an F-measure of 94.6%. Conclusion: We present a skeleton-guided 3D convolutional network to segment tubular structures from 3D medical images. Our skeleton-guided 3D convolutional network could effectively segment small tubular structures, outperforming previous methods.

DOI： 10.1007/s11548-024-03215-x

整形外科領域の画像診断技術の深化深層学習を用いたMRI運動器腫瘍診断

伊藤定之, 中島宏彰, 世木直喜, 大内田隼, 小田昌宏, 山内一平, 森田圭則, 大出幸史, 長谷康弘, 森健策, 今釜史郎

日本整形外科学会雑誌 98 巻 ( 8 ) 頁： S1732 - S1732 2024年9月

Analysis of the performance of the CorneAI for iOS in the classification of corneal diseases and cataracts based on journal photographs.

Taki Y, Ueno Y, Oda M, Kitaguchi Y, Ibrahim OMA, Aketa N, Yamaguchi T

Scientific reports 14 巻 ( 1 ) 頁： 15517 2024年7月

記述言語：英語出版者・発行元：Scientific Reports

CorneAI for iOS is an artificial intelligence (AI) application to classify the condition of the cornea and cataract into nine categories: normal, infectious keratitis, non-infection keratitis, scar, tumor, deposit, acute primary angle closure, lens opacity, and bullous keratopathy. We evaluated its performance to classify multiple conditions of the cornea and cataract of various races in images published in the Cornea journal. The positive predictive value (PPV) of the top classification with the highest predictive score was 0.75, and the PPV for the top three classifications exceeded 0.80. For individual diseases, the highest PPVs were 0.91, 0.73, 0.42, 0.72, 0.77, and 0.55 for infectious keratitis, normal, non-infection keratitis, scar, tumor, and deposit, respectively. CorneAI for iOS achieved an area under the receiver operating characteristic curve of 0.78 (95% confidence interval [CI] 0.5–1.0) for normal, 0.76 (95% CI 0.67–0.85) for infectious keratitis, 0.81 (95% CI 0.64–0.97) for non-infection keratitis, 0.55 (95% CI 0.41–0.69) for scar, 0.62 (95% CI 0.27–0.97) for tumor, and 0.71 (95% CI 0.53–0.89) for deposit. CorneAI performed well in classifying various conditions of the cornea and cataract when used to diagnose journal images, including those with variable imaging conditions, ethnicities, and rare cases.

DOI： 10.1038/s41598-024-66296-3

Federated 3D multi-organ segmentation with partially labeled and unlabeled data

Zheng, Z; Hayashi, Y; Oda, M; Kitasaka, T; Misawa, K; Mori, K

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY 2024年5月

記述言語：英語出版者・発行元：International Journal of Computer Assisted Radiology and Surgery

Purpose: This paper considers a new problem setting for multi-organ segmentation based on the following observations. In reality, (1) collecting a large-scale dataset from various institutes is usually impeded due to privacy issues; (2) many images are not labeled since the slice-by-slice annotation is costly; and (3) datasets may exhibit inconsistent, partial annotations across different institutes. Learning a federated model from these distributed, partially labeled, and unlabeled samples is an unexplored problem. Methods: To simulate this multi-organ segmentation problem, several distributed clients and a central server are maintained. The central server coordinates with clients to learn a global model using distributed private datasets, which comprise a small part of partially labeled images and a large part of unlabeled images. To address this problem, a practical framework that unifies partially supervised learning (PSL), semi-supervised learning (SSL), and federated learning (FL) paradigms with PSL, SSL, and FL modules is proposed. The PSL module manages to learn from partially labeled samples. The SSL module extracts valuable information from unlabeled data. Besides, the FL module aggregates local information from distributed clients to generate a global statistical model. With the collaboration of three modules, the presented scheme could take advantage of these distributed imperfect datasets to train a generalizable model. Results: The proposed method was extensively evaluated with multiple abdominal CT datasets, achieving an average result of 84.83% in Dice and 41.62 mm in 95HD for multi-organ (liver, spleen, and stomach) segmentation. Moreover, its efficacy in transfer learning further demonstrated its good generalization ability for downstream segmentation tasks. Conclusion: This study considers a novel problem of multi-organ segmentation, which aims to develop a generalizable model using distributed, partially labeled, and unlabeled CT images. A practical framework is presented, which, through extensive validation, has proved to be an effective solution, demonstrating strong potential in addressing this challenging problem.

DOI： 10.1007/s11548-024-03139-6

A clinical assessment of three-dimensional-printed liver model navigation for thrice or more repeated hepatectomy based on a conversation analysis

Igami, T; Maehigashi, A; Nakamura, Y; Hayashi, Y; Oda, M; Yokoyama, Y; Mizuno, T; Yamaguchi, J; Onoe, S; Sunagawa, M; Watanabe, N; Baba, T; Kawakatsu, S; Mori, K; Miwa, K; Ebata, T

SURGERY TODAY 2024年4月

記述言語：英語出版者・発行元：Surgery Today

Purposes: We performed a conversation analysis of the speech conducted among the surgical team during three-dimensional (3D)-printed liver model navigation for thrice or more repeated hepatectomy (TMRH). Methods: Seventeen patients underwent 3D-printed liver navigation surgery for TMRH. After transcription of the utterances recorded during surgery, the transcribed utterances were coded by the utterer, utterance object, utterance content, sensor, and surgical process during conversation. We then analyzed the utterances and clarified the association between the surgical process and conversation through the intraoperative reference of the 3D-printed liver. Results: In total, 130 conversations including 1648 segments were recorded. Utterance coding showed that the operator/assistant, 3D-printed liver/real liver, fact check (F)/plan check (Pc), visual check/tactile check, and confirmation of planned resection or preservation target (T)/confirmation of planned or ongoing resection line (L) accounted for 791/857, 885/763, 1148/500, 1208/440, and 1304/344 segments, respectively. The utterance’s proportions of assistants, F, F of T on 3D-printed liver, F of T on real liver, and Pc of L on 3D-printed liver were significantly higher during non-expert surgeries than during expert surgeries. Confirming the surgical process with both 3D-printed liver and real liver and performing planning using a 3D-printed liver facilitates the safe implementation of TMRH, regardless of the surgeon’s experience. Conclusions: The present study, using a unique conversation analysis, provided the first evidence for the clinical value of 3D-printed liver for TMRH for anatomical guidance of non-expert surgeons.

DOI： 10.1007/s00595-024-02835-9

Artificial intelligence-based diagnostic imaging system with virtual enteroscopy and virtual unfolded views to evaluate small bowel lesions in Crohn's disease.

Furukawa K, Oda M, Watanabe O, Nakamura M, Yamamura T, Maeda K, Mori K, Kawashima H

Revista espanola de enfermedades digestivas 2024年3月

Anatomical attention can help to segment the dilated pancreatic duct in abdominal CT.

Shen C, Roth HR, Hayashi Y, Oda M, Sato G, Miyamoto T, Rueckert D, Mori K

International journal of computer assisted radiology and surgery 2024年3月

Use of 3D-printed model of liver by experts and novices 査読有り

Maehigashi, A; Miwa, K; Oda, M; Nakamura, Y; Mori, K; Igami, T

CURRENT PSYCHOLOGY 2024年2月

掲載種別：研究論文（学術雑誌）出版者・発行元：Current Psychology

This study investigated the influence of using three-dimensional (3D) computer and 3D-printed models on the spatial reasoning of experts and novices. The task of this study required general university students as novices in Experiment 1 and surgeons specializing in digestive surgery as experts in Experiment 2 to infer the cross sections of a liver, using a 3D-computer or 3D-printed model. The results of the experiments showed that the university students learned faster and inferred the liver structure more accurately with the 3D-printed model than with the 3D-computer model. Conversely, the surgeons showed the same task performance when using the 3D-computer and 3D-printed models; however, they performed the task with more confidence and less workload during the task with the 3D-printed model. Based on the results, the cognitive effects and advantages of using 3D-printed models for novices and experts have been discussed.

DOI： 10.1007/s12144-024-05676-4

YOLOv7-RepFPN: Improving real-time performance of laparoscopic tool detection on embedded systems 査読有り

Liu, YZ; Hayashi, Y; Oda, M; Kitasaka, T; Mori, K

HEALTHCARE TECHNOLOGY LETTERS 2024年1月

掲載種別：研究論文（学術雑誌）出版者・発行元：Healthcare Technology Letters

This study focuses on enhancing the inference speed of laparoscopic tool detection on embedded devices. Laparoscopy, a minimally invasive surgery technique, markedly reduces patient recovery times and postoperative complications. Real-time laparoscopic tool detection helps assisting laparoscopy by providing information for surgical navigation, and its implementation on embedded devices is gaining interest due to the portability, network independence and scalability of the devices. However, embedded devices often face computation resource limitations, potentially hindering inference speed. To mitigate this concern, the work introduces a two-fold modification to the YOLOv7 model: the feature channels and integrate RepBlock is halved, yielding the YOLOv7-RepFPN model. This configuration leads to a significant reduction in computational complexity. Additionally, the focal EIoU (efficient intersection of union) loss function is employed for bounding box regression. Experimental results on an embedded device demonstrate that for frame-by-frame laparoscopic tool detection, the proposed YOLOv7-RepFPN achieved an mAP of 88.2% (with IoU set to 0.5) on a custom dataset based on EndoVis17, and an inference speed of 62.9 FPS. Contrasting with the original YOLOv7, which garnered an 89.3% mAP and 41.8 FPS under identical conditions, the methodology enhances the speed by 21.1 FPS while maintaining detection accuracy. This emphasizes the effectiveness of the work.

DOI： 10.1049/htl2.12072

EndoSelf: Self-supervised Monocular 3D Scene Reconstruction of Deformable Tissues with Neural Radiance Fields on Endoscopic Videos

Deep learning model for extensive smartphone-based diagnosis and triage of cataracts and multiple corneal diseases.

Ueno Y, Oda M, Yamaguchi T, Fukuoka H, Nejima R, Kitaguchi Y, Miyake M, Akiyama M, Miyata K, Kashiwagi K, Maeda N, Shimazaki J, Noma H, Mori K, Oshika T

The British journal of ophthalmology 2024年1月

Li, WD; Hayashi, Y; Oda, M; Kitasaka, T; Misawa, K; Mori, K

MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION - MICCAI 2024, PT VI 15006 巻頁： 241 - 251 2024年

Towards better laparoscopic video segmentation: A class-wise contrastive learning approach with multi-scale feature extraction

Zhang L., Hayashi Y., Oda M., Mori K.

Healthcare Technology Letters 2024年

出版者・発行元：Healthcare Technology Letters

The task of segmentation is integral to computer-aided surgery systems. Given the privacy concerns associated with medical data, collecting a large amount of annotated data for training is challenging. Unsupervised learning techniques, such as contrastive learning, have shown powerful capabilities in learning image-level representations from unlabelled data. This study leverages classification labels to enhance the accuracy of the segmentation model trained on limited annotated data. The method uses a multi-scale projection head to extract image features at various scales. The partitioning method for positive sample pairs is then improved to perform contrastive learning on the extracted features at each scale to effectively represent the differences between positive and negative samples in contrastive learning. Furthermore, the model is trained simultaneously with both segmentation labels and classification labels. This enables the model to extract features more effectively from each segmentation target class and further accelerates the convergence speed. The method was validated using the publicly available CholecSeg8k dataset for comprehensive abdominal cavity surgical segmentation. Compared to select existing methods, the proposed approach significantly enhances segmentation performance, even with a small labelled subset (1–10%) of the dataset, showcasing a superior intersection over union (IoU) score.

DOI： 10.1049/htl2.12069

M U-Net: Intestine Segmentation Using Multi-dimensional Features for Ileus Diagnosis Assistance

An Q., Oda H., Hayashi Y., Kitasaka T., Hinoki A., Uchida H., Suzuki K., Takimoto A., Oda M., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 14313 LNCS 巻頁： 135 - 144 2024年

出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

The intestine is an essential digestive organ that can cause serious health problems once diseased. This paper proposes a method for intestine segmentation to intestine obstruction diagnosis assistance called multi-dimensional U-Net (M U-Net). We employ two encoders to extract features from two-dimensional (2D) CT slices and three-dimensional (3D) CT patches. These two encoders collaborate to enhance the segmentation accuracy of the model. Additionally, we incorporate deep supervision with the M U-Net to reduce the limitation of training with sparse label data sets. The experimental results demonstrated that the Dice of the proposed method was 73.22%, the recall was 79.89%, and the precision was 70.61%.

DOI： 10.1007/978-3-031-47076-9_14

Revisiting instrument segmentation: Learning from decentralized surgical sequences with various imperfect annotations

Zheng Z., Hayashi Y., Oda M., Kitasaka T., Mori K.

Healthcare Technology Letters 2024年

出版者・発行元：Healthcare Technology Letters

This paper focuses on a new and challenging problem related to instrument segmentation. This paper aims to learn a generalizable model from distributed datasets with various imperfect annotations. Collecting a large-scale dataset for centralized learning is usually impeded due to data silos and privacy issues. Besides, local clients, such as hospitals or medical institutes, may hold datasets with diverse and imperfect annotations. These datasets can include scarce annotations (many samples are unlabelled), noisy labels prone to errors, and scribble annotations with less precision. Federated learning (FL) has emerged as an attractive paradigm for developing global models with these locally distributed datasets. However, its potential in instrument segmentation has yet to be fully investigated. Moreover, the problem of learning from various imperfect annotations in an FL setup is rarely studied, even though it presents a more practical and beneficial scenario. This work rethinks instrument segmentation in such a setting and propose a practical FL framework for this issue. Notably, this approach surpassed centralized learning under various imperfect annotation settings. This method established a foundational benchmark, and future work can build upon it by considering each client owning various annotations and aligning closer with real-world complexities.

DOI： 10.1049/htl2.12068

Intestine Segmentation from CT Volume based on Bidirectional Teaching

An, Q; Oda, H; Hayashi, Y; Kitasaka, T; Hinoki, A; Uchida, H; Suzuki, K; Takimoto, A; Oda, M; Mori, K

MEDICAL IMAGING 2024: IMAGE PROCESSING 12926 巻 2024年

出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

This paper proposes an intestine segmentation method to segment intestines from CT volumes for helping clinicians diagnose intestine obstruction. For large-scale labeled datasets, fully-supervised methods have shown superior results. However, medical image segmentation is usually difficult to achieve accurate prediction due to the limited number of labeled data available for training. To address this challenge, we introduce a novel multi-view symmetrical network (MVS-Net) for intestine segmentation and incorporate bidirectional teaching to utilize unlabeled datasets. Specifically, we design the MVS-Net, which can use different sizes of convolution kernels instead of a fixed kernel size, enabling the network to capture multi-scale features from images’ different perceptual fields and ensure segmentation accuracy. Additionally, the pseudo-labels are generated by bidirectional teaching, which can make the network captures semantic information from large-scale unlabeled data for increasing the training data. We repeated the experiment five times, and used the averaged result on the intestines dataset to represent the segmentation accuracy of the proposed method. The experimental results showed the average Dice was 78.86%, the average recall 84.50%, and the average precision 75.94%, respectively.

DOI： 10.1117/12.3006623

Continual pretraining for enhanced multi-organ segmentation from CT images

Yang, YQ; Shen, C; Tang, YC; Roth, HR; Oda, M; Hayashi, Y; Misawa, K; Mori, K

MEDICAL IMAGING 2024: IMAGE PROCESSING 12926 巻 2024年

出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

Self-supervised pretraining has shown great performance in improving the accuracy of downstream tasks. Although pretraining on a large dataset improves performances, it becomes challenging to further optimize the model by solely enlarging the dataset. In contrast, additional adaptation of pretrained models to the target domain has shown promise in NLP. Inspired by the success of continual pretraining, we investigated the efficacy of adapting the target domain dataset to a pretrained model in medical imaging, particularly in the context of segmentation. We present a study based on a self-supervised pretraining framework using the SwinUNETR backbone. In this study, we improved the generalizability of the self-supervised pretraining by adapting a foundational model pretrained on 5k CT volumes to data of the downstream segmentation task. In detail, we employed 385 abdominal CT volumes for the continual task-adaptive pretraining and 24 abdominal CT volumes for the downstream segmentation task, all sourced from the same dataset. Additionally, we conducted comparative experiments to demonstrate the benefits of this task-adapting pretraining approach. Our method has shown that continual pretraining helps to improve the performances, achieving an average Dice score for 10-class organ segmentation of 87.8%.

DOI： 10.1117/12.3006630

Boosting substantia nigra segmentation from T2 weighted MRI via test-time normalization and distance-reweighted loss

Hu, T; Itoh, H; Oda, M; Saiki, S; Hattori, N; Kamagata, K; Sako, W; Ishikawa, K; Aoki, S; Mori, K

COMPUTER-AIDED DIAGNOSIS, MEDICAL IMAGING 2024 12927 巻 2024年

出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

Substantia nigra (SN) has been reported as significantly related to the progression of Parkinson’s Disease (PD). Fully automated segmentation of SN is an important step for developing an interpretable computer-aided diagnosis system for PD. Based on the deep learning techniques, this paper proposes a novel distance-reweighted loss function and combines it with the test-time normalization (TTN) to boost the fully automated SN segmentation accuracy from low contrast T2 weighted MRI. The proposed loss encourages the model to focus on the suspicious regions with vague boundaries, and the involved TTN narrows the gap between an input MRI volume and the reference MRI volumes in test-time. The results showed that both the proposed loss and TTN could help improve the segmentation accuracy. By combining the proposed loss and TTN, the averaged Dice coefficient achieved 70.90% from T2 weighted MRI, compared to 68.17% by the baseline method.

DOI： 10.1117/12.3008521

Adaptive Octree Cube Refinement Depending on Grasping Position for Deformable Organ Models

Miyazaki, R; Hayashi, Y; Oda, M; Mori, K

IMAGE-GUIDED PROCEDURES, ROBOTIC INTERVENTIONS, AND MODELING, MEDICAL IMAGING 2024 12928 巻 2024年

出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

This paper describes an adaptive octree cube refinement method for deformable organ models. Surgical simulation is one of the most promising ways for surgical training. Various types of surgical simulators have been researched and developed. Laparoscopic surgery simulators are already in practical use. They have been evaluated for their effectiveness in learning surgical techniques. To realize a high-quality simulator, it is important to efficiently process organ deformation models according to the content of the surgical simulation so that both high-resolution and real-Time processing. In this study, we extend adaptive mesh refinement, which increases mesh resolution in the manipulation region, and apply it to an octree cube structure. The refinement process of the octree cube structure is performed based on the distance from the grasping position of the gallbladder model. This approach improves the resolution of the octree in the area near the grasping position where relatively large deformations occur. In addition, it makes it easier to detect interference between the grasp model and the high-resolution grid of the octree. Simulation results showed that there were 199 cubes before and 339 cubes after refinement, and the FPS decreased from 44.1 FPS to 32.4 FPS on a standard CPU and GPU PC, which is still within real-Time processing.

DOI： 10.1117/12.3006816

HTSeg: Hybrid Two-Stage Segmentation Framework for Intestine Segmentation from CT Volumes

An Q., Oda H., Hayashi Y., Kitasaka T., Takimoto A., Hinoki A., Uchida H., Suzuki K., Oda M., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 15196 LNCS 巻頁： 32 - 41 2024年

出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

This paper proposes a semi-supervised intestine segmentation method from CT volumes. Our method can use densely and sparsely annotated CT volumes for training to reduce the labor of manually annotating intestines. The proposed Hybrid Two-stage Segmentation (HTSeg) framework consists of two networks, a 2D swin-transformer-based network as the first stage and a 3D network as the second stage. In the first stage, we use 6964 labeled CT slices to train the 2D Swin U-Net. The trained 2D Swin U-Net is used to generate pseudo-labels for sparse annotation data. In the second stage, we use sparsely annotated datasets with pseudo-labels and densely annotated datasets to train a 3D multi-view symmetrical network (MVSNet). Experimental results showed that the Dice score of the proposed method was 74.70%, which was 1.03% higher than just using MVSNet. Compared with the other four previous methods (3D U-Net, CPS, EM, MT), the proposed method produced competitive segmentation performance. The code can be found at: https://github.com/MoriLabNU/semi-pseudo-labels.

DOI： 10.1007/978-3-031-73083-2_4

Performance Improvement for Medical Image Classification Model by using Gradient-based Analytical Feature Selection

Toda, R; Itoh, H; Oda, M; Hayashi, Y; Otake, Y; Hashimoto, M; Akashi, T; Aoki, S; Mori, K

COMPUTER-AIDED DIAGNOSIS, MEDICAL IMAGING 2024 12927 巻 2024年

出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

This paper presents a gradient-based analytical method for improving medical image classification. The automated classification of diseases is important in computer-aided diagnosis. In addition to accurate classification, its explainability is an essential factor toward its practical application. A gradient-based visual explanation provides the explainability of a model of an convolutional neural network (CNN) by indicating important patterns in an input image. Most studies use this explanation to assess CNN’s validity in a qualitative manner. On the other hand, in addition to a model’s explainability, our motivation is to utilize the visual-explanation methods to enhance the classification accuracy of a CNN model. In this study, we propose a weight-analysis-based method to improve the classification accuracy of a trained-CNN model without additional training. The proposed method selects important patterns based on a gradient-based weight analysis of a middle layer in a trained model and suppresses irrelevant patterns in the extracted features for the classification. We applied our analytical method to a convolutional and a global-average-pooling layers in a CNN, which classifies a chest CT volume into COVID-19 typical and non-typical cases. As shown in classification results on 302 testing cases, our method improved the accuracy of the COVID-19 classification.

DOI： 10.1117/12.3006620

A Bayesian Approach to Weakly-Supervised Laparoscopic Image Segmentation

Zheng, Z; Hayashi, Y; Oda, M; Kitasaka, T; Mori, K

MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION - MICCAI 2024, PT VI 15006 巻頁： 14 - 24 2024年

Smart: Stoma Measurement, Analysis, Report Tool for Microscope Image and Its Application in Plant Phenotyping

Wang C., Sato K., Hayashi Y., Oda M., Ishimaru Y., Uozumi N., Mori K.

International Conference on Wavelet Analysis and Pattern Recognition 2024年

出版者・発行元：International Conference on Wavelet Analysis and Pattern Recognition

This manuscript describes a deep learning-based algorithm for inferring stomatal phenotypes from microscope images. Botanists spray compounds on leaf surfaces and observe their state under a microscope to study the effects of different compounds on stomatal opening and closing. We propose a stomatal orientation-based method that uses stomatal orientation to guide stomatal measurements. This method has three modules: a stomata detection module that locates the stoma region and orientation using deep learning-based object detection. A stoma segmentation module that segments the aperture, guard cell, and thick inner wall from the stoma ROI image. And a phenotype quantification module that calculates the phenotype parameters by analyzing the mask image. The experimental results show that the proposed method can resolve stomata with high accuracy (the average R2 of the previous method is 0.66, and the proposed method is 0.96). For the development of the community, we will release the algorithm and tool involved in this article in GitHub.

DOI： 10.1109/ICWAPR63074.2024.10870510

Semi-supervised Tubular Structure Segmentation with Cross Geometry and Hausdorff Distance Consistency

Zhu, RY; Oda, M; Hayashi, Y; Kitasaka, T; Mori, K

MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION - MICCAI 2024, PT VIII 15008 巻頁： 612 - 622 2024年

出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

This study introduces a novel semi-supervised method for 3D segmentation of tubular structures. Complete and automated segmentation of complex tubular structures in medical imaging remains a challenging task. Traditional supervised deep learning methods often demand a tremendous number of annotated data to train the deep model, with the high cost and difficulty of obtaining annotations. To address this, a semi-supervised approach could be a viable solution. Segmenting complex tubular structures with limited annotated data remains a formidable challenge. Many semi-supervised techniques rely on pseudo-labeling, which involves generating labels for unlabeled images based on predictions from a model trained on labeled data. Besides, several semi-supervised learning methods are proposed based on data-level consistency, which enforces consistent predictions by applying perturbations to input images. However, these methods tend to overlook the geometric shape characteristics of the segmentation targets. In our research, we introduce a task-level consistency learning approach that incorporates cross geometry consistency and the Hausdorff distance consistency, taking advantage of the geometric shape properties of both labeled and unlabeled data. Our deep learning model generates both a segmentation map and a distance transform map. By applying the proposed consistency, we ensure that the geometric shapes in both maps align closely, thereby enhancing the accuracy and performance of tubular structure segmentation. We tested our method on airway segmentation in 3D CT scans, where it outperformed the recent state-of-the-art methods, showing an 88.4% tree length detected rate, 82.8% branch detected rate, and 89.7% precision rate.

DOI： 10.1007/978-3-031-72111-3_58

Leaf Navigated Plant Growth Measurement for Automated Plant Cultivation System

Wang C., Ukai Y., Hayashi Y., Oda M., Sat Y., Mori K.

International Conference on Wavelet Analysis and Pattern Recognition 2024年

出版者・発行元：International Conference on Wavelet Analysis and Pattern Recognition

This manuscript describes a workflow for automatically measuring plant growth from scanned images. To explore the different conditions of plant growth, botanists sow seeds in trays and use imaging devices such as scanners or cameras to record the daily growth of plants. Previous methods behave poorly in root measuring. To design an automated plant growth measurement system, we propose a leaf-root distance-based plant organ measurement method that finds the paths of the root by minimizing the distance energy from the root endpoints to the leaves. This method contains three modules. The first is a deep learning-based plant detection module to detect individual plants' regions of interest (ROI). An organ mask generation module that uses Otsu-thresholding to segment root and leaf. A plant organ analysis module to generate plant phenotypes (e.g., length of primary and lateral roots) from mask images. The experimental results showed that the proposed method had a higher measurement accuracy than previous methods.

DOI： 10.1109/ICWAPR63074.2024.10870512

SGSR: style-subnets-assisted generative latent bank for large-factor super-resolution with registered medical image dataset. 国際誌

Zheng T, Oda H, Hayashi Y, Nakamura S, Mori M, Takabatake H, Natori H, Oda M, Mori K

International journal of computer assisted radiology and surgery 2023年12月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：International Journal of Computer Assisted Radiology and Surgery

Purpose: We propose a large-factor super-resolution (SR) method for performing SR on registered medical image datasets. Conventional SR approaches use low-resolution (LR) and high-resolution (HR) image pairs to train a deep convolutional neural network (DCN). However, LR–HR images in medical imaging are commonly acquired from different imaging devices, and acquiring LR–HR image pairs needs registration. Registered LR–HR images have registration errors inevitably. Using LR–HR images with registration error for training an SR DCN causes collapsed SR results. To address these challenges, we introduce a novel SR approach designed specifically for registered LR–HR medical images. Methods: We propose style-subnets-assisted generative latent bank for large-factor super-resolution (SGSR) trained with registered medical image datasets. Pre-trained generative models named generative latent bank (GLB), which stores rich image priors, can be applied in SR to generate realistic and faithful images. We improve GLB by newly introducing style-subnets-assisted GLB (S-GLB). We also propose a novel inter-uncertainty loss to boost our method’s performance. Introducing more spatial information by inputting adjacent slices further improved the results. Results: SGSR outperforms state-of-the-art (SOTA) supervised SR methods qualitatively and quantitatively on multiple datasets. SGSR achieved higher reconstruction accuracy than recently supervised baselines by increasing peak signal-to-noise ratio from 32.628 to 34.206 dB. Conclusion: SGSR performs large-factor SR while given a registered LR–HR medical image dataset with registration error for training. SGSR’s results have both realistic textures and accurate anatomical structures due to favorable quantitative and qualitative results. Experiments on multiple datasets demonstrated SGSR’s superiority over other SOTA methods. SR medical images generated by SGSR are expected to improve the accuracy of pre-surgery diagnosis and reduce patient burden.

DOI： 10.1007/s11548-023-03037-3

Automated Detection of the Thoracic Ossification of the Posterior Longitudinal Ligament Using Deep Learning and Plain Radiographs

Ito, S; Nakashima, H; Segi, N; Ouchida, J; Oda, M; Yamauchi, I; Oishi, R; Miyairi, Y; Mori, K; Imagama, S

BIOMED RESEARCH INTERNATIONAL 2023 巻頁： 8495937 2023年11月

記述言語：英語出版者・発行元：BioMed Research International

Ossification of the ligaments progresses slowly in the initial stages, and most patients are unaware of the disease until obvious myelopathy symptoms appear. Consequently, treatment and clinical outcomes are not satisfactory. This study is aimed at developing an automated system for the detection of the thoracic ossification of the posterior longitudinal ligament (OPLL) using deep learning and plain radiography. We retrospectively reviewed the data of 146 patients with thoracic OPLL and 150 control cases without thoracic OPLL. Plain lateral thoracic radiographs were used for object detection, training, and validation. Thereafter, an object detection system was developed, and its accuracy was calculated. The performance of the proposed system was compared with that of two spine surgeons. The accuracy of the proposed object detection model based on plain lateral thoracic radiographs was 83.4%, whereas the accuracies of spine surgeons 1 and 2 were 80.4% and 77.4%, respectively. Our findings indicate that our automated system, which uses a deep learning-based method based on plain radiographs, can accurately detect thoracic OPLL. This system has the potential to improve the diagnostic accuracy of thoracic OPLL.

DOI： 10.1155/2023/8495937

Automated Detection and Diagnosis of Spinal Schwannomas and Meningiomas Using Deep Learning and Magnetic Resonance Imaging

ラベル拡張法によるマイクロCT像中の小葉間隔壁の抽出

深井大輔, 小田紘久, 林雄一郎, 鄭通, 中村彰太, 小田昌宏, 森健策

日本コンピュータ外科学会誌 25 巻 ( 3 ) 頁： 259 - 259 2023年11月

Ito Sadayuki, Nakashima Hiroaki, Segi Naoki, Ouchida Jun, Oda Masahiro, Yamauchi Ippei, Oishi Ryotaro, Miyairi Yuichi, Mori Kensaku, Imagama Shiro

JOURNAL OF CLINICAL MEDICINE 12 巻 ( 15 ) 2023年8月

記述言語：英語出版者・発行元：Journal of Clinical Medicine

Spinal cord tumors are infrequently identified spinal diseases that are often difficult to diagnose even with magnetic resonance imaging (MRI) findings. To minimize the probability of overlooking these tumors and improve diagnostic accuracy, an automatic diagnostic system is needed. We aimed to develop an automated system for detecting and diagnosing spinal schwannomas and meningiomas based on deep learning using You Only Look Once (YOLO) version 4 and MRI. In this retrospective diagnostic accuracy study, the data of 50 patients with spinal schwannomas, 45 patients with meningiomas, and 100 control cases were reviewed, respectively. Sagittal T1-weighted (T1W) and T2-weighted (T2W) images were used for object detection, classification, training, and validation. The object detection and diagnosis system was developed using YOLO version 4. The accuracies of the proposed object detections based on T1W, T2W, and T1W + T2W images were 84.8%, 90.3%, and 93.8%, respectively. The accuracies of the object detection for two spine surgeons were 88.9% and 90.1%, respectively. The accuracies of the proposed diagnoses based on T1W, T2W, and T1W + T2W images were 76.4%, 83.3%, and 84.1%, respectively. The accuracies of the diagnosis for two spine surgeons were 77.4% and 76.1%, respectively. We demonstrated an accurate, automated detection and diagnosis of spinal schwannomas and meningiomas using the developed deep learning-based method based on MRI. This system could be valuable in supporting radiological diagnosis of spinal schwannomas and meningioma, with a potential of reducing the radiologist’s overall workload.

DOI： 10.3390/jcm12155075

Artificial intelligence in a prediction model for postendoscopic retrograde cholangiopancreatography pancreatitis

Takahashi Hidekazu, Ohno Eizaburo, Furukawa Taiki, Yamao Kentaro, Ishikawa Takuya, Mizutani Yasuyuki, Iida Tadashi, Shiratori Yoshimune, Oyama Shintaro, Koyama Junji, Mori Kensaku, Hayashi Yuichiro, Oda Masahiro, Suzuki Takahisa, Kawashima Hiroki

DIGESTIVE ENDOSCOPY 2023年7月

記述言語：英語出版者・発行元：Digestive Endoscopy

Objectives: In this study we aimed to develop an artificial intelligence-based model for predicting postendoscopic retrograde cholangiopancreatography (ERCP) pancreatitis (PEP). Methods: We retrospectively reviewed ERCP patients at Nagoya University Hospital (NUH) and Toyota Memorial Hospital (TMH). We constructed two prediction models, a random forest (RF), one of the machine-learning algorithms, and a logistic regression (LR) model. First, we selected features of each model from 40 possible features. Then the models were trained and validated using three fold cross-validation in the NUH cohort and tested in the TMH cohort. The area under the receiver operating characteristic curve (AUROC) was used to assess model performance. Finally, using the output parameters of the RF model, we classified the patients into low-, medium-, and high-risk groups. Results: A total of 615 patients at NUH and 544 patients at TMH were enrolled. Ten features were selected for the RF model, including albumin, creatinine, biliary tract cancer, pancreatic cancer, bile duct stone, total procedure time, pancreatic duct injection, pancreatic guidewire-assisted technique without a pancreatic stent, intraductal ultrasonography, and bile duct biopsy. In the three fold cross-validation, the RF model showed better predictive ability than the LR model (AUROC 0.821 vs. 0.660). In the test, the RF model also showed better performance (AUROC 0.770 vs. 0.663, P = 0.002). Based on the RF model, we classified the patients according to the incidence of PEP (2.9%, 10.0%, and 23.9%). Conclusion: We developed an RF model. Machine-learning algorithms could be powerful tools to develop accurate prediction models.

DOI： 10.1111/den.14622

肺マイクロCT像中の小葉間隔壁抽出のための教師データ生成に関する検討

深井大輔, 小田紘久, 林雄一郎, 鄭通, 中村彰太, 小田昌宏, 森健策

日本医用画像工学会大会予稿集 42回巻頁： 153 - 154 2023年7月

記述言語：日本語出版者・発行元：(一社)日本医用画像工学会

本稿では,肺マイクロCT像における小葉間隔壁抽出のための教師データ生成手法について述べる.マイクロCTとよばれる高解像度CT撮像装置により,肺の微細構造を撮像可能となった.肺マイクロCT像による三次元的な肺微細構造解析に基づく病態や生体機能のさらなる解明が期待される.肺微細構造の一つである小葉間隔壁の構造を明らかにするため,我々はマイクロCT像からの小葉間隔壁の構造抽出を目指す.マイクロCT像における小葉間隔壁の教師データ作成の負担は大きい.本稿では,教師データ生成の負担軽減のため,二次元画像のみのラベル付与から,三次元画像のラベル画像を得る方法を検討する.本手法で得たラベル画像から5枚の断面画像をランダムに取り出して評価したところ,平均のDice係数は0.75だった.本手法によって得られたラベル画像を元に,手動で断面画像の教師データ生成を行ったところ,作業時間が27～30分から5～7分に短縮された.(著者抄録)

2.5次元特徴抽出及び3次元特徴選択によるCOVID-19自動分類モデルの汎化性能に関する検討

遠田涼, 伊東隼人, 小田昌宏, 林雄一郎, 大竹義人, 橋本正弘, 明石敏昭, 青木茂樹, 森健策

日本医用画像工学会大会予稿集 42回巻頁： 209 - 210 2023年7月

記述言語：日本語出版者・発行元：(一社)日本医用画像工学会

本稿では,2.5次元の特徴抽出及び3次元の特徴選択によるCOVID-19自動分類用CNNを提案し,その汎化性能を複数のデータセットを用いて検証する.新型コロナウイルス感染症(COVID-19)の診断において,胸部CT検査は不可欠であるが,CT画像の読影は放射線科医らにとって負担が大きく,コンピュータ支援診断技術の開発が求められている.COVID-19の流行初期から様々な自動分類モデルが提案されてきたが,その多くは小規模な内部データセットで分類性能を検証するに留まっており,汎用的なモデルは少数である.本研究では,2次元及び3次元の特徴抽出器を組み合わせることで過学習を抑制した2.5次元のCNNと,抽出された特徴から重要なパターンのみを選択する3次元のattention機構を有する分類モデルを提案し,その汎化性能を検証する.提案モデルを含めて6種類のモデルを用いた比較の結果,提案モデルは複数のデータセットで安定した分類性能を示した.(著者抄録)

Gaussian affinity and GIoU-based loss for perforation detection and localization from colonoscopy videos

DEVELOPMENT OF A MACHINE-LEARNING MODEL FOR PREDICTING POST-ERCP PANCREATITIS

Takahashi Hidekazu, Eizaburo Ohno, Taiki Furukawa, Kentaro Yamao, Takuya Ishikawa, Yasuyuki Mizutani, Tadashi Iida, Yoshimune Shiratori, Shintaro Oyama, Junji Koyama, Kensaku Mori, Yuichiro Hayashi, Masahiro Oda, Takahisa Suzuki, Hiroki Kawashima

Gastrointestinal Endoscopy 97 巻 ( 6 ) 頁： AB656 - AB656 2023年6月

Jiang Kai, Itoh Hayato, Oda Masahiro, Okumura Taishi, Mori Yuichi, Misawa Masashi, Hayashi Takemasa, Kudo Shin-Ei, Mori Kensaku

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY 18 巻 ( 5 ) 頁： 795 - 805 2023年5月

記述言語：英語出版者・発行元：International Journal of Computer Assisted Radiology and Surgery

Purpose: Endoscopic submucosal dissection (ESD) is a minimally invasive treatment for early gastric cancer. However, perforations may happen and cause peritonitis during ESD. Thus, there is a potential demand for a computer-aided diagnosis system to support physicians in ESD. This paper presents a method to detect and localize perforations from colonoscopy videos to avoid perforation ignoring or enlarging by ESD physicians. Method: We proposed a training method for YOLOv3 by using GIoU and Gaussian affinity losses for perforation detection and localization in colonoscopic images. In this method, the object functional contains the generalized intersection over Union loss and Gaussian affinity loss. We propose a training method for the architecture of YOLOv3 with the presented loss functional to detect and localize perforations precisely. Results: To qualitatively and quantitatively evaluate the presented method, we created a dataset from 49 ESD videos. The results of the presented method on our dataset revealed a state-of-the-art performance of perforation detection and localization, which achieved 0.881 accuracy, 0.869 AUC, and 0.879 mean average precision. Furthermore, the presented method is able to detect a newly appeared perforation in 0.1 s. Conclusions: The experimental results demonstrated that YOLOv3 trained by the presented loss functional were very effective in perforation detection and localization. The presented method can quickly and precisely remind physicians of perforation happening in ESD. We believe a future CAD system can be constructed for clinical applications with the proposed method.

DOI： 10.1007/s11548-022-02821-x

Correction to: Gaussian affinity and GIoU-based loss for perforation detection and localization from colonoscopy videos.

Jiang K, Itoh H, Oda M, Okumura T, Mori Y, Misawa M, Hayashi T, Kudo SE, Mori K

International journal of computer assisted radiology and surgery 18 巻 ( 5 ) 頁： 807 2023年5月

Deep learning-based prediction model for postoperative complications of cervical posterior longitudinal ligament ossification. 査読有り

Ito S, Nakashima H, Yoshii T, Egawa S, Sakai K, Kusano K, Tsutui S, Hirai T, Matsukura Y, Wada K, Katsumi K, Koda M, Kimura A, Furuya T, Maki S, Nagoshi N, Nishida N, Nagamoto Y, Oshima Y, Ando K, Takahata M, Mori K, Nakajima H, Murata K, Miyagi M, Kaito T, Yamada K, Banno T, Kato S, Ohba T, Inami S, Fujibayashi S, Katoh H, Kanno H, Oda M, Mori K, Taneichi H, Kawaguchi Y, Takeshita K, Matsumoto M, Yamazaki M, Okawa A, Imagama S

European spine journal : official publication of the European Spine Society, the European Spinal Deformity Society, and the European Section of the Cervical Spine Research Society 2023年2月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：European Spine Journal

Purpose: Postoperative complication prediction helps surgeons to inform and manage patient expectations. Deep learning, a model that finds patterns in large samples of data, outperform traditional statistical methods in making predictions. This study aimed to create a deep learning-based model (DLM) to predict postoperative complications in patients with cervical ossification of the posterior longitudinal ligament (OPLL). Methods: This prospective multicenter study was conducted by the 28 institutions, and 478 patients were included in the analysis. Deep learning was used to create two predictive models of the overall postoperative complications and neurological complications, one of the major complications. These models were constructed by learning the patient's preoperative background, clinical symptoms, surgical procedures, and imaging findings. These logistic regression models were also created, and these accuracies were compared with those of the DLM. Results: Overall complications were observed in 127 cases (26.6%). The accuracy of the DLM was 74.6 ± 3.7% for predicting the overall occurrence of complications, which was comparable to that of the logistic regression (74.1%). Neurological complications were observed in 48 cases (10.0%), and the accuracy of the DLM was 91.7 ± 3.5%, which was higher than that of the logistic regression (90.1%). Conclusion: A new algorithm using deep learning was able to predict complications after cervical OPLL surgery. This model was well calibrated, with prediction accuracy comparable to that of regression models. The accuracy remained high even for predicting only neurological complications, for which the case number is limited compared to conventional statistical methods.

DOI： 10.1007/s00586-023-07562-2

A semantic segmentation method for laparoscopic images using semantically similar groups

Uramoto L., Hayashi Y., Oda M., Kitasaka T., Misawa K., Mori K.

Progress in Biomedical Optics and Imaging - Proceedings of SPIE 12466 巻 2023年

出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

In this paper, we present a segmentation method for laparoscopic images using semantically similar groups for multi-class semantic segmentation. Accurate semantic segmentation is a key problem for computer assisted surgeries. Common segmentation models do not explicitly learn similarities between classes. We propose a model that, in addition to learning to segment an image into classes, also learns to segment it into human-defined semantically similar groups. We modify the LinkNet34 architecture by adding a second decoder with an auxiliary task of segmenting the image into these groups. The feature maps of the second decoder are merged into the final decoder. We validate our method against our base model LinkNet34 and a larger LinkNet50. We find that our proposed modification increased the performance both with mean Dice (average +1.5%) and mean Intersection over Union metrics (average +2.8%) on two laparoscopic datasets.

DOI： 10.1117/12.2654636

Improved method for COVID-19 Classification of Complex-Architecture CNN from Chest CT volumes using Orthogonal Ensemble Networks

Toda R., Oda M., Hayashi Y., Otake Y., Hashimoto M., Akashi T., Aoki S., Mori K.

Progress in Biomedical Optics and Imaging - Proceedings of SPIE 12465 巻 2023年

出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

This paper introduces the improved method for the COVID-19 classification based on computed tomography (CT) volumes using a combination of a complex-architecture convolutional neural network (CNN) and orthogonal ensemble networks (OEN). The novel coronavirus disease reported in 2019 (COVID-19) is still spreading worldwide. Early and accurate diagnosis of COVID-19 is required in such a situation, and the CT scan is an essential examination. Various computer-aided diagnosis (CAD) methods have been developed to assist and accelerate doctors' diagnoses. Although one of the effective methods is ensemble learning, existing methods combine some major models which do not specialize in COVID-19. In this study, we attempted to improve the performance of a CNN for the COVID-19 classification based on chest CT volumes. The CNN model specializes in feature extraction from anisotropic chest CT volumes. We adopt the OEN, an ensemble learning method considering inter-model diversity, to boost its feature extraction ability. For the experiment, We used chest CT volumes of 1283 cases acquired in multiple medical institutions in Japan. The classification result on 257 test cases indicated that the combination could improve the classification performance.

DOI： 10.1117/12.2653792

Classification of COVID-19 cases from chest CT volumes using hybrid model of 3D CNN and 3D MLP-Mixer

Oda M., Zheng T., Hayashi Y., Otake Y., Hashimoto M., Akashi T., Aoki S., Mori K.

Progress in Biomedical Optics and Imaging - Proceedings of SPIE 12465 巻 2023年

出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

This paper proposes an automated classification method of COVID-19 chest CT volumes using improved 3D MLP-Mixer. Novel coronavirus disease 2019 (COVID-19) spreads over the world, causing a large number of infected patients and deaths. Sudden increase in the number of COVID-19 patients causes a manpower shortage in medical institutions. Computer-aided diagnosis (CAD) system provides quick and quantitative diagnosis results. CAD system for COVID-19 enables efficient diagnosis workflow and contributes to reduce such manpower shortage. In image-based diagnosis of viral pneumonia cases including COVID-19, both local and global image features are important because viral pneumonia cause many ground glass opacities and consolidations in large areas in the lung. This paper proposes an automated classification method of chest CT volumes for COVID-19 diagnosis assistance. MLP-Mixer is a recent method of image classification using Vision Transformer-like architecture. It performs classification using both local and global image features. To classify 3D CT volumes, we developed a hybrid classification model that consists of both a 3D convolutional neural network (CNN) and a 3D version of the MLP-Mixer. Classification accuracy of the proposed method was evaluated using a dataset that contains 1205 CT volumes and obtained 79.5% of classification accuracy. The accuracy was higher than that of conventional 3D CNN models consists of 3D CNN layers and simple MLP layers.

DOI： 10.1117/12.2654706

Real Bronchoscopic Images-based Bronchial Nomenclature: a Preliminary Study

Wang C., Hayashi Y., Oda M., Kitasaka T., Takabatake H., Mori M., Honma H., Natori H., Mori K.

Progress in Biomedical Optics and Imaging - Proceedings of SPIE 12466 巻 2023年

出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

This article describes a method for bronchial nomenclature using real bronchoscopic (RB) images and pre-built knowledge base of branches. The bronchus has a complex tree-like structure, which increases the difficulty of bronchoscopy. Therefore, a bronchoscopic navigation system is used to help physicians during examination. Conventional navigation system used preoperative CT images and real bronchoscopic images to obtain the camera pose for navigation, whose accuracy is influenced by organ deformation. We propose a bronchial nomenclature method to estimate branch names for bronchoscopic navigation. This method consists of a bronchus knowledge base construction model, a camera motion estimation module, an anatomical structure tracking module, and a branch name estimation module. The knowledge base construction module is used to find the relationship of each branch. The anatomical tracking module is used to track the bronchial orifice (BO) extracted in each RB frame. The camera motion estimation module is used to estimate the camera motion between two frames. The branch name estimation module uses the pre-built bronchus knowledge base and BO tracking results to find the name of each branch. Experimental results showed that it is possible to estimate branch names using only RB images and the pre-built knowledge base of branches.

DOI： 10.1117/12.2654508

Priority attention network with Bayesian learning for fully automatic segmentation of substantia nigra from neuromelanin MRI

Hu Tao, Itoh Hayato, Oda Masahiro, Saiki Shinji, Hattori Nobutaka, Kamagata Koji, Aoki Shigeki, Mori Kensaku

MEDICAL IMAGING 2023 12464 巻 2023年

出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

Neuromelanin magnetic resonance imaging (NM-MRI) has been widely used in the diagnosis of Parkinson’s disease (PD) for its significantly enhanced contrast between the PD-related structure, the substantia nigra (SN) and surrounding tissues. To develop the computer-aided diagnosis (CAD) system of PD and reduce the labor burden of clinicians, precise and automatic segmentation of SN is becoming more and more desired. This paper proposes a novel network combining the priority gating attention and Bayesian learning for improving the accuracy of fully automatic SN segmentation from NM-MRI. Different from the conventional gated attention model, the proposed network uses the prior SN probability map for guiding the attention computation and reducing the potential disruptions introduced by the background. Additionally, to lower the risks of over-fitting and estimate the confidence scores for the segmentation results, Bayesian learning with Monte Carlo dropout is applied in the training and testing phases. The quantitative results showed that the proposed network acquired the averaged Dice score of 79.46% in comparison with the baseline model 77.93%.

DOI： 10.1117/12.2655112

Octree Cube Constraints in PBD Method for High Resolution Surgical Simulation

Miyazaki Rintaro, Hayashi Yuichiro, Oda Masahiro, Mori Kensaku

MEDICAL IMAGING 2023 12464 巻 2023年

出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

This paper proposes a deformable tissue model that introduces octree lattice vertex layout and cubic constraints to the orthodox PBD (Position Based Dynamics) method. Surgical simulation is expected to provide a safe method for training in surgery, which is especially useful for preoperative education of inexperienced surgeons and/or for the case a prior attempt is required. To build a surgical simulator, it is necessary to develop organ models with deformations and interaction algorithms between surgical instruments and organ models, all of which must be performed in real time. Since existing surgical simulators focus on real-time performance, the resolution of organ models is limited. The proposed method restricts the vertex locations of the PBD method to the vertices of the octree lattice to save computation time while maintaining a high deformation resolution. To obtain appropriate results even for large deformations, three-dimensional constraints are applied to each octree cube as the constraints of the PBD method. In the simulations, we tested the overall deformation by dropping a liver model and the local deformation scene by laparoscopic clipping. As a result, we achieved deformation simulations at 26.5 fps for the model with approximately 2,672 cube elements and 20,659 vertices.

DOI： 10.1117/12.2654092

L-former : A Lightweight Transformer for Realistic Medical Image Generation and its Application to Super-resolution

Zheng Tong, Oda Hirohisa, Hayashi Yuichiro, Nakamura Shota, Mori Masaki, Takabatake Hirotsugu, Natori Hiroshi, Oda Masahiro, Mori Kensaku

MEDICAL IMAGING 2023 12464 巻 2023年

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

Medical image analysis approaches such as data augmentation and domain adaption need huge amounts of realistic medical images. Generating realistic medical images by machine learning is a feasible approach. We propose L-former, a lightweight Transformer for realistic medical image generation. L-former can generate more reliable and realistic medical images than recent generative adversarial networks (GANs). Meanwhile, L-former does not consume as high computational cost as conventional Transformer-based generative models. L-former uses Transformers to generate low-resolution feature vectors at shallow layers, and uses convolutional neural networks to generate high-resolution realistic medical images at deep layers. Experimental results showed that L-former outperformed conventional GANs by FID scores 33.79 and 76.85 on two datasets, respectively. We further conducted a downstream study by using the images generated by L-former to perform a super-resolution task. A high PSNR score of 27.87 proved L-former’s ability to generate reliable images for super-resolution and showed its potential for applications in medical diagnosis.

DOI： 10.1117/12.2653776

Oesophagus Achalasia Diagnosis from Esophagoscopy Based on a Serial Multi-scale Network 査読有り

Jiang K., Oda M., Hayashi Y., Shiwaku H., Misawa M., Mori K.

Computer Methods in Biomechanics and Biomedical Engineering: Imaging and Visualization 2023年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Computer Methods in Biomechanics and Biomedical Engineering: Imaging and Visualization

Oesophageal achalasia is a primary oesophageal motility disorder disease. To diagnose oesophagus achalasia, physicians recommend endoscopic evaluation of the oesophagus. However, a low sensitivity still accompanies esophagoscopy on oesophagus achalasia diagnosis. Thus, a quantitative diagnosis system is needed to support physicians diagnose achalasia from the esophagoscopy video. This paper proposes a Serial Multi-scale Network for classifying achalasia images from the esophagoscopy video. The proposed method contains two main components, a Dense-pooling Net, and a Serial Multi-scale Dilated encoder. We construct the Dense-pooling Net using a convolution neural network with dense mixed-pooling connections to extract features. We design the Serial Multi-scale Dilated encoder based on a residual-style dilated encoder. We combine the dilated encoder and spatial attention modules to focus on features we need. We trained and evaluated our method with a dataset that was extracted from several esophagoscopy videos of achalasia patients. The evaluation results reveal a state-of-the-art accuracy of achalasia diagnosis. Furthermore, we developed a real-time computer-aided achalasia diagnosis system with the trained network. In the real-time test, the achalasia diagnosis system can stably output the diagnosis results in only (Formula presented.) seconds. The extended experiments demonstrate that the constructed diagnosis system can diagnose achalasia from esophagoscopy videos.

DOI： 10.1080/21681163.2022.2159534

KST-Mixer: kinematic spatio-temporal data mixer for colon shape estimation 査読有り国際共著

Oda M., Furukawa K., Navab N., Mori K.

Computer Methods in Biomechanics and Biomedical Engineering: Imaging and Visualization 2023年

担当区分：筆頭著者,　責任著者記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Computer Methods in Biomechanics and Biomedical Engineering: Imaging and Visualization

We propose a spatio-temporal mixing kinematic data estimation method to estimate the shape of the colon with deformations caused by colonoscope insertion. Endoscope tracking or a navigation system that navigates physicians to target positions is needed to reduce such complications as organ perforations. Although many previous methods focused to track bronchoscopes and surgical endoscopes, few number of colonoscope tracking methods were proposed because the colon largely deforms during colonoscope insertion. The deformation causes significant tracking errors. Colon deformation should be considered in the tracking process. We propose a colon shape estimation method using a Kinematic Spatio-Temporal data Mixer (KST-Mixer) that can be used during colonoscope insertions to the colon. Kinematic data of a colonoscope and the colon, including positions and directions of their centerlines, are obtained using electromagnetic and depth sensors. The proposed method separates the data into sub-groups along the spatial and temporal axes. The KST-Mixer extracts kinematic features and mix them along the axes multiple times. We evaluated colon shape estimation accuracies in phantom studies. The proposed method achieved 11.92 mm mean Euclidean distance error, the smallest of the previous methods. Statistical analysis indicated that the proposed method significantly reduced the error compared to the previous methods.

DOI： 10.1080/21681163.2022.2151938

Thrombosis region extraction and quantitative analysis in confocal laser scanning microscopic image sequence in in-vivo imaging

Wu Y., Oda M., Hayashi Y., Kawamura S., Takebe T., Mori K.

Progress in Biomedical Optics and Imaging - Proceedings of SPIE 12468 巻 2023年

出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

In this paper, we propose a scheme that includes automated extraction of thrombus regions and quantitative analysis of thrombosis in confocal laser scanning microscope (CLSM) blood flow image sequence. Making thrombosis model in animal models play an important role in the development of antithrombotic drugs and ascertaining thrombosis mechanisms. Making thrombosis model in cerebral cortex of mice is usually observed using a CLSM in the fluorescence mode. However, some small changes of thrombus regions are not easily observed in CLSM blood flow image sequences. In addition, it is not easy for researchers to quantitatively analyze the degree of thrombosis. Therefore, we propose a scheme to achieve automatic thrombosis region extraction and quantitative analysis. In which, our thrombosis region extraction method uses analysis of changing pattern of thrombosis regions in CLSM blood flow image sequence. Experimental results showed that our scheme can help biological researchers observe and analyze the changes of thrombosis in animal models and reduced the use of fluorescent thrombus markers.

DOI： 10.1117/12.2654632

TriMix: A General Framework for Medical Image Segmentation from Limited Supervision

Zheng Z., Hayashi Y., Oda M., Kitasaka T., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 13846 LNCS 巻頁： 185 - 202 2023年

出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

We present a general framework for medical image segmentation from limited supervision, reducing the reliance on fully and densely labeled data. Our method is simple, jointly trains triple diverse models, and adopts a mix augmentation scheme, and thus is called TriMix. TriMix imposes consistency under a more challenging perturbation, i.e., combining data augmentation and model diversity on the tri-training framework. This straightforward strategy enables TriMix to serve as a strong and general learner learning from limited supervision using different kinds of imperfect labels. We conduct extensive experiments to show TriMix’s generic purpose for semi- and weakly-supervised segmentation tasks. Compared to task-specific state-of-the-arts, TriMix achieves competitive performance and sometimes surpasses them by a large margin. The code is available at https://github.com/MoriLabNU/TriMix.

DOI： 10.1007/978-3-031-26351-4_12

ConDistFL: Conditional Distillation for Federated Learning from Partially Annotated Data

AI画像解析による内視鏡外科手術手技のビデオ評価及び手術支援システムの構築

安井昭洋, 内田広夫, 森健策, 石田昇平, 出家亨一, 檜顕成, 城田千代栄, 小田昌宏, 林雄一郎

生体医工学 Annual61 巻 ( Abstract ) 頁： 127_2 - 127_2 2023年

記述言語：日本語出版者・発行元：公益社団法人日本生体医工学会

【はじめに】術後成長発達する小児患者にとって、低侵襲手術は非常に重要である。しかし患者数は限られているため、しっかりとした手術を行うためにoff the job-training(OJT)が重要である。さらにOJTでの効率的な手技獲得には、手技を客観的に評価しfeed backを行うシステムが必須である。また安全で効率的な内視鏡手術を行うためには、臓器の位置関係の把握が必要であるため、術中ナビゲーションは重要な要件となる。これらの課題に対して、AIを用いた内視鏡手技評価および手術支援システムの構築に着手しており現状の成果を報告する。【方法と結果】食道閉鎖症モデルを用いた吻合手技を被験者に課し、各被験者の手技を最初に人の目で「check 表」「エラー項目」「時間」を用いて評価した。次にビデオから検出した鉗子の動きと人が判定した手技優劣の関係性をAIで学習させ、上位88％・下位95％の精度で手技優劣が自動判定可能となった。この結果を解析することで今まで必要だった50項目以上の肉眼チェックが、わずか７項目チェックするだけで手技の優劣を判断できることが明らかになった。現在食道閉鎖症の手術画像を用いて、食道・迷走神経・気管を深層学習させ、各種構造物の自動認識を進めている。【まとめ】AI画像解析により内視鏡手技の優劣をビデオで判定可能となった。この結果から新たに効率的な手技判断基準を定めることができた。術中ナビゲーションは現在精度のさらなる向上を目指している。

DOI： 10.11239/jsmbe.annual61.127_2

CiNii Research

Wang P., Shen C., Wang W., Oda M., Fuh C.S., Mori K., Roth H.R.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 14393 巻頁： 311 - 321 2023年

出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Developing a generalized segmentation model capable of simultaneously delineating multiple organs and diseases is highly desirable. Federated learning (FL) is a key technology enabling the collaborative development of a model without exchanging training data. However, the limited access to fully annotated training data poses a major challenge to training generalizable models. We propose “ConDistFL”, a framework to solve this problem by combining FL with knowledge distillation. Local models can extract the knowledge of unlabeled organs and tumors from partially annotated data from the global model with an adequately designed conditional probability representation. We validate our framework on four distinct partially annotated abdominal CT datasets from the MSD and KiTS19 challenges. The experimental results show that the proposed framework significantly outperforms FedAvg and FedOpt baselines. Moreover, the performance on an external test dataset demonstrates superior generalizability compared to models trained on each dataset separately. Our ablation study suggests that ConDistFL can perform well without frequent aggregation, reducing the communication cost of FL. Our implementation will be available at https://github.com/NVIDIA/NVFlare/tree/main/research/condist-fl.

DOI： 10.1007/978-3-031-47401-9_30

Masked Frequency Consistency for Domain-Adaptive Semantic Segmentation of Laparoscopic Images

Zhao X., Hayashi Y., Oda M., Kitasaka T., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 14220 LNCS 巻頁： 663 - 673 2023年

出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Semantic segmentation of laparoscopic images is an important issue for intraoperative guidance in laparoscopic surgery. However, acquiring and annotating laparoscopic datasets is labor-intensive, which limits the research on this topic. In this paper, we tackle the Domain-Adaptive Semantic Segmentation (DASS) task, which aims to train a segmentation network using only computer-generated simulated images and unlabeled real images. To bridge the large domain gap between generated and real images, we propose a Masked Frequency Consistency (MFC) module that encourages the network to learn frequency-related information of the target domain as additional cues for robust recognition. Specifically, MFC randomly masks some high-frequency information of the image to improve the consistency of the network’s predictions for low-frequency images and real images. We conduct extensive experiments on existing DASS frameworks with our MFC module and show performance improvements. Our approach achieves comparable results to fully supervised learning method on the CholecSeg8K dataset without using any manual annotation. The code is available at github.com/MoriLabNU/MFC.

DOI： 10.1007/978-3-031-43907-0_63

Multi-view Guidance for Self-supervised Monocular Depth Estimation on Laparoscopic Images via Spatio-Temporal Correspondence

Li W., Hayashi Y., Oda M., Kitasaka T., Misawa K., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 14228 LNCS 巻頁： 429 - 439 2023年

出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

This work proposes an innovative self-supervised approach to monocular depth estimation in laparoscopic scenarios. Previous methods independently predicted depth maps ignoring spatial coherence in local regions and temporal correlation between adjacent images. The proposed approach leverages spatio-temporal coherence to address the challenges of textureless areas and homogeneous colors in such scenes. This approach utilizes a multi-view depth estimation model to guide monocular depth estimation when predicting depth maps. Moreover, the minimum reprojection error is extended to construct a cost volume for the multi-view model using adjacent images. Additionally, a 3D consistency of the point cloud back-projected from predicted depth maps is optimized for the monocular depth estimation model. To benefit from spatial coherence, deformable patch-matching is introduced to the monocular and multi-view models to smooth depth maps in local regions. Finally, a cycled prediction learning for view synthesis and relative poses is designed to exploit the temporal correlation between adjacent images fully. Experimental results show that the proposed method outperforms existing methods in both qualitative and quantitative evaluations. Our code is available at https://github.com/MoriLabNU/MGMDepthL.

DOI： 10.1007/978-3-031-43996-4_41

膀胱鏡画像におけるtiny-YOLOを用いた腫瘍検出

牟田口淳, 小田昌宏, 猪口淳一, 森健策, 江藤正俊

生体医工学 Annual61 巻 ( Abstract ) 頁： 255_2 - 255_2 2023年

記述言語：日本語出版者・発行元：公益社団法人日本生体医工学会

【背景】膀胱癌は経尿道手術後に再発が多い腫瘍であり、膀胱鏡での腫瘍の見落としが原因とされている。内視鏡での観察は、従来の白色光(WLI)の他に、NBIを使用するが、いずれの腫瘍検出精度は検者の技量・経験に依存するため、検査の再現性・客観性が少ないことが課題である。近年、人工知能(AI)が多くの医療分野で活用されており、AIによる検査は、客観性・再現性を持った上で、エキスパートレベルと同程度の診断能を持つ可能性があるとされている。今回、WLI/NBI膀胱鏡画像を用いて、AIによる腫瘍検出の精度を検証した。【方法】2019年から2021年まで、経尿道的膀胱腫瘍切除術(TURBT)の際に、WLI/NBIを用いて観察を行った症例の手術動画から膀胱鏡画像を作成し、腫瘍を含む画像を腫瘍画像、腫瘍を含まない画像を正常画像と定義した。腫瘍画像内の膀胱腫瘍を矩形でアノテーションを行い、テストデータ用の画像を用いてAIによる感度、特異度、陽性的中率を評価した。AIでの物体検出はtiny-YOLOを用い、腫瘍検出精度の検証を行った。【結果】WLIとNBIから、それぞれ腫瘍画像をそれぞれtiny-YOLOで学習を行い、腫瘍画像(WLI: 525枚、NBI:219枚)と正常画像(WLI:98枚、NBI:108枚)で精度検証を行った。AIによる物体検出の感度/特異度/陽性的中率は、WLIで87.8%/88.8%/97.7%、NBIで82.2%/81.4%/90.0%であった。【結論】膀胱鏡画像において、AIにより比較的良好に腫瘍検出が可能であった。更なる精度改善、リアルタイム検出への課題について、文献的考察を加え報告する。

DOI： 10.11239/jsmbe.annual61.255_2

CiNii Research

Boundary-aware feature and prediction refinement for polyp segmentation 査読有り

Qiu Jie, Hayashi Yuichiro, Oda Masahiro, Kitasaka Takayuki, Mori Kensaku

COMPUTER METHODS IN BIOMECHANICS AND BIOMEDICAL ENGINEERING-IMAGING AND VISUALIZATION 2022年12月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Computer Methods in Biomechanics and Biomedical Engineering: Imaging and Visualization

Polyp segmentation from colonoscopy videos is an essential task in medical image processing for detecting early cancer. However, segmenting a precise boundary is still challenging, even with powerful deep neural networks. We consider the difficulty can be caused by: (1) the ambiguity boundary and (2) some complicated shape makes polyps hard to segment. To address these problems, we propose the Boundary-aware Feature and Prediction Refinement framework (BaFPR) for polyp segmentation. Specifically, we design a segmentation decoder for representation learning with boundary prior and propose a novel consistency loss to learn clues from the polar coordinate. The decoder mainly consists of a boundary prior module (BPM) and a bi-directional fusion module (BiFM). BPM is designed to learn the boundary prior, while BiFM learns to fuse representations of BPM and multi-scale representations from an encoder. To handle these complicated shapes of polyps, we maintain an extra segmentation network that learns with polar transformations of data to provide extra clues for the main segmentation network by our proposed consistency loss. We evaluated BaFPR with five challenging datasets for polyp segmentation and the results showed that our proposal consistently improves the segmentation performance of polyps. Code available at: https://github.com/MoriLabNU/BaFPR.

DOI： 10.1080/21681163.2022.2155579

Anatomy aware-based 2.5D bronchoscope tracking for image-guided bronchoscopic navigation 査読有り

Wang Cheng, Oda Masahiro, Hayashi Yuichiro, Kitasaka Takayuki, Itoh Hayato, Honma Hirotoshi, Takebatake Hirotsugu, Mori Masaki, Natori Hiroshi, Mori Kensaku

COMPUTER METHODS IN BIOMECHANICS AND BIOMEDICAL ENGINEERING-IMAGING AND VISUALIZATION 2022年12月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Computer Methods in Biomechanics and Biomedical Engineering: Imaging and Visualization

Physicians use an endoscopic navigation system during bronchoscopy to decrease the risk of getting lost in complex tree-structure like bronchus. Most existing navigation systems based on the camera pose estimated from bronchoscope tracking and/or deep learning. However, bronchoscope tracking-based method exists tracking error, and the pre-training of the model needs massive data. This paper describes an improved bronchoscope tracking procedure by adopting image domain translation technique to improve tracking performance. Specifically, our scheme consists of three modules, an RGB-D image domain translation module, an anatomical structure classification module and a structure-aware bronchoscope tracking module. The RGB-D image domain translation module translates a real bronchoscope (RB) image to its corresponding virtual bronchoscope image and depth image. The anatomical dependency module classifies the current scene into two categories: structureless and rich structure. The bronchoscope tracking module uses a modified video-CT bronchoscope tracking approach to estimate camera pose. Experimental results showed that the proposed method achieved higher tracking accuracy than the current state-of-the-art bronchoscope tracking methods.

DOI： 10.1080/21681163.2022.2152728

Classification and visual explanation for COVID-19 pneumonia from CT images using triple learning. 査読有り

Kato S, Oda M, Mori K, Shimizu A, Otake Y, Hashimoto M, Akashi T, Hotta K

Scientific reports 12 巻 ( 1 ) 頁： 20840 2022年12月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Scientific Reports

This study presents a novel framework for classifying and visualizing pneumonia induced by COVID-19 from CT images. Although many image classification methods using deep learning have been proposed, in the case of medical image fields, standard classification methods are unable to be used in some cases because the medical images that belong to the same category vary depending on the progression of the symptoms and the size of the inflamed area. In addition, it is essential that the models used be transparent and explainable, allowing health care providers to trust the models and avoid mistakes. In this study, we propose a classification method using contrastive learning and an attention mechanism. Contrastive learning is able to close the distance for images of the same category and generate a better feature space for classification. An attention mechanism is able to emphasize an important area in the image and visualize the location related to classification. Through experiments conducted on two-types of classification using a three-fold cross validation, we confirmed that the classification accuracy was significantly improved; in addition, a detailed visual explanation was achieved comparison with conventional methods.

DOI： 10.1038/s41598-022-24936-6

MICCAI 2022参加報告招待有り

小田昌宏

医用画像情報学会雑誌 39 巻 ( 4 ) 頁： 78 - 81 2022年12月

特集整形外科領域における人工知能の応用各論深層学習を用いたMRIでの脊髄腫瘍自動位置検出システムの構築招待有り

伊藤定之, 中島宏彰, 町野正明, 世木直喜, 小田昌宏, 大内田隼, 森下和明, 森健策, 今釜史郎

臨床整形外科 57 巻 ( 10 ) 頁： 1189 - 1195 2022年10月

A skeleton context-aware 3D fully convolutional network for abdominal artery segmentation. 査読有り

Zhu R, Oda M, Hayashi Y, Kitasaka T, Misawa K, Fujiwara M, Mori K

International journal of computer assisted radiology and surgery 2022年10月

Class-wise confidence-aware active learning for laparoscopic images segmentation. 査読有り

Qiu J, Hayashi Y, Oda M, Kitasaka T, Mori K

International journal of computer assisted radiology and surgery 2022年10月

Pattern Analysis of Substantia Nigra in Parkinson Disease by Fifth-Order Tensor Decomposition and Multi-sequence MRI 査読有り

Itoh H., Hu T., Oda M., Saiki S., Kamagata K., Hattori N., Aoki S., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 13594 LNCS 巻頁： 63 - 75 2022年10月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

This work proposes a new feature extraction method to analyse patterns of the substantia nigra in Parkinson disease. Recent imaging techniques such that neuromelanin-sensitive MRI enable us to recognise the region of the substantia nigra and capture early Parkinson-disease-related changes. However, automated feature extraction of Parkinson-disease-related changes and their geometrical interpretation are still challenging. To tackle these challenges, we introduce a fifth-order tensor expression of multi-sequence MRI data such as T1-weighted, T2-weighted, and neuromelanin images and its tensor decomposition. Reconstruction from the selected components of the decomposition visualises the discriminative patterns of the substantia nigra between normal and Parkinson-disease patients. We collected multi-sequence MRI data from 155 patients for experiments. Using the collected data, we validate the proposed method and analyse discriminative patterns in the substantia nigra. Experimental results show that the geometrical interpretation of selected features coincides with neuropathological characteristics.

DOI： 10.1007/978-3-031-18814-5_7

Joint Multi Organ and Tumor Segmentation from Partial Labels Using Federated Learning 査読有り国際共著

Shen C., Wang P., Yang D., Xu D., Oda M., Chen P.T., Liu K.L., Liao W.C., Fuh C.S., Mori K., Wang W., Roth H.R.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 13573 LNCS 巻頁： 58 - 67 2022年10月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Segmentation studies in medical image analysis are always associated with a particular task scenario. However, building datasets to train models to segment multiple types of organs and pathologies is challenging. For example, a dataset annotated for the pancreas and pancreatic tumors will result in a model that cannot segment other organs, like the liver and spleen, visible in the same abdominal computed tomography image. The lack of a well-annotated dataset is one limitation resulting in a lack of universal segmentation models. Federated learning (FL) is ideally suited for addressing this issue in the real-world context. In this work, we show that each medical center can use training data for distinct tasks to collaboratively build more generalizable segmentation models for multiple segmentation tasks without the requirement to centralize datasets in one place. The main challenge of this research is the heterogeneity of training data from various institutions and segmentation tasks. In this paper, we propose a multi-task segmentation framework using FL to learn segmentation models using several independent datasets with different annotations of organs or tumors. We include experiments on four publicly available single-task datasets, including MSD liver (w/ tumor), MSD spleen, MSD pancreas (w/ tumor), and KITS19. Experimental results on an external validation set to highlight the advantages of employing FL in multi-task organ and tumor segmentation.

DOI： 10.1007/978-3-031-18523-6_6

Geometric Constraints for Self-supervised Monocular Depth Estimation on Laparoscopic Images with Dual-task Consistency 査読有り

重症COVID-19肺炎における深層学習を用いたCT画像評価と臨床パラメーターの検討査読有り

坂東皓介, 春日井大介, 後藤縁, 小田昌宏, 森健策

日本救急医学会雑誌 33 巻 ( 10 ) 頁： 760 - 760 2022年10月

特集整形外科領域における人工知能の応用各論深層学習を用いたMRIでの脊髄腫瘍自動位置検出システムの構築招待有り

伊藤定之, 中島宏彰, 町野正明, 世木直喜, 小田昌宏, 大内田隼, 森下和明, 森健策, 今釜史郎

臨床整形外科 57 巻 ( 10 ) 頁： 1189 - 1195 2022年10月

Li Wenda, Hayashi Yuichiro, Oda Masahiro, Kitasaka Takayuki, Misawa Kazunari, Mori Kensaku

MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION, MICCAI 2022, PT IV 13434 巻頁： 467 - 477 2022年9月

Depth values are essential information to automate surgical robots and achieve Augmented Reality technology for minimally invasive surgery. Although depth-pose self-supervised monocular depth estimation performs impressively for autonomous driving scenarios, it is more challenging to predict accurate depth values for laparoscopic images due to the following two aspects: (i) the laparoscope’s motions contain many rotations, leading to pose estimation difficulties for the depth-pose learning strategy; (ii) the smooth surface reduces photometric error even if the matching pixels are inaccurate between adjacent frames. This paper proposes a novel self-supervised monocular depth estimation for laparoscopic images with geometric constraints. We predict the scene coordinates as an auxiliary task and construct dual-task consistency between the predicted depth maps and scene coordinates under a unified camera coordinate system to achieve pixel-level geometric constraints. We extend the pose estimation into a Siamese process to provide stronger and more balanced geometric constraints in a depth-pose learning strategy by leveraging the order of the adjacent frames in a video sequence. We also design a weight mask for depth estimation based on our consistency to alleviate the interference from predictions with low confidence. The experimental results showed that the proposed method outperformed the baseline on depth and pose estimation. Our code is available at https://github.com/MoriLabNU/GCDepthL.

DOI： 10.1007/978-3-031-16440-8_45

Positive-gradient-weighted object activation mapping: visual explanation of object detector towards precise colorectal-polyp localisation 査読有り

Itoh Hayato, Misawa Masashi, Mori Yuichi, Kudo Shin-Ei, Oda Masahiro, Mori Kensaku

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY 2022年8月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：International Journal of Computer Assisted Radiology and Surgery

Purpose: Precise polyp detection and localisation are essential for colonoscopy diagnosis. Statistical machine learning with a large-scale data set can contribute to the construction of a computer-aided diagnosis system for the prevention of overlooking and miss-localisation of a polyp in colonoscopy. We propose new visual explaining methods for a well-trained object detector, which achieves fast and accurate polyp detection with a bounding box towards a precise automated polyp localisation. Method: We refine gradient-weighted class activation mapping for more accurate highlighting of important patterns in processing a convolutional neural network. Extending the refined mapping into multiscaled processing, we define object activation mapping that highlights important object patterns in an image for a detection task. Finally, we define polyp activation mapping to achieve precise polyp localisation by integrating adaptive local thresholding into object activation mapping. We experimentally evaluate the proposed visual explaining methods with four publicly available databases. Results: The refined mapping visualises important patterns in each convolutional layer more accurately than the original gradient-weighted class activation mapping. The object activation mapping clearly visualises important patterns in colonoscopic images for polyp detection. The polyp activation mapping localises the detected polyps in ETIS-Larib, CVC-Clinic and Kvasir-SEG database with mean Dice scores of 0.76, 0.72 and 0.72, respectively. Conclusions: We developed new visual explaining methods for a convolutional neural network by refining and extending gradient-weighted class activation mapping. Experimental results demonstrated the validity of the proposed methods by showing that accurate visualisation of important patterns and localisation of polyps in a colonoscopic image. The proposed visual explaining methods are useful for the interpreting and applying a trained polyp detector.

DOI： 10.1007/s11548-022-02696-y

テンソル分解を用いた黒質緻密部の3次元パターン表現に関する初期的検討

伊東隼人, 小田昌宏, 斉木臣二, 服部信考, 鎌形康司, 青木茂樹, 森健策

日本医用画像工学会大会予稿集 41回巻頁： 124 - 125 2022年7月

記述言語：日本語掲載種別：研究論文（研究会，シンポジウム資料等）出版者・発行元：(一社)日本医用画像工学会

黒質緻密部の3次元パターン表現に関する初期的検討について報告する.黒質緻密部における神経細胞の減少はパーキンソン病疾患者において観察される特徴のひとつである.この黒質緻密部の変化はT2強調画像や神経メラニン強調画像を介して捉えることができ,黒質緻密部の左右差や体積の減少として観察できると報告されている.一方で,得られた画像データを用いて黒質緻密部の3次元パターンの変化を計算機上で解析するためには,何らかの特徴表現が必要となる.本稿では黒質緻密部の神経メラニンMRIの信号強度比をテンソルにて表し,これらテンソルの分解に基づいた特徴表現を提案する.提案した特徴表現にて健常者・パーキンソン病患者のパターン分布を調査した結果,2つのクラスで異なる分布を表現できていることを確認した.(著者抄録)

Comprehensive Diagnostic Performance of Real-Time Characterization of Colorectal Lesions Using an Artificial Intelligence–Assisted System: A Prospective Study 査読有り

Minegishi Y., Kudo S.E., Miyata Y., Nemoto T., Mori K., Misawa M., Mori Y., Mochida K., Akimoto Y., Ishiyama M., Ogura Y., Abe M., Sato Y., Ogawa Y., Yasuharu M., Tanaka K., Ichimasa K., Nakamura H., Ogata N., Hisayuki T., Kudo T., Hayashi T., Wakamura K., Miyachi H., Baba T., Ishida F., Itoh H., Oda M.

Gastroenterology 163 巻 ( 1 ) 頁： 323 - 325.e3 2022年7月

Blood Vessel Segmentation from Low-Contrast and Wide-Field Optical Microscopic Images of Cranial Window by Attention-Gate-Based Network 査読有り

Wu Yunheng, Oda Masahiro, Hayashi Yuichiro, Takebe Takanori, Nagata Shogo, Wang Cheng, Mori Kensaku

2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION WORKSHOPS (CVPRW 2022) 2022-June 巻頁： 1863 - 1872 2022年6月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）出版者・発行元：IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops

The stereomicroscope, which is an optical microscope, is used to observe the organoids cultured in cranial windows. A cranial window is a light accessible observation window made on the brain of mice through craniotomy. Organoids research is often conducted on cranial windows. Hence, the observation of blood vessels in them is important for organoid research, like organoid vascularization. Therefore, achieving a simple, low-cost method that extracts blood vessel structures would significantly help researchers observe the blood vessels in cranial windows from microscopic images. However, wide-field optical microscopic images taken by stereomicroscope suffer from low contrast and dura mater occlusion, complicating the observation of the blood vessels in such images. To address such problems and assist researchers who are observing vascular structures, we propose a method that segments blood vessels in cranial windows from low-contrast and wide-field microscopic images. Our method is based on the Attention U-Net framework and clDice, which considers the connectivity of blood vessels. In addition, for low-contrast and partial occlusion problems, we used contrast enhancement and dehazing as preprocessing steps. Our method achieved a Dice score of 75.56%, a clDice score of 79.95%, and the Accuracy of 91.41% on our microscopic image dataset, suggesting that our method can extract blood vessels from low-contrast and wide-field microscopic images better than other methods.

DOI： 10.1109/CVPRW56347.2022.00203

JACLS ALL-02 SR protocol reduced-intensity chemotherapy produces excellent outcomes in patients with low-risk childhood acute lymphoblastic leukemia

BONE MARROW EDEMA SCORE IN HAND X-RAY FILM BY AI DEEP LEARNING ASSOCIATE WITH MRI BONE EDEMA IN RHEUMATOID ARTHRITIS.

Katayama K., Pan D., Oda M., Okubo T., Mori K.

ANNALS OF THE RHEUMATIC DISEASES 81 巻頁： 1773 - 1774 2022年6月

Takahashi Y., Ishida H., Imamura T., Tamefusa K., Suenobu S., Usami I., Yumura-Yagi K., Hasegawa D., Nishimura S., Suzuki N., Hashii Y., Deguchi T., Moriya-Saito A., Kosaka Y., Kato K., Kobayashi R., Kawasaki H., Hori H., Sato A., Kudo T., Nakahata T., Oda M., Hara J., Horibe K.

International Journal of Hematology 115 巻 ( 6 ) 頁： 890 - 897 2022年6月

出版者・発行元：International Journal of Hematology

Acute lymphoblastic leukemia (ALL) is the most common childhood cancer. As overall cure rates of childhood ALL have improved, reduction of overall treatment intensity while still ensuring excellent outcomes is imperative for low-risk patients. We report the outcomes of patients treated following the standard-risk protocol from the prospective Japan Association of Childhood Leukemia Study (JACLS) ALL-02 study, which was conducted between 2002 and 2008 for patients with newly diagnosed ALL aged 1–18 years. Of 1138 patients with B-cell precursor ALL, 388 (34.1%) were allocated to this protocol. Excellent outcomes were achieved despite the overall treatment intensity being lower than that of most contemporary protocols: 4 years event-free survival (EFS) was 92.3% and 4 years overall survival 98.2%. Patients with high hyperdiploidy (HHD) involving triple trisomy (trisomy of chromosomes 4, 10, and 17) or ETV6-RUNX1 had even better outcomes (4 years EFS 97.6% and 100%, respectively). Unique characteristics of this protocol include a selection of low-risk patients with a low initial WBC count and good early treatment response and reduction of cumulative doses of chemotherapeutic agents while maintaining dose density. In Japan, we are currently investigating the feasibility of this protocol while incorporating minimal residual disease into the patient stratification strategy.

DOI： 10.1007/s12185-022-03315-x

SR-CycleGAN: super-resolution of clinical CT to micro-CT level with multi-modality super-resolution loss. 査読有り

Zheng T, Oda H, Hayashi Y, Moriya T, Nakamura S, Mori M, Takabatake H, Natori H, Oda M, Mori K

Journal of Medical Imaging 9 巻 ( 2 ) 頁： 024003 2022年3月

A cascaded fully convolutional network framework for dilated pancreatic duct segmentation 査読有り国際共著

Shen Chen, Roth Holger R., Hayashi Yuichiro, Oda Masahiro, Miyamoto Tadaaki, Sato Gen, Mori Kensaku

International Journal of Computer Assisted Radiology and Surgery 17 巻 ( 2 ) 頁： 343 - 354 2022年2月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：International Journal of Computer Assisted Radiology and Surgery

Purpose: Pancreatic duct dilation can be considered an early sign of pancreatic ductal adenocarcinoma (PDAC). However, there is little existing research focused on dilated pancreatic duct segmentation as a potential screening tool for people without PDAC. Dilated pancreatic duct segmentation is difficult due to the lack of readily available labeled data and strong voxel imbalance between the pancreatic duct region and other regions. To overcome these challenges, we propose a two-step approach for dilated pancreatic duct segmentation from abdominal computed tomography (CT) volumes using fully convolutional networks (FCNs). Methods: Our framework segments the pancreatic duct in a cascaded manner. The pancreatic duct occupies a tiny portion of abdominal CT volumes. Therefore, to concentrate on the pancreas regions, we use a public pancreas dataset to train an FCN to generate an ROI covering the pancreas and use a 3D U-Net-like FCN for coarse pancreas segmentation. To further improve the dilated pancreatic duct segmentation, we deploy a skip connection on each corresponding resolution level and an attention mechanism in the bottleneck layer. Moreover, we introduce a combined loss function based on Dice loss and Focal loss. Random data augmentation is adopted throughout the experiments to improve the generalizability of the model. Results: We manually created a dilated pancreatic duct dataset with semi-automated annotation tools. Experimental results showed that our proposed framework is practical for dilated pancreatic duct segmentation. The average Dice score and sensitivity were 49.9% and 51.9%, respectively. These results show the potential of our approach as a clinical screening tool. Conclusions: We investigate an automated framework for dilated pancreatic duct segmentation. The cascade strategy effectively improved the segmentation performance of the pancreatic duct. Our modifications to the FCNs together with random data augmentation and the proposed combined loss function facilitate automated segmentation.

DOI： 10.1007/s11548-021-02530-x

Real-time Esophagus Achalasia Detection Method for Esophagoscopy Assistance 査読有り

Jiang Kai, Oda Masahiro, Shiwaku Hironari, Misawa Masashi, Mori Kensaku

MEDICAL IMAGING 2022: COMPUTER-AIDED DIAGNOSIS 12033 巻 2022年2月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

This paper presents an automated real-time esophagus achalasia (achalasia) detection method for esophagoscopy assistance. Achalasia is a well-recognized primary esophageal motor disorder of unknown etiology. To diagnose the achalasia, endoscopic evaluation of the esophagus and stomach is recommended to ensure that there is not a malignancy causing the disease or esophageal squamous cell carcinoma complicating achalasia. However, esophagoscopy is low sensitive in the early-stage of achalasia, only about half of patients with early-stage achalasia can be identified. Thus, a quantitative detection system of real-time esophagoscopy video is required for diagnosis assistance of achalasia. This paper presents to use of a convolutional neural network (CNN) to detect all achalasia frames in esophagoscopy videos. The features of achalasia cannot be easily distinguished. To better extract features from esophagoscopy frames, we introduce dense pooling connections and dilated convolutions in the CNN. We trained and evaluated our network with an original dataset that is extracted from several esophagoscopy videos of achalasia patients. Furthermore, we develop a real-time achalasia detection ComputerAided Diagnosis (CAD) system with the trained network. The CAD system can detect each frame from the input esophagoscopy videos with only 0.1 milliseconds delay. The real-time achalasia detection system achieved 0.872 accuracy, and 0.943 AUC score on our dataset.

DOI： 10.1117/12.2613289

Uncertainty meets 3D-spatial feature in colonoscopic polyp-size determination 査読有り

Itoh Hayato, Oda Masahiro, Jiang Kai, Mori Yuichi, Misawa Masashi, Kudo Shin-Ei, Imai Kenichiro, Ito Sayo, Hotta Kinichi, Mori Kensaku

Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization 2022年1月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Computer Methods in Biomechanics and Biomedical Engineering: Imaging and Visualization

This paper presents a new 3D-spatial feature extraction from a 2D colonoscopic image for polyp-size estimation. The polyp-size estimation poses potential demands on colonoscopy, since an endoscopist’s subjective estimation is apt to result in uncertain polyp-size determination. This uncertain determination is derived from the lack of 3D-spatial information. Even though a previous work clarified that depth estimation mitigates uncertainty in binary polyp-size classification, precise estimation of 3D structure from a colonoscopic image(s) remains an unsolved challenge in medical image analysis. This work proposes a 3D-spatial feature that expresses a polyp’s precise 3D shape to mitigate uncertainty in polyp-size determination. First, we introduce an accurate depth estimation method to capture the 3D structure of a colon. Next, we integrate depth estimation and polyp localisation to extract a 3D polyp shape as a feature. Finally, we achieve polyp-size estimation by statistical learning of extracted features. The experimental results demonstrated the validity both of our depth estimation and 3D-spatial feature. Compared with deep RGB and RGB-D convolutional neural networks (CNNs), a shallow CNN with the proposed 3D-spatial feature achieved a more accurate polyp-size estimation with a mean absolute error of 1.36 mm, whereas the one of the deep CNN is 3.11 mm.

DOI： 10.1080/21681163.2021.2004445

Impact of the clinical use of artificial intelligence-assisted neoplasia detection for colonoscopy: a large-scale prospective propensity score-matched study (with video) 査読有り

Ishiyama Misaki, Kudo Shin-ei, Misawa Masashi, Mori Yuichi, Maeda Yasuhara, Ichimasa Katsuro, Kudo Toyoki, Hayashi Takemasa, Wakamura Kunihiko, Miyachi Hideyuki, Ishida Fumio, Itoh Hayato, Oda Masahiro, Mori Kensaku

Gastrointestinal Endoscopy 95 巻 ( 1 ) 頁： 155 - 163 2022年1月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Gastrointestinal Endoscopy

Background and Aims: Recently, the use of computer-aided detection (CADe) for colonoscopy has been investigated to improve the adenoma detection rate (ADR). We aimed to assess the efficacy of a regulatory-approved CADe in a large-scale study with high numbers of patients and endoscopists. Methods: This was a propensity score–matched prospective study that took place at a university hospital between July 2020 and December 2020. We recruited patients aged ≥20 years who were scheduled for colonoscopy. Patients with polyposis, inflammatory bowel disease, or incomplete colonoscopy were excluded. We used a regulatory-approved CADe system and conducted a propensity score matching–based comparison of the ADR between patients examined with and without CADe as the primary outcome. Results: During the study period, 2261 patients underwent colonoscopy with the CADe system or routine colonoscopy, and 172 patients were excluded in accordance with the exclusion criteria. Thirty endoscopists (9 nonexperts and 21 experts) were involved in this study. Propensity score matching was conducted using 5 factors, resulting in 1836 patients included in the analysis (918 patients in each group). The ADR was significantly higher in the CADe group than in the control group (26.4% vs 19.9%, respectively; relative risk, 1.32; 95% confidence interval, 1.12-1.57); however, there was no significant increase in the advanced neoplasia detection rate (3.7% vs 2.9%, respectively). Conclusions: The use of the CADe system for colonoscopy significantly increased the ADR in a large-scale prospective study including 30 endoscopists (Clinical trial registration number: UMIN000040677.)

DOI： 10.1016/j.gie.2021.07.022

Enhancing Model Generalization for Substantia Nigra Segmentation Using a Test-time Normalization-Based Method

Hu T., Itoh H., Oda M., Hayashi Y., Lu Z., Saiki S., Hattori N., Kamagata K., Aoki S., Kumamaru K.K., Akashi T., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 13437 LNCS 巻頁： 736 - 744 2022年

掲載種別：研究論文（国際会議プロシーディングス）出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Automatic segmentation of substantia nigra (SN), which is Parkinson’s disease-related tissue, is an important step toward accurate computer-aided diagnosis systems. Conventional methods for SN segmentation depend heavily on limited magnetic resonance imaging (MRI) modalities such as neuromelanin and quantitative susceptibility mapping, which require longer imaging times and are rare in public datasets. To enable a multi-modal investigation for SN anatomic alterations based on medical bigdata researches, the need for automated SN segmentation arises from commonly investigated T2-weighted MRIs. To improve the performance of the automated SN segmentation from a T2-weighted MRI and enhance the model generalization for cross-center researches, this paper proposes a novel test-time normalization (TTN) method to increase the geometric and intensity similarity between the query data and the model’s trained data. Our proposed method requires no additional training procedure or extra annotation for the unseen data. Our results showed that our proposed TTN achieved a mean Dice score of 71.08% in comparison with the baseline model’s 69.87% score with in-house dataset. Additionally, improved SN segmentation performance was observed from the unseen and unlabeled datasets.

DOI： 10.1007/978-3-031-16449-1_70

その他リンク： https://dblp.uni-trier.de/db/conf/miccai/miccai2022-7.html#HuIOHLSHKAKAM22

Spatially variant biases considered self-supervised depth estimation based on laparoscopic videos 査読有り

Li Wenda, Hayashi Yuichiro, Oda Masahiro, Kitasaka Takayuki, Misawa Kazunari, Mori Kensaku

Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization 2021年12月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Computer Methods in Biomechanics and Biomedical Engineering: Imaging and Visualization

Depth estimation is an essential tool in obtaining depth information for robotic surgery and augmented reality technology in the current laparoscopic surgery robot system. Since there is a lack of ground-truth for depth values and laparoscope motions during operation, depth estimation networks have difficulties in predicting depth maps from laparoscopic images under the supervised strategy. It is challenging to generate the correct depth maps for the different environments from abdominal images. To tackle these problems, we propose a novel monocular self-supervised depth estimation network with sparse nest architecture. We design a non-local block to capture broader and deeper context features that can further enhance the scene-variant generalisation capacity of the network for the differences in datasets. Moreover, we introduce an improved multi-mask feature in the loss function to tackle the classical occlusion problem based on the time-series information from stereo videos. We also use heteroscedastic aleatoric uncertainty to reduce the effect of noisy data for depth estimation. We compared our proposed method with other existing methods for different scenes in datasets. The experimental results show that the proposed model outperformed the state-of-the-art models qualitatively and quantitatively.

DOI： 10.1080/21681163.2021.2015723

Depth estimation from single-shot monocular endoscope image using image domain adaptation and edge-aware depth estimation 査読有り

Oda Masahiro, Itoh Hayato, Tanaka Kiyohito, Takabatake Hirotsugu, Mori Masaki, Natori Hiroshi, Mori Kensaku

Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization 2021年12月

担当区分：筆頭著者記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Computer Methods in Biomechanics and Biomedical Engineering: Imaging and Visualization

We propose a depth estimation method from a single-shot monocular endoscopic image using Lambertian surface translation by domain adaptation and depth estimation using multi-scale edge loss. We employ a two-step estimation process including Lambertian surface translation from unpaired data and depth estimation. The texture and specular reflection on the surface of an organ reduce the accuracy of depth estimations. We apply Lambertian surface translation to an endoscopic image to remove these texture and reflections. Then, we estimate the depth by using a fully convolutional network (FCN). During the training of the FCN, improvement of the object edge similarity between an estimated image and a ground truth depth image is important for getting better results. We introduced a muti-scale edge loss function to improve the accuracy of depth estimation. We quantitatively evaluated the proposed method using real colonoscopic images. The estimated depth values were proportional to the real depth values. Furthermore, we applied the estimated depth images to automated anatomical location identification of colonoscopic images using a convolutional neural network. The identification accuracy of the network improved from 69.2% to 74.1% by using the estimated depth images.

DOI： 10.1080/21681163.2021.2012835

Attention-Guided Pancreatic Duct Segmentation from Abdominal CT Volumes 査読有り国際共著

Shen C., Roth H.R., Yuichiro H., Oda M., Miyamoto T., Sato G., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 12969 LNCS 巻頁： 46 - 55 2021年11月

Pancreatic duct dilation indicates a high risk of pancreatic ductal adenocarcinoma (PDAC), the deadliest cancer with a poor prognosis. Segmentation of dilated pancreatic duct from CT taken from patients without PDAC shows the potential to assist the early detection of PDAC. Most current researches include pancreatic duct segmentation as one additional class for patients who have already detected PDAC. However, the dilated pancreatic duct for people who have not yet developed PDAC is typically much smaller, making the segmentation difficult. Deep learning-based segmentation on tiny components is challenging because of the large imbalance between the target object and irrelevant regions. In this work, we explore an attention-guided approach for dilated pancreatic duct segmentation as a screening tool for pre-PDAC patients, enhancing the pancreas regions’ concentration and ignoring the unnecessary features. We employ a multi-scale aggregation to combine the information at different scales to improve the segmentation performance further. Our proposed multi-scale pancreatic attention-guided approach achieved a Dice score of 54.16% on dilated pancreatic duct dataset, which shows a significant improvement over prior techniques.

DOI： 10.1007/978-3-030-90874-4_5

Multi-task Federated Learning for Heterogeneous Pancreas Segmentation 査読有り国際共著

Shen C., Wang P., Roth H.R., Yang D., Xu D., Oda M., Wang W., Fuh C.S., Chen P.T., Liu K.L., Liao W.C., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 12969 LNCS 巻頁： 101 - 110 2021年11月

Federated learning (FL) for medical image segmentation becomes more challenging in multi-task settings where clients might have different categories of labels represented in their data. For example, one client might have patient data with “healthy” pancreases only while datasets from other clients may contain cases with pancreatic tumors. The vanilla federated averaging algorithm makes it possible to obtain more generalizable deep learning-based segmentation models representing the training data from multiple institutions without centralizing datasets. However, it might be sub-optimal for the aforementioned multi-task scenarios. In this paper, we investigate heterogeneous optimization methods that show improvements for the automated segmentation of pancreas and pancreatic tumors in abdominal CT images with FL settings.

DOI： 10.1007/978-3-030-90874-4_10

Intestine Segmentation with Small Computational Cost for Diagnosis Assistance of Ileus and Intestinal Obstruction 査読有り

Oda H., Hayashi Y., Kitasaka T., Takimoto A., Hinoki A., Uchida H., Suzuki K., Oda M., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 12969 LNCS 巻頁： 3 - 12 2021年11月

This paper proposes an intestine segmentation method from CT volumes for the intestinal obstruction and the ileus diagnosis assistance. The previous method was built based on the 3D U-Net, whose computational cost was high. Nevertheless, there was no confirmation that the 3-dimensional network contributed to the segmentation performance. In this paper, we propose a method utilizing the 2D U-Net on behalf of the previous methods’ 3D U-Net. Experimental results using 110 CT volumes showed that both the proposed (2D U-Net) and previous (3D U-Net) methods achieved similar scores for the segmentation accuracy and system usefulness. In addition, the proposed method’s inference was 0.5 min on average, around 8-times faster than the 3D U-Net’s. Although subsequent processes still require more than 20 min for each case, utilizing lightweight networks is essential for practical use in the future, especially for emergency diagnosis.

DOI： 10.1007/978-3-030-90874-4_1

COVID-19 Infection Segmentation from Chest CT Images Based on Scale Uncertainty 査読有り

Oda M., Zheng T., Hayashi Y., Otake Y., Hashimoto M., Akashi T., Aoki S., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 12969 LNCS 巻頁： 88 - 97 2021年11月

担当区分：筆頭著者,　責任著者記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

This paper proposes a segmentation method of infection regions in the lung from CT volumes of COVID-19 patients. COVID-19 spread worldwide, causing many infected patients and deaths. CT image-based diagnosis of COVID-19 can provide quick and accurate diagnosis results. An automated segmentation method of infection regions in the lung provides a quantitative criterion for diagnosis. Previous methods employ whole 2D image or 3D volume-based processes. Infection regions have a considerable variation in their sizes. Such processes easily miss small infection regions. Patch-based process is effective for segmenting small targets. However, selecting the appropriate patch size is difficult in infection region segmentation. We utilize the scale uncertainty among various receptive field sizes of a segmentation FCN to obtain infection regions. The receptive field sizes can be defined as the patch size and the resolution of volumes where patches are clipped from. This paper proposes an infection segmentation network (ISNet) that performs patch-based segmentation and a scale uncertainty-aware prediction aggregation method that refines the segmentation result. We design ISNet to segment infection regions that have various intensity values. ISNet has multiple encoding paths to process patch volumes normalized by multiple intensity ranges. We collect prediction results generated by ISNets having various receptive field sizes. Scale uncertainty among the prediction results is extracted by the prediction aggregation method. We use an aggregation FCN to generate a refined segmentation result considering scale uncertainty among the predictions. In our experiments using 199 chest CT volumes of COVID-19 cases, the prediction aggregation method improved the dice similarity score from 47.6% to 62.1%.

DOI： 10.1007/978-3-030-90874-4_9

Aorta-aware GAN for non-contrast to artery contrasted CT translation and its application to abdominal aortic aneurysm detection. 査読有り

Hu T, Oda M, Hayashi Y, Lu Z, Kumamaru KK, Akashi T, Aoki S, Mori K

International Journal of Computer Assisted Radiology and Surgery 17 巻 ( 1 ) 頁： 97 - 105 2021年10月

Graph Cuts Loss to Boost Model Accuracy and Generalizability for Medical Image Segmentation 査読有り

Zheng Zhou, Oda Masahiro, Mori Kensaku

2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION WORKSHOPS (ICCVW 2021) 2021-October 巻頁： 3297 - 3306 2021年10月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Proceedings of the IEEE International Conference on Computer Vision

Segmentation accuracy and generalization ability are essential for deep learning models, especially in medical image segmentation. We present a novel, robust yet straightforward loss function to boost model accuracy and generalizability for medical image segmentation. We reformulate the graph cuts cost function to a loss function for supervised learning. The graph cuts loss innately focuses on a dual penalty to optimize the regional properties and boundary regularization. We benchmark the proposed loss on six public retinal vessel segmentation datasets with a comprehensive intra-dataset and cross-dataset evaluation. Results reveal that the proposed loss is more generalizable, narrowing the performance gap between different architectures. Besides, models trained with our loss show higher segmentation accuracy and better generalization ability than those trained with other mainstream losses. Moreover, we extend our loss to other segmentation tasks, e.g., left atrium and liver tumor segmentation. The proposed loss still achieves comparable performance to the state-of-the-art, demonstrating its potential for any N-D segmentation problem. The code is available at https://github.com/zzh_enggit/graphcutsloss.

DOI： 10.1109/ICCVW54120.2021.00369

Super-Resolution by Latent Space Exploration: Training with Poorly-Aligned Clinical and Micro CT Image Dataset 査読有り

Zheng T., Oda H., Hayashi Y., Nakamura S., Oda M., Mori K.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 12965 LNCS 巻頁： 24 - 33 2021年9月

This paper proposes a super-resolution (SR) method, for performing SR on a poorly-aligned dataset. Super-resolution methods commonly needs aligned low-resolution (LR) and high-resolution (HR) images for training. For obtaining paired LR and HR images in medical imaging, we need to align low and high-resolution data using image registration technology. However, since the hardness of aligning LR and HR images, the aligned LR-HR dataset is always low quality. Conventional SR methods always fail to train using poorly-aligned datasets since these methods need high-quality LR-HR datasets. To tackle this problem, we propose a two-step framework for SR using poorly-aligned datasets. In the first step, we decompose image representation into two parts: one is a content code that captures the image content; the other is a style code that captures the image style and anatomy difference between LR / HR images. To perform SR of a given LR image, we input the content code and a latent variable simultaneously into the SR network to obtain an SR result. In the second step, using the trained SR network and an LR image, we search for a content code, and a style code for generating the most proper SR image. This is conducted by searching for the best content code and the best style code by latent space exploration. We conducted experiments using a poorly-aligned clinical-micro CT lung specimen dataset. Experimental results illustrated the proposed method outperformed conventional SR methods by increasing SSIM from 0.309 to 0.312, and have much more convincing perceptual quality than conventional SR methods.

DOI： 10.1007/978-3-030-87592-3_3

Performance improvement of weakly supervised fully convolutional networks by skip connections for brain structure segmentation 査読有り

Sugino Takaaki, Roth Holger R., Oda Masahiro, Kin Taichi, Saito Nobuhito, Nakajima Yoshikazu, Mori Kensaku

Medical Physics 48 巻 ( 11 ) 頁： 7215 - 7227 2021年9月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Medical Physics

Purpose: For the planning and navigation of neurosurgery, we have developed a fully convolutional network (FCN)-based method for brain structure segmentation on magnetic resonance (MR) images. The capability of an FCN depends on the quality of the training data (i.e., raw data and annotation data) and network architectures. The improvement of annotation quality is a significant concern because it requires much labor for labeling organ regions. To address this problem, we focus on skip connection architectures and reveal which skip connections are effective for training FCNs using sparsely annotated brain images. Methods: We tested 2D FCN architectures with four different types of skip connections. The first was a U-Net architecture with horizontal skip connections that transfer feature maps at the same scale from the encoder to the decoder. The second was a U-Net++ architecture with dense convolution layers and dense horizontal skip connections. The third was a full-resolution residual network (FRRN) architecture with vertical skip connections that pass feature maps between each downsampled scale path and the full-resolution scale path. The last one was a hybrid architecture with a combination of horizontal and vertical skip connections. We validated the effect of skip connections on medical image segmentation from sparse annotation based on these four FCN architectures, which were trained under the same conditions. Results: For multiclass segmentation of the cerebrum, cerebellum, brainstem, and blood vessels from sparsely annotated MR images, we performed a comparative evaluation of segmentation performance among the above four FCN approaches: U-Net, U-Net++, FRRN, and hybrid architectures. The experimental results show that the horizontal skip connections in the U-Net architectures were effective for the segmentation of larger sized objects, whereas the vertical skip connections in the FRRN architecture improved the segmentation of smaller sized objects. The hybrid architecture with both horizontal and vertical skip connections achieved the best results of the four FCN architectures. We then performed an ablation study to explore which skip connections in the FRRN architecture contributed to the improved segmentation of blood vessels. In the ablation study, we compared the segmentation performance between architectures with a horizontal path (HP), an HP and vertical up paths (HP+VUPs), an HP and vertical down paths (HP+VDPs), and an HP and vertical up and down paths (FRRN). We found that the vertical up paths were effective in improving the segmentation of smaller sized objects. Conclusions: This paper investigated which skip connection architectures were effective for multiclass brain segmentation from sparse annotation. Consequently, using vertical skip connections with horizontal skip connections allowed FCNs to improve segmentation performance.

DOI： 10.1002/mp.15192

Binary polyp-size classification based on deep-learned spatial information 査読有り

Itoh Hayato, Oda Masahiro, Jiang Kai, Mori Yuichi, Misawa Masashi, Kudo Shin-Ei, Imai Kenichiro, Ito Sayo, Hotta Kinichi, Mori Kensaku

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERYInternational Journal of Computer Assisted Radiology and Surgery 16 巻頁： 1817 - 1828 2021年9月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：International Journal of Computer Assisted Radiology and Surgery

Purpose: The size information of detected polyps is an essential factor for diagnosis in colon cancer screening. For example, adenomas and sessile serrated polyps that are ≥ 10 mm are considered advanced, and shorter surveillance intervals are recommended for smaller polyps. However, sometimes the subjective estimations of endoscopists are incorrect and overestimate the sizes. To circumvent these difficulties, we developed a method for automatic binary polyp-size classification between two polyp sizes: from 1 to 9 mm and ≥ 10 mm. Method: We introduce a binary polyp-size classification method that estimates a polyp’s three-dimensional spatial information. This estimation is comprised of polyp localisation and depth estimation. The combination of location and depth information expresses a polyp’s three-dimensional shape. In experiments, we quantitatively and qualitatively evaluate the proposed method using 787 polyps of both protruded and flat types. Results: The proposed method’s best classification accuracy outperformed the fine-tuned state-of-the-art image classification methods. Post-processing of sequential voting increased the classification accuracy and achieved classification accuracy of 0.81 and 0.88 for polyps ranging from 1 to 9 mm and others that are ≥ 10 mm. Qualitative analysis revealed the importance of polyp localisation even in polyp-size classification. Conclusions: We developed a binary polyp-size classification method by utilising the estimated three-dimensional shape of a polyp. Experiments demonstrated accurate classification for both protruded- and flat-type polyps, even though the flat type have ambiguous boundary between a polyp and colon wall.

DOI： 10.1007/s11548-021-02477-z

Depth-based branching level estimation for bronchoscopic navigation 査読有り

Wang Cheng, Hayashi Yuichiro, Oda Masahiro, Kitasaka Takayuki, Takabatake Hirotsugu, Mori Masaki, Honma Hirotoshi, Natori Hiroshi, Mori Kensaku

International Journal of Computer Assisted Radiology and Surgery 16 巻 ( 10 ) 頁： 1795 - 1804 2021年8月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：International Journal of Computer Assisted Radiology and Surgery

Purpose: Bronchoscopists rely on navigation systems during bronchoscopy to reduce the risk of getting lost in the complex bronchial tree-like structure and the homogeneous bronchus lumens. We propose a patient-specific branching level estimation method for bronchoscopic navigation because it is vital to identify the branches being examined in the bronchus tree during examination. Methods: We estimate the branching level by integrating the changes in the number of bronchial orifices and the camera motions among the frames. We extract the bronchial orifice regions from a depth image, which is generated using a cycle generative adversarial network (CycleGAN) from real bronchoscopic images. We calculate the number of orifice regions using the vertical and horizontal projection profiles of the depth images and obtain the camera-moving direction using the feature point-based camera motion estimation. The changes in the number of bronchial orifices are combined with the camera-moving direction to estimate the branching level. Results: We used three in vivo and one phantom case to train the CycleGAN model and four in vivo cases to validate the proposed method. We manually created the ground truth of the branching level. The experimental results showed that the proposed method can estimate the branching level with an average accuracy of 87.6%. The processing time per frame was about 61 ms. Conclusion: Experimental results show that it is feasible to estimate the branching level using the number of bronchial orifices and camera-motion estimation from real bronchoscopic images.

DOI： 10.1007/s11548-021-02460-8

Unsupervised colonoscopic depth estimation by domain translations with a Lambertian-reflection keeping auxiliary task 査読有り

深層学習を用いた医用画像処理と画像を超えた応用招待有り

小田昌宏

生体医工学 59 巻 ( 0 ) 頁： 256 - 256 2021年6月

担当区分：筆頭著者記述言語：日本語掲載種別：研究論文（研究会，シンポジウム資料等）出版者・発行元：公益社団法人日本生体医工学会

深層学習は医用画像処理において重要な手法の一つとなった．医用画像処理に含まれる画像分類，検出，セグメンテーションにおいては深層学習が多くの問題を解決し，高い成果を挙げている．画像モダリティに着目すると，眼底画像や放射線画像を対象とした深層学習の適用例が多く，実用化レベルに達したものもある．本発表では，手法や画像モダリティの面から深層学習を用いた医用画像処理を概観し，応用例を紹介する．また，深層学習の新たな応用についても紹介する．深層学習は画像に限らず様々なデータを説明変数と目的変数とすることが可能である．この性質を利用し，画像，テキスト，センサ計測データ等を用いる新たな手法を構築できる．応用例と共に深層学習の新たな可能性について考える．

DOI： 10.11239/jsmbe.Annual59.256

Artificial intelligence-assisted colonic endocytoscopy for cancer recognition: a multicenter study 査読有り

Yuichi Mori , Shin-ei Kudo , Masashi Misawa , Kinichi Hotta , Ohtsuka Kazuo , Shoichi Saito , Hiroaki Ikematsu , Yutaka Saito , Takahisa Matsuda , Takeda Kenichi , Toyoki Kudo , Tetsuo Nemoto , Hayato Itoh , Kensaku Mori

Endoscopy International Open 頁： E1004 - E1011 2021年6月

Itoh Hayato, Oda Masahiro, Mori Yuichi, Misawa Masashi, Kudo Shin-Ei, Imai Kenichiro, Ito Sayo, Hotta Kinichi, Takabatake Hirotsugu, Mori Masaki, Natori Hiroshi, Mori Kensaku

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERYInternational Journal of Computer Assisted Radiology and Surgery 16 巻 ( 6 ) 頁： 989 - 1001 2021年6月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：International Journal of Computer Assisted Radiology and Surgery

Purpose: A three-dimensional (3D) structure extraction technique viewed from a two-dimensional image is essential for the development of a computer-aided diagnosis (CAD) system for colonoscopy. However, a straightforward application of existing depth-estimation methods to colonoscopic images is impossible or inappropriate due to several limitations of colonoscopes. In particular, the absence of ground-truth depth for colonoscopic images hinders the application of supervised machine learning methods. To circumvent these difficulties, we developed an unsupervised and accurate depth-estimation method. Method: We propose a novel unsupervised depth-estimation method by introducing a Lambertian-reflection model as an auxiliary task to domain translation between real and virtual colonoscopic images. This auxiliary task contributes to accurate depth estimation by maintaining the Lambertian-reflection assumption. In our experiments, we qualitatively evaluate the proposed method by comparing it with state-of-the-art unsupervised methods. Furthermore, we present two quantitative evaluations of the proposed method using a measuring device, as well as a new 3D reconstruction technique and measured polyp sizes. Results: Our proposed method achieved accurate depth estimation with an average estimation error of less than 1 mm for regions close to the colonoscope in both of two types of quantitative evaluations. Qualitative evaluation showed that the introduced auxiliary task reduces the effects of specular reflections and colon wall textures on depth estimation and our proposed method achieved smooth depth estimation without noise, thus validating the proposed method. Conclusions: We developed an accurate depth-estimation method with a new type of unsupervised domain translation with the auxiliary task. This method is useful for analysis of colonoscopic images and for the development of a CAD system since it can extract accurate 3D information.

DOI： 10.1007/s11548-021-02398-x

Extremely imbalanced Subarachnoid Hemorrhage detection based on DenseNet-LSTM network with Class-Balanced Loss and Transfer Learning 査読有り

Deep learning system for automatic detection of bladder tumors in cystoscopic images 査読有り

Mutaguchi J., Oda M., Ueda S., Kinoshita F., Naganuma H., Matsumoto T., Lee K., Monji K., Kashiwagi K., Takeuchi A., Shiota M., Inokuchi J., Mori K., Eto M.

EUROPEAN UROLOGY 79 巻頁： S1022 - S1023 2021年6月

COVID-19診断支援AI開発における名古屋大学の取り組み招待有り

小田昌宏, 鄭通, 林雄一郎, 森健策

Medical Imaging Technology 39 巻 ( 1 ) 頁： 13 - 19 2021年2月

Lu Zhongyang, Oda Masahiro, Hayashi Yuichiro, Hu Tao, Itoh Hayato, Watadani Takeyuki, Abe Osamu, Hashimoto Masahiro, Jinzaki Masahiro, Mori Kensaku

MEDICAL IMAGING 2021: COMPUTER-AIDED DIAGNOSIS 11597 巻 2021年2月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）出版者・発行元：Progress in Biomedical Optics and Imaging - Proceedings of SPIE

Subarachnoid Hemorrhage (SAH) detection is a critical, severe problem that confused clinical residents for a long time. With the rise of deep learning technologies, SAH detection made a significant breakthrough in recent ten years. Whereas, the performances are significantly degraded on imbalanced data, makes deep learning models have always suffered criticism. In this study, we present a DenseNet-LSTM network with Class-Balanced Loss and the transfer learning strategy to solve the SAH detection problem on an extremely imbalanced dataset. Compared to the previous works, the proposed framework not merely effectively integrate greyscale features the and spatial information from the consecutive CT scans, but also employ Class-Balanced loss and transfer learning to alleviate the adverse effects and broaden feature diversity respectively on an extreme SAH cases scarcity dataset, mimicking the actual situation of emergency departments. Comprehensive experiments are conducted on a dataset, consisted of 2,519 cases without hemorrhage cases and only 33 cases with SAH. Experimental results demonstrate the F-measure score of SAH detection achieved a remarkable improvement, the backbone DenseNet121 gained around 33% promotion after transfer learning, and on this basis, importing the Class-Balanced Loss and the LSTM structure, the F-measure score further increased 6.1% and 2.7% sequentially.

DOI： 10.1117/12.2582088

Automated Detection of Spinal Schwannomas Utilizing Deep Learning Based on Object Detection from MRI 査読有り

Sadayuki Ito, Kei Ando, Kazuyoshi Kobayashi, Hiroaki Nakashima Masahiro Oda, Masaaki Machino, Shunsuke Kanbara, Taro Inoue, Hidetoshi Yamaguchi, Hiroyuki Koshimizu, Kensaku Mori, Naoki Ishiguro, Shiro Imagama

Spine 46 巻 ( 2 ) 頁： 95 - 100 2021年1月

Current status and future perspective on artificial intelligence for lower endoscopy 査読有り

Misawa Masashi, Kudo Shin-ei, Mori Yuichi, Maeda Yasuharu, Ogawa Yushi, Ichimasa Katsuro, Kudo Toyoki, Wakamura Kunihiko, Hayashi Takemasa, Miyachi Hideyuki, Baba Toshiyuki, Ishida Fumio, Itoh Hayato, Oda Masahiro, Mori Kensaku

DIGESTIVE ENDOSCOPY 33 巻 ( 2 ) 頁： 273 - 284 2021年1月

医療支援AIの現状と情報横断型AIへの期待招待有り

小田昌宏

日本整形外科学会雑誌 95 巻 ( 1 ) 頁： 3 - 8 2021年1月

Current status and future perspective on artificial intelligence for lower endoscopy 査読有り

Misawa M., Kudo S.E., Mori Y., Maeda Y., Ogawa Y., Ichimasa K., Kudo T., Wakamura K., Hayashi T., Miyachi H., Baba T., Ishida F., Itoh H., Oda M., Mori K.

Gastroenterological Endoscopy 63 巻 ( 7 ) 頁： 1402 - 1416 2021年1月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Gastroenterological Endoscopy

The global incidence and mortality rate of colorectal cancer remains high. Colonoscopy is regarded as the gold standard examination for detecting and eradicating neoplastic lesions. However, there are some uncertainties in colonoscopy practice that are related to limitations in human performance. First, approximately one-fourth of colorectal neoplasms are missed on a single colonoscopy. Second, it is still difficult for nonexperts to perform adequately regarding optical biopsy. Third, recording of some quality indicators (e.g. cecal intubation, bowel preparation, and withdrawal speed）which are related to adenoma detection rate, is sometimes incomplete. With recent improvements in machine learning techniques and advances in computer performance, artificial intelligence-assisted computer-aided diagnosis is being increasingly utilized by endoscopists. In particular, the emergence of deep-learning, data-driven machine learning techniques have made the development of computer-aided systems easier than that of conventional machine learning techniques, the former currently being considered the standard artificial intelligence engine of computer-aided diagnosis by colonoscopy. To date, computer-aided detection systems seem to have improved the rate of detection of neoplasms. Additionally, computer-aided characterization systems may have the potential to improve diagnostic accuracy in real-time clinical practice. Furthermore, some artificial intelligence-assisted systems that aim to improve the quality of colonoscopy have been reported. The implementation of computer-aided system clinical practice may provide additional benefits such as helping in educational poorly performing endoscopists and supporting real-time clinical decision-making. In this review, we have focused on computer-aided diagnosis during colonoscopy reported by gastroenterologists and discussed its status, limitations, and future prospects.

DOI： 10.11280/gee.63.1402

ディープラーニングによる物体検出：膀胱内視鏡における膀胱癌診断

第1回JSCAS AI Challenge開催報告招待有り

森健策, 小田昌宏, 平尾彰浩, 澁澤喜人, 竹下修由, 伊藤雅昭

日本コンピュータ外科学会誌 23 巻 ( 2 ) 頁： 74 - 75 2021年

牟田口淳, 小田昌宏, 小林聡, 猪口淳一, 森健策, 江藤正俊

生体医工学 59 巻 ( 0 ) 頁： 299 - 299 2021年

記述言語：日本語出版者・発行元：公益社団法人日本生体医工学会

【背景】膀胱癌は経尿道手術後に再発が多い腫瘍であり、膀胱鏡での腫瘍の見落としが原因とされている。従来の白色光(WLI)に加え、NBIは腫瘍検出の精度が改善する報告があるが、検者の経験に依存するため、再現性・客観性が乏しいことが課題である。近年、客観的な診断ツールとして人工知能(AI)が活用されているが、膀胱鏡においてAIによる腫瘍検出の研究はあまり行われていない。今回、我々はWLI/NBI膀胱鏡画像を用いて、AIによる腫瘍検出の精度を検証した。【方法】2019年12月から2020年11月まで、経尿道的膀胱腫瘍切除術(TURBT)の際に、WLI/NBIを用いて観察を行った症例の手術動画から膀胱鏡画像を作成し、腫瘍を含む画像を腫瘍画像、腫瘍を含まない画像を正常画像と定義した。腫瘍画像を学習データ、テストデータに分類し、AIによる感度、特異度、陽性的中率を評価した。AIでの物体検出はtiny-YOLOを用い、医師個人のコンピューター上で環境構築を行った。【結果】腫瘍画像(WLI:809枚、NBI:684枚)をそれぞれtiny-YOLOで学習を行い、腫瘍画像(WLI:280枚、NBI:87枚)と正常画像(WLI:104枚、NBI:104枚)で精度検証を行った。AIによる物体検出の感度/特異度/陽性的中率は、WLIで84.6%/76.9%/90.8%、NBIで78.1%/97.1%/95.8%であった。【結論】膀胱鏡画像において、AIにより比較的良好に腫瘍検出が可能であった。更なる精度改善、リアルタイム検出への課題について、文献的考察を加え報告する。

DOI： 10.11239/jsmbe.Annual59.299

深層学習による画像認識入門（2） GPU 環境構築と画像の領域分割招待有り

原武史, 小田昌宏

Medical Imaging Technology 39 巻 ( 3 ) 頁： 124 - 130 2021年

記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：日本医用画像工学会

深層学習において本格的な実験や研究を行うためには，GPUを使った環境構築が必要となる．これは，CPUのみで計算した場合と比較して数十倍の計算速度が得られるためである．今回の講座では，GPU環境の構築方法を説明し，画像の領域分割を例にして，計算時間がどの程度変化するかを明らかにする．同時に，U-Netを用いた領域分割の方法，その正解データの構築についても言及する．

DOI： 10.11409/mit.39.124

特集／説明可能なAIの実現に向けて―序文― 招待有り

藤吉弘亘, 小田昌宏

Medical Imaging Technology 39 巻 ( 3 ) 頁： 97 - 98 2021年

Robust endocytoscopic image classification based on higher-order symmetric tensor analysis and multi-scale topological statistics 査読有り

Hayato Itoh, Yukitaka Nimura, Yuichi Mori, Masashi Misawa, Shin-Ei Kudo, Kinichi Hotta, Kazuo Ohtsuka, Shoichi Saito, Yutaka Saito, Hiroaki Ikematsu, Yuichiro Hayashi, Masahiro Oda, Kensaku Mori

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY 15 巻 ( 12 ) 頁： 2049 - 2059 2020年12月

Synthetic laparoscopic video generation for machine learning-based surgical instrument segmentation from real laparoscopic video and virtual surgical instruments 査読有り

Ozawa Takuya, Hayashi Yuichiro, Oda Hirohisa, Oda Masahiro, Kitasaka Takayuki, Takeshita Nobuyoshi, Ito Masaaki, Mori Kensaku

COMPUTER METHODS IN BIOMECHANICS AND BIOMEDICAL ENGINEERING-IMAGING AND VISUALIZATION 2020年11月

大腸鏡画像にYOLO-v3を用いた、穿孔の検出と局在化の比率損失設計に関する予備的研究(Preliminary Study of Loss-Functions Design for Detection and Localization of Perforations with YOLO-v3 in Colonoscopic Images)

蒋凱, 伊東隼人, 小田昌宏, 奥村大志, 森悠一, 三澤将史, 林武雅, 工藤進英, 森健策

日本コンピュータ外科学会誌 22 巻 ( 4 ) 頁： 348 - 349 2020年11月

SUN database 大腸ポリープ自動検出器の精度評価に向けた試験用画像

伊東隼人, 三澤将史, 森悠一, 小田昌宏, 工藤進英, 森健策

日本コンピュータ外科学会誌 22 巻 ( 4 ) 頁： 346 - 347 2020年11月

【多元計算解剖学の診断・治療・医工学への展開】人工知能に基づく医療機器EndoBRAINの臨床導入薬事承認取得・保険算定への挑戦

森悠一, 工藤進英, 三澤将史, 伊東隼人, 小田昌宏, 森健策

MEDICAL IMAGING TECHNOLOGY 38 巻 ( 5 ) 頁： 213 - 216 2020年11月

腹腔鏡動画像用オンラインアノテーションツールの開発

屠芸豪, 伊東隼人, 小澤卓也, 小田昌宏, 竹下修由, 伊藤雅昭, 森健策

日本コンピュータ外科学会誌 22 巻 ( 4 ) 頁： 306 - 307 2020年11月

Artificial Intelligence System to Determine Risk of T1 Colorectal Cancer Metastasis to Lymph Node. 査読有り

Kudo SE, Ichimasa K, Villard B, Mori Y, Misawa M, Saito S, Hotta K, Saito Y, Matsuda T, Yamada K, Mitani T, Ohtsuka K, Chino A, Ide D, Imai K, Kishida Y, Nakamura K, Saiki Y, Tanaka M, Hoteya S, Yamashita S, Kinugasa Y, Fukuda M, Kudo T, Miyachi H, Ishida F, Itoh H, Oda M, Mori K

Gastroenterology 2020年9月

Prediction of dose distribution from luminescence image of water using a deep convolutional neural network for particle therapy 査読有り

Takuya Yabe, Seiichi Yamamoto, Masahiro Oda, Kensaku Mori, Toshiyuki Toshito, Takashi Akagi

MEDICAL PHYSICS 47 巻 ( 9 ) 頁： 3882 - 3891 2020年9月

大腸内視鏡における穿孔の自動検出および位置推定に関する予備的検討

蒋凱, 伊東隼人, 小田昌宏, 奥村大志, 森悠一, 三澤将史, 林武雅, 工藤進英, 森健策

日本医用画像工学会大会予稿集 39回巻頁： 29 - 29 2020年9月

位置特徴量とマルチスケール特徴量による胃壁マイクロCT像からの胃壁の層構造及び腫瘍の抽出

御手洗翠, 小田紘久, 杉野貴明, 守谷享泰, 伊東隼人, 小田昌宏, 小宮山琢真, 古川和宏, 宮原良二, 藤城光弘, 森雅樹, 高畠博嗣, 名取博, 森健策

日本医用画像工学会大会予稿集 39回巻頁： 569 - 572 2020年9月

記述言語：日本語出版者・発行元：日本医用画像工学会

本稿では,Spherical K-means(SpK)により抽出された複数スケールの特徴量と位置特徴量を用いた胃壁μCT画像からの粘膜層,粘膜下層,筋層及び腫瘍の抽出手法について報告する.胃壁のμCT画像から解剖学的構造を抽出することで腫瘍の立体構造把握が可能である.従来手法では複数スケールの特徴抽出フィルタをSpKによって学習し,パッチから特徴抽出を行った.その後得られた特徴量に基づいてSVMによって分類を行っていた.しかし,SpKにより生成された特徴抽出フィルタは濃度値に基づく特徴量を抽出するため,粘膜層と筋層のように濃度値が類似した領域同士の特徴量が近しい値になる.本手法ではSpKにより得られる複数スケールの特徴量にパッチの位置特徴量を追加して分類する事によって,濃度値に基づく特徴量では分類できない領域の分類を可能にする.パッチの位置特徴量としてパッチの中心座標を用いる.本手法を胃壁μCT像に適用した結果,粘膜層,粘膜下層,筋層及び腫瘍の抽出に対するDICE係数は68.3%,63.2%,82.2%,69.1%であった.特に従来手法と比べ,腫瘍の抽出に対するDICE係数は26.1%向上した.(著者抄録)

Development of a computer-aided detection system for colonoscopy and a publicly accessible large colonoscopy video database (with video). 査読有り

A visual SLAM-based bronchoscope tracking scheme for bronchoscopic navigation 査読有り

Cheng Wang, Masahiro Oda, Yuichiro Hayashi, Benjamin Villard, Takayuki Kitasaka, Hirotsugu Takabatake, Masaki Mori, Hirotoshi Honma, Hiroshi Natori, Kensaku Mori

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY 15 巻 ( 10 ) 頁： 1619 - 1630 2020年8月

Misawa M, Kudo SE, Mori Y, Hotta K, Ohtsuka K, Matsuda T, Saito S, Kudo T, Baba T, Ishida F, Itoh H, Oda M, Mori K

Gastrointestinal endoscopy 2020年7月

A deformable model for navigated laparoscopic gastrectomy based on finite elemental method 査読有り

Chen Tao, Wei Guodong, Xu Lili, Shi Weili, Xu Yikai, Zhu Yongyi, Hayashi Yuichiro, Oda Hirohisa, Oda Masahiro, Hu Yanfeng, Yu Jiang, Jiang Zhengang, Li Guoxin, Mori Kensaku

MINIMALLY INVASIVE THERAPY & ALLIED TECHNOLOGIES 29 巻 ( 4 ) 頁： 210 - 216 2020年7月

Artificial Intelligence-assisted System Improves Endoscopic Identification of Colorectal Neoplasms 査読有り

Kudo Shin-ei, Misawa Masashi, Mori Yuichi, Hotta Kinichi, Ohtsuka Kazuo, Ikematsu Hiroaki, Saito Yutaka, Takeda Kenichi, Nakamura Hiroki, Ichimasa Katsuro, Ishigaki Tomoyuki, Toyoshima Naoya, Kudo Toyoki, Hayashi Takemasa, Wakamura Kunihiko, Baba Toshiyuki, Ishida Fumio, Inoue Haruhiro, Itoh Hayato, Oda Masahiro, Mori Kensaku

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 18 巻 ( 8 ) 頁： 1874 - + 2020年7月

PREDICTION OF LYMPH NODE METASTASIS IN T2 COLORECTAL CANCER BASED ON ARTIFICIAL INTELLIGENCE -PROPOSAL OF AN INDICATION FOR FUTURE FULL-THICKNESS ENDOSCOPIC RESECTION- 査読有り

Ichimasa Katsuro, Kudo Shinei, Benjamin Villard, Nakahara Kenta, Mori Yuichi, Misawa Masashi, Yasuharu Maeda, Toyoshima Naoya, Ogata Noriyuki, Kudo Toyoki, Hayashi Takemasa, Wakamura Kunihiko, Miyachi Hideyuki, Sawada Naruhiko, Itoh Hayato, Oda Masahiro, Mori Kensaku, Ishida Fumio

GASTROINTESTINAL ENDOSCOPY 91 巻 ( 6 ) 頁： AB43 - AB43 2020年6月

内視鏡診断支援ソフトウェアEndoBRAINの臨床的有効性および医療費抑制効果

森悠一, 工藤進英, 三澤将史, 武田健一, 前田康晴, 小川悠史, 一政克朗, 若村邦彦, 林武雅, 工藤豊樹, 宮地英行, 馬場俊之, 伊東隼人, 小田昌宏, 森健策

日本大腸検査学会雑誌 36 巻 ( 2 ) 頁： 77 - 82 2020年5月

記述言語：日本語出版者・発行元：(社)日本大腸検査学会

EndoBRAIN(製造:サイバーネットシステム株式会社・販売:オリンパス株式会社)は、人工知能による診断支援システムとして、本邦初の薬機法承認を得た医療機器である。EndoBRAINは超拡大内視鏡(CF-H290ECI、オリンパス株式会社)によって取得される520倍の拡大画像をAIが解析することで、大腸病変が腫瘍なのかどうかを瞬時に予測し、医師に情報提供を行う。2012年から開始したEndoBRAINの研究開発は、昭和大学-名古屋大学-サイバーネットシステム株式会社での医工産連携のもと、日本医療研究開発機構(AMED)からのサポート下に継続的に実施され、2018年12月に薬機法承認を取得した。EndoBRAINの臨床的有用性については、2018年に791症例を対象とした大規模前向き試験の結果が公表されており、感度92.7%で腫瘍の鑑別が可能であると報告されている。また、最近公表された前向き試験の副次解析結果によると、EndoBRAINを使用することで、本来切除すべきでない非腫瘍性ポリープの治療件数を抑制することができるため、これにより最大164億円/年の医療費が削減しうることが分かってきた。本稿では、EndoBRAINの特徴・使用法、および薬機法承認取得までの経緯を紹介するとともに、その臨床的有用性および期待される医療費抑制効果について、最新の研究成果を総括する。(著者抄録)

Super-resolution of clinical CT volumes with modified CycleGAN using micro CT volumes

AIによるSSA/Pの超拡大内視鏡診断

小川悠史, 工藤進英, 森悠一, 三澤将史, 片岡伸一, 前田康晴, 一政克朗, 石垣智之, 工藤豊樹, 若村邦彦, 林武雅, 馬場俊之, 石田文生, 伊東隼人, 小田昌宏, 森健策

日本大腸検査学会雑誌 36 巻 ( 2 ) 頁： 125 - 125 2020年5月

Tong ZHENG, Hirohisa ODA, Takayasu MORIYA, Takaaki SUGINO, Shota NAKAMURA, Masahiro ODA, Masaki MORI, Hirotsugu TAKABATAKE, Hiroshi NATORI, Kensaku MORI

CoRR abs/2004.03272 巻頁： 454 - 459 2020年4月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：日本医用画像工学会

This paper presents a super-resolution (SR) method with unpaired training
dataset of clinical CT and micro CT volumes. For obtaining very detailed
information such as cancer invasion from pre-operative clinical CT volumes of
lung cancer patients, SR of clinical CT volumes to $\m$}CT level is desired.
While most SR methods require paired low- and high- resolution images for
training, it is infeasible to obtain paired clinical CT and {\mu}CT volumes. We
propose a SR approach based on CycleGAN, which could perform SR on clinical CT
into $\mu$CT level. We proposed new loss functions to keep cycle consistency,
while training without paired volumes. Experimental results demonstrated that
our proposed method successfully performed SR of clinical CT volume of lung
cancer patients into $\mu$CT level.

arXiv

Artificial intelligence for magnifying endoscopy, endocytoscopy, and confocal laser endomicroscopy of the colorectum

その他リンク： http://arxiv.org/pdf/2004.03272v1

How Far Will Clinical Application of AI Applications Advance for Colorectal Cancer Diagnosis? 査読有り

Mori Yuichi, Kudo Shin-ei, Misawa Masashi, Takeda Kenichi, Kudo Toyoki, Itoh Hayato, Oda Masahiro, Mori Kensaku

JOURNAL OF THE ANUS RECTUM AND COLON 4 巻 ( 2 ) 頁： 47 - 50 2020年4月

Mori Yuichi, Kudo Shin-ei, Misawa Masashi, Itoh Hayato, Oda Masahiro, Mori Kensaku

TECHNIQUES AND INNOVATIONS IN GASTROINTESTINAL ENDOSCOPY 22 巻 ( 2 ) 頁： 56 - 60 2020年4月

記述言語：日本語出版者・発行元：Techniques and Innovations in Gastrointestinal Endoscopy

Because magnifying endoscopy is considered to be more accurate at predicting the histology of colorectal polyps than nonmagnifying endoscopy, it has been attracting a lot of attention, especially in Japan. However, use of magnifying endoscopy is not yet widespread because of its limited availability and the difficulty in interpreting the acquired images. Application of artificial intelligence (AI) is now changing this situation because it helps less-skilled endoscopists to accurately interpret magnified images. Research in this field initially focused on magnifying endoscopy with narrow-band imaging as the target of AI. Most previously published retrospective studies have reported over 90% sensitivity in differentiation of neoplastic lesions; however, automatically indicating the region of interest (ROI) of the polyps that AI should analyze has been found to be challenging. To address this practical problem, some researchers have started to adopt contact endomicroscopy as a target for AI. Contact endomicroscopy includes endocytoscopy (520-fold magnification, Olympus, Tokyo, Japan) and confocal laser endomicroscopy (1000-fold magnification, Mauna Kea, Paris, France). These forms of contact endomicroscopy provide ultramagnified images that make it unnecessary to manually select the ROI because the entire image acquired by contact endomicroscopy is the ROI of the targeted polyps. This strength of contact endomicroscopy has contributed to early implementation of this technology into clinical practice, which may change the utility of magnifying endoscopy in clinical settings and help increase its use globally in the near future.

DOI： 10.1016/j.tgie.2019.150632

Tensor-cut: A tensor-based graph-cut blood vessel segmentation method and its application to renal artery segmentation 査読有り

Cardiac fiber tracking on super high-resolution CT images: a comparative study 査読有り

Hirohisa Oda, Holger R. Roth, Takaaki Sugino, Naoki Sunaguchi, Noriko Usami, Masahiro Oda, Daisuke Shimao, Shu Ichihara, Tetsuya Yuasa, Masami Ando, Toshiaki Akita, Yuji Narita, Kensaku Mori

JOURNAL OF MEDICAL IMAGING 7 巻 ( 2 ) 頁： 026001 2020年3月

粒子線治療の線量測定のためU-Netを使用した水の発光画像からの線量分布推定(Estimation of Dose Distributions from the Luminescence Image of Water Using U-Net for Particle Therapy Dosimetry)

Yabe Takuya, Oda Masahiro, Mori Kensaku, Akagi Takashi, Yamamoto Seiichi

日本放射線技術学会総会学術大会予稿集 76回巻頁： 336 - 337 2020年3月

Chenglong Wang, Masahiro Oda, Yuichiro Hayashi, Yasushi Yoshino, Tokunori Yamamoto, Alejandro F. Frangi, Kensaku Mori

MEDICAL IMAGE ANALYSIS 60 巻頁： 101623 2020年2月

Micro-computed tomography images of lung adenocarcinoma: detection of lepidic growth patterns 査読有り

Nakamura Shota, Mori Kensaku, Iwano Shingo, Kawaguchi Koji, Fukui Takayuki, Hakiri Shuhei, Ozeki Naoki, Oda Masahiro, Yokoi Kohei

NAGOYA JOURNAL OF MEDICAL SCIENCE 82 巻 ( 1 ) 頁： 25 - 31 2020年2月

人工知能に基づく医療機器EndoBRAINの臨床導入 ―薬事承認取得・保険算定への挑戦― 招待有り

森悠一, 工藤進英, 三澤将史, 伊東隼人, 小田昌宏, 森健策

Ｍｅｄｉｃａｌ　Ｉｍａｇｉｎｇ　Ｔｅｃｈｎｏｌｏｇｙ 38 巻 ( 5 ) 頁： 213-216 2020年

記述言語：英語掲載種別：研究論文（学術雑誌）

EndoBRAIN（製造：サイバネットシステム社，販売：オリンパス社）は，内視鏡検査中に発見された大腸ポリープが腫瘍なのかどうかを瞬時に解析予測し，医師の診断支援を行うシステムである．EndoBRAINは人工知能に基づく診断支援システムとして，本邦初の薬事承認を2018年12月に得た医療機器であるが，承認達成は緊密な医工産官連携研究の賜物であった．昭和大学（医）-名古屋大学（工）-サイバネットシステム株式会社（産）での連携体制を構築し，複数の公的研究費（文部科学省科学研究費（新学術領域研究・多元計算解剖学を含む）・日本医療研究開発機構（AMED）研究費）のサポートを得ることで，着実に研究成果を蓄積し，薬事承認に至った．本稿では，EndoBRAINの研究開発概要を示すとともに，医療機器開発における鬼門である薬事承認と，その後の普及において不可欠とされる（しかしながらいまだAI医療機器については実現していない）保険承認に対するアプローチについて紹介したい．

DOI： 10.11409/mit.38.213

コンピュータ外科における画像ベース深層学習の利用招待有り

小田昌宏

日本コンピュータ外科学会誌 22 巻 ( 1 ) 頁： 54-58 2020年

Organ Segmentation From Full-size CT Images Using Memory-Efficient FCN 査読有り

Wang Chenglong, Oda Masahiro, Mori Kensaku

MEDICAL IMAGING 2020: COMPUTER-AIDED DIAGNOSIS 11314 巻 2020年

Usefulness of fine-tuning for deep learning based multi-organ regions segmentation method from non-contrast CT volumes using small training dataset 査読有り

Hayashi Yuichiro, Shen Chen, Roth Holger R., Oda Masahiro, Misawa Kazunari, Jinzaki Masahiro, Hashimoto Masahiro, Kumamaru Kanako K., Aoki Shigeki, Mori Kensaku

MEDICAL IMAGING 2020: COMPUTER-AIDED DIAGNOSIS 11314 巻 2020年

Visualising decision-reasoning regions in computer-aided pathological pattern diagnosis of endoscytoscopic images based on CNN weights analysis 査読有り

Itoh Hayato, Lu Zhongyang, Mori Yuichi, Misawa Masashi, Oda Masahiro, Kudo Shin-ei, Mori Kensaku

MEDICAL IMAGING 2020: COMPUTER-AIDED DIAGNOSIS 11314 巻 2020年

Visualizing intestines for diagnostic assistance of ileus based on intestinal region segmentation from 3D CT images 査読有り

Oda Hirohisa, Nishio Kohei, Kitasaka Takayuki, Amano Hizuru, Takimoto Aitaro, Uchida Hiroo, Suzuki Kojiro, Itoh Hayato, Oda Masahiro, Mori Kensaku

MEDICAL IMAGING 2020: COMPUTER-AIDED DIAGNOSIS 11314 巻 2020年

Spatial information-embedded fully convolutional networks for multi-organ segmentation with improved data augmentation and instance normalization 査読有り

Chen Shen, Chenglong Wang, Holger R. Roth, Masahiro Oda, Yuichiro Hayashi, Kazunari Misawa, Kensaku Mori

Progress in Biomedical Optics and Imaging - Proceedings of SPIE 11313 巻 2020年

Multi-modality super-resolution loss for GAN-based super-resolution of clinical CT images using micro CT image database 査読有り

Tong Zheng, Hirohisa Oda, Takayasu Moriya, Takaaki Sugino, Shota Nakamura, Masahiro Oda, Masaki Mori, Hirotsugu Takabatake, Hiroshi Natori, Kensaku Mori

Progress in Biomedical Optics and Imaging - Proceedings of SPIE 11313 巻 2020年

Improved visual SLAM for bronchoscope tracking and registration with pre-operative CT images 査読有り

Cheng Wang, Masahiro Oda, Yuichiro Hayashi, Takayuki Kitasaka, Hirotoshi Honma, Hirotsugu Takabatake, Masaki Mori, Hiroshi Natori, Kensaku Mori

Proceedings of SPIE - The International Society for Optical Engineering 11315 巻 2020年

Automated eye disease classification method from anterior eye image using anatomical structure focused image classification technique 査読有り

Oda Masahiro, Yamaguchi Takefumi, Fukuoka Hideki, Ueno Yuta, Mori Kensaku

MEDICAL IMAGING 2020: COMPUTER-AIDED DIAGNOSIS 11314 巻 2020年

Abdominal artery segmentation method from CT volumes using fully convolutional neural network 査読有り

Masahiro Oda, Holger R. Roth, Takayuki Kitasaka, Kazunari Misawa, Michitaka Fujiwara, Kensaku Mori

International Journal of Computer Assisted Radiology and Surgery 14 巻 ( 12 ) 頁： 2069 - 2081 2019年12月

Realistic Endoscopic Image Generation Method Using Virtual-to-real Image-domain Translation 査読有り

Masahiro Oda, Kiyohito Tanaka, Hirotsugu Takabatake, Masaki Mori, Hiroshi Natori, Kensaku Mori

Healthcare Technology Letters 6 巻 ( 6 ) 頁： 214 - 219 2019年12月

Stable polyp-scene classification via subsampling and residual learning from an imbalanced large dataset 査読有り

Hayato Itoh, Holger Roth, Masahiro Oda, Masashi Misawa, Yuichi Mori, Shin-Ei Kudo, Kensaku Mori

HEALTHCARE TECHNOLOGY LETTERS 6 巻 ( 6 ) 頁： 237 - 242 2019年12月

Artificial intelligence and computer-aided diagnosis for colonoscopy: where do we stand now?

Kudo Shin-ei, Mori Yuichi, Abdel-aal Usama M., Misawa Masashi, Itoh Hayato, Oda Masahiro, Mori Kensaku

TRANSLATIONAL GASTROENTEROLOGY AND HEPATOLOGY 2019年11月

超拡大大腸内視鏡画像における施設間データ分布の差異を考慮した分類法に関する初期的検討

伊東隼人, 森悠一, 三澤将史, 小田昌宏, 工藤進英, 森健策

日本コンピュータ外科学会誌 21 巻 ( 4 ) 頁： 332 - 333 2019年11月

臨床肺CT画像と切除肺マイクロCT画像の初期位置合わせ手法の検討査読有り

波多腰慎矢, 小田紘久, 杉野貴明, 林雄一郎, Roth Holger R., 中村彰太, 小田昌宏, 森雅樹, 高畠博嗣, 名取博, 森健策

日本コンピュータ外科学会誌 21 巻 ( 4 ) 頁： 318 - 319 2019年11月

内視鏡的粘膜下層剥離術中の自動穿孔検出に関する初期的検討

大石仁美, 伊東隼人, 森悠一, 三澤将史, 林武雅, 奥村大志, 小田昌宏, 工藤進英, 森健策

日本コンピュータ外科学会誌 21 巻 ( 4 ) 頁： 232 - 233 2019年11月

Grad-CAMを用いた脳CT像からのくも膜下出血の出血領域可視化に関する検討(Visualization of subarachnoid hemorrhage area from brain CT images using Grad-CAM)

魯仲陽, 伊東隼人, 小田昌宏, 林雄一郎, 渡谷岳行, 阿部修, 橋本正弘, 陣崎雅弘, 森健策

日本コンピュータ外科学会誌 21 巻 ( 4 ) 頁： 229 - 230 2019年11月

小児腸閉塞患者のCT像における電子洗浄手法の評価

西尾光平, 小田紘久, 千馬耕亮, 北坂孝幸, 林雄一郎, 伊東隼人, 小田昌宏, 檜顕成, 内田広夫, 森健策

日本コンピュータ外科学会誌 21 巻 ( 4 ) 頁： 321 - 321 2019年11月

グラフ畳み込みニューラルネットワークによる腹部動脈血管名自動命名におけるデータ拡張による精度改善

日比裕太, 林雄一郎, 北坂孝幸, 伊東隼人, 小田昌宏, 三澤一成, 森健策

日本コンピュータ外科学会誌 21 巻 ( 4 ) 頁： 226 - 227 2019年11月

特集今知りたい、AIの歴史とこれからディープラーニング実践の環境構築

小田昌宏

臨床画像 35 巻 ( 10 ) 頁： 1120 - 1128 2019年10月

Precise estimation of renal vascular dominant regions using spatially aware fully convolutional networks, tensor-cut and Voronoi diagrams 査読有り

Chenglong Wang, Holger R. Roth, Takayuki Kitasaka, Masahiro Oda, Yuichiro Hayashi, Yasushi Yoshino, Tokunori Yamamoto, Naoto Sassa, Momokazu Goto, Kensaku Mori

COMPUTERIZED MEDICAL IMAGING AND GRAPHICS 77 巻頁： 101642 2019年10月

SSA/Pの人工知能支援下超拡大内視鏡診断

小川悠史, 工藤進英, 森悠一, 三澤将史, 武田健一, 片岡伸一, 前田康晴, 一政克朗, 工藤豊樹, 若村邦彦, 林武雅, 馬場俊之, 石田文生, 伊東隼人, 小田昌宏, 森健策

Gastroenterological Endoscopy 61 巻 ( Suppl.2 ) 頁： 2167 - 2167 2019年10月

A view of three dimensional unit structures of alveoli in peripheral lung 査読有り

Natori Hiroshi, Takabatake Hirotsugu, Mori Masaki, Oda Masahiro, Mori Kensaku, Koba Hiroyuki, Takahashi Hiroki

EUROPEAN RESPIRATORY JOURNAL 54 巻 2019年9月

Artificial intelligence and colonoscopy: Current status and future perspectives 査読有り

Kudo Shin-ei, Mori Yuichi, Misawa Masashi, Takeda Kenichi, Kudo Toyoki, Itoh Hayato, Oda Masahiro, Mori Kensaku

DIGESTIVE ENDOSCOPY 31 巻 ( 4 ) 頁： 363 - 371 2019年7月

Artificial intelligence and upper gastrointestinal endoscopy: Current status and future perspective 査読有り

Yuichi Mori, Shin‐ei Kudo, Hussein Ebaid Naeem Mohmed, Masashi Misawa, Noriyuki Ogata, Hayato Itoh, Masahiro Oda, Kensaku Mori

DIGESTIVE ENDOSCOPY 31 巻 ( 4 ) 頁： 378 - 388 2019年7月

3D fully convolutional network を用いた腎腫瘍の定量評価における初期検討

王成龍, 小田昌宏, 林雄一郎, 佐々直人, 山本徳則, 森健策

第38回日本医用画像工学会大会予稿集頁： OP5-23 2019年7月

AMED 大規模データベースを用いた CT 画像解析と病変検出への応用

森健策, 小田昌宏

第38回日本医用画像工学会大会予稿集頁： SY1-5 2019年7月

Polyp size classification in colorectal cancer using a Siamese network

ヴィラードベンジャミン, 伊東隼人, 小田昌宏, 森悠一, 一政克朗, 三澤将史, 工藤進英, 森健策

第38回日本医用画像工学会大会予稿集頁： OP2-14 2019年7月

少量のラベルデータを用いた学習によるイレウス症例 CT 像における拡張腸管の自動抽出

小田紘久, 西尾光平, 北坂孝幸, 天野日出, 千馬耕亮, 内田広夫, 鈴木耕次郞, 伊東隼人, 小田昌宏, 森健策

第38回日本医用画像工学会大会予稿集頁： OP1-15 2019年7月

小児腸閉塞患者の CT 像における CycleGAN を用いた電子洗浄手法の検討

西尾光平, 小田紘久, 千馬耕亮, 北坂孝幸, 伊東隼人, 小田昌宏, 檜顕成, 内田広夫, 森健策

第38回日本医用画像工学会大会予稿集頁： OP3-20 2019年7月

グラフ畳み込みニューラルネットワークを用いた腹部動脈血管名自動命名の初期検討

日比裕太, 林雄一郎, 北坂孝幸, 伊東隼人, 小田昌宏, 三澤一成, 森健策

第38回日本医用画像工学会大会予稿集頁： OP5-11 2019年7月

深層学習における学習データセット規模拡大に応じた分類精度向上に関する実験的検討～超拡大大腸内視鏡画像における腫瘍性病変分類に向けた特徴量抽出～

伊東隼人, 森悠一, 三澤将史, 小田昌宏, 工藤進英, 森健策

第38回日本医用画像工学会大会予稿集頁： OP1-24 2019年7月

表現学習と SVM による胃壁マイクロ CT 像の半教師ありセグメンテーション手法

御手洗翠, 小田紘久, 杉野貴明, 守谷享泰, 伊東隼人, 小田昌宏, 小宮山琢真, 森雅樹, 高畠博嗣, 名取博, 森健策

第38回日本医用画像工学会大会予稿集頁： OP2-08 2019年7月

腹腔鏡動画像からの Fully Convolutional Network による血管領域抽出

盛満慎太郎, 小澤卓也, 北坂孝幸, 林雄一郎, 小田昌宏, 伊藤雅昭, 竹下修由, 三澤一成, 森健策

第38回日本医用画像工学会大会予稿集頁： OP3-12 2019年7月

移学習を用いた腹部 thick-slice CT 像における多臓器領域の自動抽出の初期検討

申忱, ロスホルガー, 林雄一郎, 小田紘久, 小田昌宏, 三澤一成, 森健策

第38回日本医用画像工学会大会予稿集頁： OP4-15 2019年7月

深層学習を用いた非造影 CT 画像からの複数臓器領域の抽出に関する検討

林雄一郎, 申忱, Roth Holger, 小田昌宏, 三澤一成, 森健策

第38回日本医用画像工学会大会予稿集頁： OP5-14 2019年7月

深層学習を用いた腹腔鏡手術動画像の出血領域自動セグメンテーション

山本翔太, 小田紘久, 林雄一郎, 北坂孝幸, 小田昌宏, 伊藤雅昭, 竹下修由, 森健策

第38回日本医用画像工学会大会予稿頁： OP4-13 2019年7月

ARTIFICIAL INTELLIGENCE-ASSISTED POLYP DETECTION SYSTEM FOR COLONOSCOPY, BASED ON THE LARGEST AVAILABLE COLLECTION OF CLINICAL VIDEO DATA FOR MACHINE LEARNING 査読有り

Misawa Masashi, Kudo Shinei, Mori Yuichi, Cho Tomonari, Kataoka Shinichi, Maeda Yasuharu, Ogawa Yushi, Takeda Kenichi, Nakamura Hiroki, Ichimasa Katsuro, Toyoshima Naoya, Ogata Noriyuki, Kudo Toyoki, Hisayuki Tomokazu, Hayashi Takemasa, Wakamura Kunihiko, Baba Toshiyuki, Ishida Fumio, Itoh Hayato, Oda Masahiro, Mori Kensaku

GASTROINTESTINAL ENDOSCOPY 89 巻 ( 6 ) 頁： AB646 - AB647 2019年6月

PERFORMANCE OF NON-EXPERT ENDOSCOPISTS IN OPTICAL BIOPSY OF DIMINUTIVE COLORECTAL POLYPS WITH REAL-TIME USE OF ARTIFICIAL INTELLIGENCE 査読有り

Mori Yuichi, Kudo Shinei, Misawa Masashi, Kataoka Shinichi, Takeda Kenichi, Suzuki Kenichi, Ichimasa Katsuro, Ogawa Yushi, Maeda Yasuharu, Hayashi Takemasa, Wakamura Kunihiko, Kudo Toyoki, Ishida Fumio, Inoue Haruhiro, Itoh Hayato, Oda Masahiro, Mori Kensaku

GASTROINTESTINAL ENDOSCOPY 89 巻 ( 6 ) 頁： AB89 - AB89 2019年6月

COMPUTER-AIDED DIAGNOSIS FOR SAMLL COLORECTAL LESIONS: A MULTI-CENTER VALIDATION "ENDOBRAIN STUDY" DESIGNED TO OBTAIN REGULATORY APPROVAL 査読有り

Hotta Kinichi, Kudo Shinei, Mori Yuichi, Ikematsu Hiroaki, Saito Yutaka, Ohtsuka Kazuo, Misawa Masashi, Itoh Hayato, Oda Masahiro, Mori Kensaku

GASTROINTESTINAL ENDOSCOPY 89 巻 ( 6 ) 頁： AB76 - AB76 2019年6月

ARTIFICIAL INTELLIGENCE-ASSISTED POLYP DETECTION SYSTEM FOR COLONOSCOPY, BASED ON THE LARGEST AVAILABLE COLLECTION OF CLINICAL VIDEO DATA FOR MACHINE LEARNING 査読有り

GASTROINTESTINAL ENDOSCOPY 89 巻 ( 6 ) 頁： AB646-AB647 2019年6月

Evaluation on econstruction accuracy of visual SLAM based bronchoscope tracking

C. Wang, Masahiro Oda, Yuichiro Hayashi, Takayuki Kitasaka, Hayato Itoh, H. Honma, H. Takabatake, M. Mori, H. Natori, Kensaku Mori

International Journal of Computer Assisted Radiology and Surgery CARS 2019 14 巻 ( 1 ) 頁： S24-25 2019年6月

深層学習を用いた脳CT像からの出血検出におけるデータ拡張とネットワーク構造の影響に関する考察

魯仲陽, 小田昌宏, 鄭通, 申忱, 胡涛, 渡谷岳行, 阿部修, 橋本正弘, 陣崎雅弘, 森健策

電子情報通信学会技術研究報告(MI) 119 巻 ( 51 ) 頁： 65-70 2019年5月

Development of a New Laparoscopic Detection System for Gastric Cancer Using Near-Infrared Light-Emitting Clips with Glass Phosphor 査読有り

Inada Shunko A, Nakanishi Hayao, Oda Masahiro, Mori Kensaku, Ito Akihiro, Hasegawa Junichi, Misawa Kazunari, Fuchi Shingo

MICROMACHINES 10 巻 ( 2 ) 2019年2月

Fully automated diagnostic system with artificial intelligence using endocytoscopy to identify the presence of histologic inflammation associated with ulcerative colitis 査読有り

Yasuharu Maeda, Shin-eiKudo, Yuichi Mori, Masashi Misawa, Noriyuki Ogata, Seiko Sasanuma, Kunihiko Wakamura, Masahiro Oda, Kensaku Mori, Kazuo Ohtsuka

GASTROINTESTINAL ENDOSCOPY 89 巻 ( 2 ) 頁： 408 - 415 2019年2月

Fully automated diagnostic system with artificial intelligence using endocytoscopy to identify the presence of histologic inflammation associated with ulcerative colitis (with video) 査読有り

Yasuharu Maeda, Shin-eiKudo, Yuichi Mori, Masashi Misawa, Noriyuki Ogata, Seiko Sasanuma, Kunihiko Wakamura, Masahiro Oda, Kensaku Mori, Kazuo Ohtsuka

Gastrointestinal endoscopy 89 巻 ( 2 ) 頁： 408-415 2019年2月

Multi-class abdominal organs segmentation with improved V-Nets

Chen Shen, Fausto Milletari, Holger R. Roth, Hirohisa Oda, Masahiro Oda, Yuichiro Hayashi, Kazunari Misawa, Kensaku Mori

Proceedings of SPIE 10949, Medical Imaging 2019 頁： 109490B-1-7 2019年2月

CTからの腹部多臓器抽出におけるgroup normalizationの影響に関する考察

申忱, Fausto Milletari, Holger R. Roth, 小田紘久, 小田昌宏, 林雄一郎, 三澤一成, 森健策

電子情報通信学会技術研究報告(MI) 118 巻 ( 412 ) 頁： 143-148 2019年1月

機械学習を用いた腹部動脈血管名自動命名におけるデータ拡張法の適用に関する検討

鉄村悠介, 林雄一郎, 小田昌宏, 北坂孝幸, 三澤一成, 森健策

電子情報通信学会技術研究報告(MI) 118 巻 ( 412 ) 頁： 191-196 2019年1月

敵対的Dense U-netを用いた切除肺マイクロCT像の超解像

鄭通, 小田紘久, Holger R. Roth, 小田昌宏, 中村彰太, 森健策

電子情報通信学会技術研究報告(MI) 118 巻 ( 412 ) 頁： 7-12 2019年1月

切除肺のマイクロCT像における3D-DBPNを用いた超解像の検討

鄭通, 小田紘久, 小田昌宏, 守谷享泰, 中村彰太, 森健策

第11回呼吸機能イメージング研究会学術集会, プログラム抄録集頁： 82 2019年1月

不均衡データからの特徴選択超拡大内視鏡画像の病理類型分類に向けて

伊東隼人, 森悠一, 三澤将史, 小田昌宏, 工藤進英, 森健策

電子情報通信学会技術研究報告(MI) 118 巻 ( 412 ) 頁： 109-114 2019年1月

マルチモーダル画像を用いた深層学習ベースの頭部解剖構造抽出少量画像データ学習における抽出精度検証

杉野貴明, Holger R. Roth, 小田昌宏, 金太一, 森健策

電子情報通信学会技術研究報告(MI) 118 巻 ( 412 ) 頁： 65-70 2019年1月

Investigation of extracting the interlobular septa with combination of Hessian analysis and radial structure tensor in micro-CT volume

Xiaotian Zhao, Hirohisa Oda, Shota Nakamura, Yuichiro Hayashi, Hayato Itoh, Masahiro Oda, Kensaku Mori

IFMIA 2019 頁： 0 2019年1月

経時CT像間の腹部臓器の変形を考慮したリンパ節自動対応付け手法の検討

舘高基, 小田昌宏, 林雄一郎, 伊東隼人, 中村嘉彦, 北坂孝幸, 三澤一成, 森健策

電子情報通信学会技術研究報告(MI) 118 巻 ( 412 ) 頁： 97-102 2019年1月

Automated Hand Eye Calibration in Laparoscope Holding Robot for Robot Assisted Surgery 査読有り

Jiang Shuai, Hayashi Yuichiro, Wang Cheng, Oda Masahiro, Kitasaka Takayuki, Misawa Kazunari, Mori Kensaku

INTERNATIONAL WORKSHOP ON ADVANCED IMAGE TECHNOLOGY (IWAIT) 2019 11049 巻 2019年

特集／医用画像処理におけるGenerative Adversarial Networks の利用 − 序文 − 招待有り

小田昌宏

ＭｅｄｉｃａｌＩｍａｇｉｎｇＴｅｃｈｎｏｌｏｇｙ 37 巻 ( 3 ) 頁： 123-124 2019年

Weakly-supervised Deep Learning of Interstitial Lung Disease Types on CT Images

Wang Chenglong, Moriya Takayasu, Hayashi Yuichiro, Roth Holger, Lu Le, Oda Masahiro, Ohkubo Hirotugu, Mori Kensaku

MEDICAL IMAGING 2019: COMPUTER-AIDED DIAGNOSIS 10950 巻 2019年

Discriminative Feature Selection by Optimal Manifold Search for Neoplastic Image Recognition 査読有り

Itoh Hayato, Mori Yuichi, Misawa Masashi, Oda Masahiro, Kudo Shin-Ei, Mori Kensaku

COMPUTER VISION - ECCV 2018 WORKSHOPS, PT IV 11132 巻頁： 534-549 2019年

外科支援におけるAI・深層学習招待有り

小田昌宏

日本コンピュータ外科学会誌 21 巻 ( 3 ) 頁： 143-146 2019年

名古屋大学の取り組み：放射線画像診断支援と内視鏡画像解析招待有り

小田昌宏, 申忱, 小田紘久, 森健策

Ｍｅｄｉｃａｌ　Ｉｍａｇｉｎｇ　Ｔｅｃｈｎｏｌｏｇｙ 37 巻 ( 2 ) 頁： 84-88 2019年

担当区分：筆頭著者記述言語：英語掲載種別：研究論文（研究会，シンポジウム資料等）

本稿では，われわれがAMED プロジェクトにおいて行ってきた医用画像解析研究について紹介する．具体的には，非造影腹部CT 像からの血管診断支援およびCT 像からの腹部複数臓器自動セグメンテーションを含む放射線画像診断支援研究，胃および大腸の内視鏡画像の観察部位分類を含む内視鏡画像解析研究を行った．これらの研究を概説し，得られた成果等について述べる．

DOI： 10.11409/mit.37.84

Micro-CT in the Analysis of Formalin-Fixed Paraffin-Embedded Blocks of Resected Pancreatic Lesions 査読有り

Shindo K., Ohuchida K., Roth H. R., Oda H., Iwamoto C., Oda M., Ohtsuka T., Mori K., Hashizume M., Nakamura M.

PANCREAS 48 巻 ( 10 ) 頁： 1523 - 1523 2019年

Polyp-Size Classification with RGB-D features for Colonoscopy 査読有り

Itoh Hayato, Roth Holger R., Mori Yuichi, Misawa Masashi, Oda Masahiro, Kudo Shin-ei, Mori Kensaku

MEDICAL IMAGING 2019: COMPUTER-AIDED DIAGNOSIS 10950 巻 2019年

Scanning, registration, and fiber estimation of rabbit hearts using micro-focus and refraction-contrast X-ray CT 査読有り

Oda Hirohisa, Roth Holger R., Sugino Takaaki, Sunaguchi Naoki, Usami Noriko, Oda Masahiro, Shimao Daisuke, Ichihara Shu, Yuasa Tetsuya, Ando Masami, Akita Toshiaki, Narita Yuji, Mori Kensaku

MEDICAL IMAGING 2019: BIOMEDICAL APPLICATIONS IN MOLECULAR, STRUCTURAL, AND FUNCTIONAL IMAGING 10953 巻 2019年

Spaciousness Filters for Non-contrast CT Volume Segmentation of the Intestine Region for Emergency Ileus Diagnosis 査読有り

Oda Hirohisa, Nishio Kohei, Kitasaka Takayuki, Villard Benjamin, Amano Hizuru, Chiba Kosuke, Hinoki Akinari, Uchida Hiroo, Suzuki Kojiro, Itoh Hayato, Oda Masahiro, Mori Kensaku

MICCAI 2019, Workshop: CLIP2019 11840 巻頁： 104-114 2019年

Tubular Structure Segmentation Using Spatial Fully Connected Network with Radial Distance Loss for 3D Medical Images 査読有り

Wang Chenglong, Hayashi Yuichiro, Oda Masahiro, Itoh Hayato, Kitasaka Takayuki, Frangi Alejandro F., Mori Kensaku

MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION - MICCAI 2019, PT VI 11769 巻頁： 348-356 2019年

Unsupervised Segmentation of Micro-CT Images Based on a Hybrid of Variational Inference and Adversarial Learning 査読有り

Moriya Takayasu, Roth Holger R., Nakamura Shota, Oda Hirohisa, Oda Masahiro, Mori Kensaku

MEDICAL IMAGING 2019: BIOMEDICAL APPLICATIONS IN MOLECULAR, STRUCTURAL, AND FUNCTIONAL IMAGING 10953 巻 2019年

Unsupervised Segmentation of Micro-CT Images of Lung Cancer Specimen Using Deep Generative Models 査読有り

Moriya Takayasu, Oda Hirohisa, Mitarai Midori, Nakamura Shota, Roth Holger R., Oda Masahiro, Mori Kensaku

MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION - MICCAI 2019, PT VI 11769 巻頁： 240-248 2019年

Visual SLAM for bronchoscope tracking and bronchus reconstruction in bronchoscopic navigation 査読有り

Wang Cheng, Oda Masahiro, Hayashi Yuichiro, Kitasaka Takayuki, Honma Hirotoshi, Takabatake Hirotsugu, Mori Masaki, Natori Hiroshi, Mori Kensaku

MEDICAL IMAGING 2019: IMAGE-GUIDED PROCEDURES, ROBOTIC INTERVENTIONS, AND MODELING 10951 巻 2019年

Improving V-Nets for multi-class abdominal organ segmentation 査読有り

Shen Chen, Milletari Fausto, Roth Holger R., Oda Hirohisa, Oda Masahiro, Hayashi Yuichiro, Misawa Kazunari, Mori Kensaku

MEDICAL IMAGING 2019: IMAGE PROCESSING 10949 巻 2019年

Investigation of extracting interlobular septa with Hessian analysis and Radial Structure Tensor combined with roundness error in micro-CT volume 査読有り

Zhao Xiaotian, Oda Hirohisa, Nakamura Shota, Hayashi Yuichiro, Itoh Hayato, Oda Masahiro, Mori Kensaku

INTERNATIONAL FORUM ON MEDICAL IMAGING IN ASIA 2019 11050 巻 2019年

ディープラーニング実践の環境構築

小田昌宏

臨床画像 35 巻頁： 1120 - 1128 2019年

大腸内視鏡治療誘導のためのRecurrent Neural Networkを用いた大腸内視鏡トラッキング手法の開発査読有り

小田昌宏, Holger R. Roth, 北坂孝幸, 古川和宏, 宮原良二, 廣岡芳樹, Nassir Navab, 森健策

日本バーチャルリアリティ学会論文誌 23 巻 ( 4 ) 頁： 249-252 2018年12月

CT 像より自動抽出された動脈領域に対応した機械学習に基づく腹部動脈血管名自動命名法

鉄村悠介, 林雄一郎, 小田昌宏, 北坂孝幸, 三澤一成, 森健策

日本コンピュータ外科学会誌第27回日本コンピュータ外科学会大会特集号 20 巻 ( 4 18(Ⅶ)-4 ) 頁： 322-323 2018年11月

不均衡データセットからの学習データセット構築法 ―機械学習に基づく医用画像分類に向けて―

伊東隼人, 森悠一, 三澤将史, 小田昌宏, 工藤進英, 森健策

日本コンピュータ外科学会誌第27回日本コンピュータ外科学会大会特集号 20 巻 ( 4 18(Ⅲ)-5 ) 頁： 261-262 2018年11月

ディープラーニングを用いた腹腔鏡映像からの腹腔鏡下胃切除術の手術工程解析の検討

林雄一郎, 杉野貴明, 小田昌宏, 三澤一成, 森健策

日本コンピュータ外科学会誌第27回日本コンピュータ外科学会大会特集号 20 巻 ( 4 18(ⅩⅠ)-10 ) 頁： 368-369 2018年11月

イレウス診断支援システムにおける閉塞部位の誤検出修正及び改善ツールの構築

西尾光平, 小田紘久, 千馬耕亮, 北坂孝幸, Holger R. Roth, 伊東隼人, 林雄一郎, 小田昌宏, 檜顕成, 内田広夫, 森健策

日本コンピュータ外科学会誌第27回日本コンピュータ外科学会大会特集号 20 巻 ( 4 18(Ⅹ)-3 ) 頁： 348-349 2018年11月

U‒Net を用いた腹腔鏡動画像における出血領域検出に関する検討

小澤卓也, 小田紘久, 伊東隼人, 北坂孝幸, Holger R. Roth, 小田昌宏, 林雄一郎, 三澤一成, 伊藤雅昭, 竹下修由, 森健策

日本コンピュータ外科学会誌第27回日本コンピュータ外科学会大会特集号 20 巻 ( 4 18(6)-10 ) 頁： 370 2018年11月

SLAM ベースのビジュアルトラッキングにおける隣接フレーム利用再構成手法の評価

王成, 小田昌宏, 林雄一郎, 北坂孝幸, 本間裕敏, 高畠博嗣, 森雅樹, 名取博, 森健策

日本コンピュータ外科学会誌第27回日本コンピュータ外科学会大会特集号 20 巻 ( 4 18(Ⅸ)-4 ) 頁： 342-343 2018年11月

Micro CT and Histopathological Image Registration Based on Deep-Learning Assisted Image Registration

Kensaku Mori, Kai Nagara, Shota Nakamura, Hirohisa Oda, MENG, Holger R. Roth, Masahiro Oda

RSNA2018 頁： CH218-ED-X 2018年11月

Investigation on the condition of using adjacent reconstruction in visual bronchoscope tracking

Cheng Wang, Masahiro Oda, Yuichiro Hayashi, Takayuki Kitasaka, Hirotoshi Honma, Hirotsugu Takabatake, Masaki Mori, Hiroshi Natori, Kensaku Mori

電子情報通信学会技術研究報告(MI), MI2018-43 118 巻 ( 286 ) 頁： 27-32 2018年11月

深層学習を用いたマイクロ CT 画像の超解像に関する初期的検討

鄭通, Holger R. Roth, 小田昌宏, 小田紘久, 中村彰太, 森健策

日本コンピュータ外科学会誌第27回日本コンピュータ外科学会大会特集号 20 巻 ( 4 18(Ⅱ)-8 ) 頁： 252-253 2018年11月

腹腔鏡把持ロボットのための自動ハンドアイキャリブレーションの検討

蒋帥, 林雄一郎, 小田昌宏, 北坂孝幸, 三澤一成, 森健策

日本コンピュータ外科学会誌第27回日本コンピュータ外科学会大会特集号 20 巻 ( 4 18(Ⅹ)-10 ) 頁： 359 2018年11月

生成モデルを利用したマイクロ CT 画像の半教師ありセグメンテーション

守谷享泰, Holger R. Roth, 中村彰太, 小田紘久, 小田昌宏, 森健策

日本コンピュータ外科学会誌第27回日本コンピュータ外科学会大会特集号 20 巻 ( 4 18(Ⅵ)-9 ) 頁： 312-313 2018年11月

深層学習を用いた屈折 X 線 CT 画像からの眼球構造抽出 ―Sparse annnotation データの学習法に関する検討―

杉野貴明, Holger R. Roth, 小田昌宏, 砂口尚輝, 島雄大介, 森健策

日本コンピュータ外科学会誌第27回日本コンピュータ外科学会大会特集号 20 巻 ( 4 18(Ⅲ)-4 ) 頁： 259-260 2018年11月

Real-Time Use of Artificial Intelligence in Identification of Diminutive Polyps During Colonoscopy A Prospective Study 査読有り

Yuichi Mori, Shin-ei Kudo, Masashi Misawa, Yutaka Saito, Hiroaki Ikematsu, Kinichi Hotta, Kazuo Ohtsuka, Fumihiko Urushibara, Shinichi Kataoka, Yushi Ogawa, Yasuharu Maeda, Kenichi Takeda, Hiroki Nakamura, Katsuro Ichimasa, Toyoki Kudo, Takemasa Hayashi, Kunihiko Wakamura, Fumio Ishida, Haruhiro Inoue, Hayato Itoh, Masahiro Oda, Kensaku Mori

ANNALS OF INTERNAL MEDICINE 169 巻 ( 6 ) 頁： 357 - 366 2018年9月

Layout of alveoli and pores of Kohn on magnified 3D printed model of the peripheral lung specimen by micro CT 査読有り

Natori Hiroshi, Takabatake Hirotsugu, Mori Masaki, Mori Kensaku, Oda Masahiro, Koba Hiroyuki, Takahashi Hiroki

EUROPEAN RESPIRATORY JOURNAL 52 巻 2018年9月

Application of three-dimensional print in minor hepatectomy following liver partition between anterior and posterior sectors 査読有り

Tsuyoshi Igami, Yoshihiko Nakamura, Masahiro Oda, Hiroshi Tanaka, Motoi Nojiri, Tomoki Ebata, Yukihiro Yokoyama, Gen Sugawara, Takashi Mizuno, Junpei Yamaguchi, Kensaku Mori, Masato Nagino

ANZ JOURNAL OF SURGERY 88 巻 ( 9 ) 頁： 882 - 885 2018年9月

Sparse annotationによる深層学習ベースの解剖構造抽出：屈折X線CT像からの精密な眼球セグメンテーション

杉野貴明, Holger R. Roth, 小田昌宏, 砂口尚輝, 島雄大介, 市原周, 湯浅哲也, 安藤正海, 森健策

MIRU2018 頁： PS3-11 2018年8月

超拡大内視鏡における病理画像分類のための特徴選択法

伊東隼人, 森悠一, 三澤将史, 小田昌宏, 工藤進英, 森健策

MIRU2018 頁： PS2-17 2018年8月

Fast Marching Algorithmに基づく小児CT像からの腸管閉塞部位検出手法

西尾光平, 小田紘久, 千馬耕亮, 北坂孝幸, Holger Roth, 伊東隼人, 小田昌宏, 檜顕成, 内田広夫, 森健策

第37回日本医用画像工学会大会予稿集頁： OP1-6 2018年7月

機械学習による内視鏡動画インスタンスセグメンテーションのための手動アノテーションツールの開発

小澤卓也, 小田紘久, 伊東隼人, 北坂孝幸, Holger R. Roth, 小田昌宏, 林雄一郎, 三澤一成, 伊藤雅昭, 竹下修由, 森健

第37回日本医用画像工学会大会予稿集頁： OP1-7 2018年7月

教師なし深度推定を利用したRGB-D 特徴抽出に基づくポリープのトリナリサイズ推定

伊東隼人, Holger Roth, 三澤将史, 森悠一, 小田昌宏, 工藤進英, 森健策

第37回日本医用画像工学会大会予稿集頁： OP14-4 2018年7月

マイクロCT画像からのRSTを用いた小葉壁抽出手法の検討

趙笑添, Holger R. Roth, 中村彰太, 小田紘久, 林雄一郎, 守谷享泰, 長柄快, 小田昌宏, 森健策

電子情報通信学会技術研究報告(MI) 118 巻 ( 150 ) 頁： 11-16 2018年7月

ウサギ心臓の屈折CT像における線維配向の可視化ならびに評価

小田紘久, Holger R. Roth, 砂口尚輝, 宇佐美紀子, 小田昌宏, 島雄大介, 市原周, 湯浅哲也, 安藤正海, 秋田利明, 成田裕司, 森健策

第37回日本医用画像工学会大会予稿集頁： OP1-8 2018年7月

機械学習を用いた腹部動脈血管名自動命名における臓器情報および多血管相互関係利用方法の検討

鉄村悠介, Holger Roth, 林雄一郎, 小田昌宏, 三澤一成, 森健策

第37回日本医用画像工学会大会予稿集頁： OP14-2 2018年7月

隣接復元を用いたSLAMベースの気管支鏡追跡の改善

王成, 小田昌宏, 林雄一郎, 本間裕敏, 高畑博嗣, 森雅樹, 名取博, 森健策

第37回日本医用画像工学会大会予稿集頁： OP13-6 2018年7月

胃の変形情報を利用した経時リンパ節の自動対応付け手法の精度向上に関する研究

舘高基, 小田昌宏, 林雄一郎, 中村嘉彦, 北坂孝幸, 三澤一成, 森健策

第37回日本医用画像工学会大会予稿集頁： OP4-2 2018年7月

Micro-focus X-ray CT of the heart:A comparison with X-ray refraction-contrast CT,

Hirohisa Oda, Holger R. Roth, Naoki Sunaguchi, Daisuke Shimao, Takaaki Sugino, Masahiro Oda, Toshiaki Akita, Yuji Narita, Shu Ichihara, Tetsuya Yuasa, Masami Ando, Kensaku Mori

International Journal of Computer Assisted Radiology and Surgery 13 巻 ( 1 ) 頁： s140-142 2018年6月

Polyp detection in colonoscopic videos by using spatio-temporal feature

Hayato Itoh, Holger R. Roth, Masashi Misawa, Yuichi Mori, Masahiro Oda, Shin-ei Kudo, Kensaku Mori

International Journal of Computer Assisted Radiology and Surgery 13 巻 ( 1 ) 頁： s97-98 2018年6月

Automated ganglion cell detection using fully convolutional networks and evaluation under different training losses

Hirohisa Oda, Kosuke Chiba, Holger R. Roth, Takayuki Kitasaka, Masahiro Oda, Akinari Hinoki, Hiroo Uchida, Kensaku Mori

International Journal of Computer Assisted Radiology and Surgery 13 巻 ( 1 ) 頁： s104-106 2018年6月

Evaluation of 3D fully convolutional networks for multi-class organ segmentation in contrast-enhanced CT

Chen Shen, Holger R. Roth, Hirohisa Oda, Masahiro Oda, Yuichiro Hayashi, Kazunari Misawa, Tadaaki Miyamoto, Kensaku Mori

International Journal of Computer Assisted Radiology and Surgery 13 巻 ( 1 ) 頁： s21-22 2018年6月

Improvement of robustness of SLAM-based bronchoscope tracking by posture guided feature matching

Cheng Wang, Masahiro Oda, Yuichiro Hayashi, Hirotoshi Honma, Hirotsugu Takabatake, Masaki Mori, Hiroshi Natori, Kensaku Mori

International Journal of Computer Assisted Radiology and Surgery 13 巻 ( 1 ) 頁： s11-12 2018年6月

Artificial Intelligence-Assisted Polyp Detection for Colonoscopy: Initial Experience 査読有り

Misawa Masashi, Kudo Shin-ei, Mori Yuichi, Cho Tomonari, Kataoka Shinichi, Yamauchi Akihiro, Ogawa Yushi, Maeda Yasuharu, Takeda Kenichi, Ichimasa Katsuro, Nakamura Hiroki, Yagawa Yusuke, Toyoshima Naoya, Ogata Noriyuki, Kudo Toyoki, Hisayuki Tomokazu, Hayashi Takemasa, Wakamura Kunihiko, Baba Toshiyuki, Ishida Fumio, Itoh Hayato, Roth Holger, Oda Masahiro, Mori Kensaku

GASTROENTEROLOGY 154 巻 ( 8 ) 頁： 2027 - + 2018年6月

DIAGNOSTIC IMAGING SYSTEM WITH VIRTUAL ENTEROSCOPY AND COMPUTER-AIDED DETECTION FOR EVALUATION OF SMALL BOWEL LESIONS OF CROHN'S DISEASE 査読有り

Furukawa Kazuhiro, Miyahara Ryoji, Funasaka Kohei, Suhara Hiroki, Matsushita Masanobu, Yamamura Takeshi, Ishikawa Takuya, Ohno Eizaburo, Nakamura Masanao, Kawashima Hiroki, Watanabe Osamu, Oda Masahiro, Mori Kensaku, Hirooka Yoshiki, Goto Hidemi

GASTROINTESTINAL ENDOSCOPY 87 巻 ( 6 ) 頁： AB304-AB304 2018年6月

An application of cascaded 3D fully convolutional networks for medical image segmentation 査読有り

Holger R. Roth, Hirohisa Oda, Xiangrong Zhou, Natsuki Shimizu, Ying Yang, Yuichiro Hayashi, Masahiro Oda, Michitaka Fujiwara, Kazunari Misawa, Kensaku Mori

COMPUTERIZED MEDICAL IMAGING AND GRAPHICS 66 巻頁： 90 - 99 2018年6月

DIAGNOSTIC IMAGING SYSTEM WITH VIRTUAL ENTEROSCOPY AND COMPUTER-AIDED DETECTION FOR EVALUATION OF SMALL BOWEL LESIONS OF CROHN'S DISEASE 査読有り

GASTROINTESTINAL ENDOSCOPY 87 巻 ( 6 ) 頁： AB304-AB304 2018年6月

3D U-Netと測地距離カーネルを取り入れた全連結条件付き確率場に基づく医用画像からの多臓器自動抽出

楊瀛, Roth Holger, 小田昌宏, 北坂孝幸, 三澤一成, 森健策

電子情報通信学会技術研究報告(MI) 117 巻 ( 518 ) 頁： 75-80 2018年3月

CNNによる回帰を用いた臓器領域の位置推定手法の初期的検討

清水南月, 小田昌宏, ロスホルガー, 林雄一郎, 三澤一成, 藤原道隆, 森健策

電子情報通信学会技術研究報告(MI) 117 巻 ( 518 ) 頁： 81-86 2018年3月

Pre/intra-operative diagnosis and navigational assistance based on multidisciplinary computational anatomy

Kensaku Mori, Masahiro Oda, Holger R roth, Yoshihiko Nakamura, Yoshito Mekada, Takayuki Kitasaka, Kazunari Misawa, Michitaka Fujiwara, Kazuhiro Durukawa, Shu Ichihara

第4回多元計算解剖学国際シンポジウム頁： 87-105 2018年3月

ディープラーニングを用いた教師なし学習によるレジストレーション手法の初期的検討

長柄快, Holger R. Roth, 中村彰太, 小田昌宏, 森健策

電子情報通信学会技術研究報告(MI) 117 巻 ( 518 ) 頁： 7-12 2018年3月

複数のステレオ内視鏡画像からの臓器形状復元の定量評価

柴田睦実, 林雄一郎, 小田昌宏, 三澤一成, 森健策

電子情報通信学会技術研究報告(MI) 117 巻 ( 518 ) 頁： 117-122 2018年3月

超拡大内視鏡画像における腫瘍性ポリープ分類に向けたグラスマン距離に基づく特徴選択法

伊東隼人, 森悠一, 三澤将史, 小田昌宏, 工藤進英, 森健策

電子情報通信学会技術研究報告(MI) 117 巻 ( 518 ) 頁： 51-56 2018年3月

Develop and Validate a Finite Element Method Model for Deformation Matching of Laparoscopic Gastrectomy Navigation 査読有り

Chen Tao, Wei Guodong, Shi Weili, Hayashi Yuichiro, Oda Masahiro, Jiang Zhengang, Li Guoxin, Mori Kensaku

MEDICAL IMAGING 2018: IMAGE-GUIDED PROCEDURES, ROBOTIC INTERVENTIONS, AND MODELING 10576 巻 2018年

デスクトップ型マイクロCTによる微細解剖構造イメージング招待有り

森健策, 中村彰太, 秋田利明, 小田紘久, ホルガーロス, 小田昌宏

Ｍｅｄｉｃａｌ　Ｉｍａｇｉｎｇ　Ｔｅｃｈｎｏｌｏｇｙ 36 巻 ( 3 ) 頁： 127-132 2018年

記述言語：日本語掲載種別：研究論文（研究会，シンポジウム資料等）

本稿では，デスクトップ型マイクロCTを用いた微細構造イメージングについて述べる．臨床の場で利用されるX線CT装置は，その解像度がおおよそ1ボクセルあたり0.5mmから1mm程度である．このようなイメージング装置を用いて得ることができる画像は，このボクセル解像度に準じた解剖構造を得ることができる．一方，マイクロCT装置を用いれば，1μmから50μm程度の解像度で撮影できる．本稿では，マイクロCTによって撮影された肺標本ならびに心臓標本を示し，その可能性について述べる．

DOI： 10.11409/mit.36.127

Fine Segmentation of Tiny Blood Vessel Based on Fully-Connected Conditional Random Field 査読有り

Wang Chenglong, Oda Masahiro, Yoshino Yasushi, Yamamoto Tokunori, Mori Kensaku

MEDICAL IMAGING 2018: IMAGE PROCESSING 10574 巻 2018年

Dense volumetric detection and segmentation of mediastinal lymph nodes in chest CT images 査読有り

Oda Hirohisa, Roth Holger R., Bhatia Kanwal K., Oda Masahiro, Kitasaka Takayuki, Iwano Shingo, Homma Hirotoshi, Takabatake Hirotsugu, Mori Masaki, Natori Hiroshi, Schnabel Julia A., Mori Kensaku

MEDICAL IMAGING 2018: COMPUTER-AIDED DIAGNOSIS 10575 巻 2018年

Deep Learning and Its Application to Medical Image Segmentation 招待有り

ROTH Holger R., SHEN Chen, ODA Hirohisa, ODA Masahiro, HAYASHI Yuichiro, MISAWA Kazunari, MORI Kensaku

Medical Imaging Technology 36 巻 ( 2 ) 頁： 63-71 2018年

記述言語：英語掲載種別：研究論文（研究会，シンポジウム資料等）

One of the most common tasks in medical imaging is semantic segmentation. Achieving this segmentation automatically has been an active area of research, but the task has been proven very challenging due to the large variation of anatomy across different patients. However, recent advances in deep learning have made it possible to significantly improve the performance of image recognition and semantic segmentation methods in the field of computer vision. Due to the data driven approaches of hierarchical feature learning in deep learning frameworks, these advances can be translated to medical images without much difficulty. Several variations of deep convolutional neural networks have been successfully applied to medical images. Especially fully convolutional architectures have been proven efficient for segmentation of 3D medical images. In this article, we describe how to build a 3D fully convolutional network (FCN) that can process 3D images in order to produce automatic semantic segmentations. The model is trained and evaluated on a clinical computed tomography (CT) dataset and shows stateof-the-art performance in multi-organ segmentation.

DOI： 10.11409/mit.36.63

Cascade classification of endocytoscopic images of colorectal lesions for automated pathological diagnosis 査読有り

Itoh Hayato, Mori Yuichi, Misawa Masashi, Oda Masahiro, Kudo Shin-ei, Mori Kensaku

MEDICAL IMAGING 2018: COMPUTER-AIDED DIAGNOSIS 10575 巻 2018年

BESNet: Boundary-Enhanced Segmentation of Cells in Histopathological Images 査読有り

Oda Hirohisa, Roth Holger R., Chiba Kosuke, Sokolic Jure, Kitasaka Takayuki, Oda Masahiro, Hinoki Akinari, Uchida Hiroo, Schnabel Julia A., Mori Kensaku

MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION - MICCAI 2018, PT II 11071 巻頁： 228-236 2018年

Automatic segmentation of eyeball structures from micro-CT images based on sparse annotation 査読有り

Sugino Takaaki, Roth Holger R., Oda Masahiro, Omata Seiji, Sakuma Shinya, Arai Fumihito, Mori Kensaku

MEDICAL IMAGING 2018: BIOMEDICAL APPLICATIONS IN MOLECULAR, STRUCTURAL, AND FUNCTIONAL IMAGING 10578 巻 2018年

Airway Segmentation from 3D Chest CT Volumes Based on Volume of Interest Using Gradient Vector Flow

MENG Qier, KITASAKA Takayuki, ODA Masahiro, UENO Junji, MORI Kensaku

Ｍｅｄｉｃａｌ　Ｉｍａｇｉｎｇ　Ｔｅｃｈｎｏｌｏｇｙ 36 巻 ( 3 ) 頁： 133-146 2018年

記述言語：英語掲載種別：研究論文（学術雑誌）

In this paper, we propose a new airway segmentation algorithm from 3D chest CT volumes based on the volume of interest (VOI). The algorithm segments each bronchial branch by recognizing the airway regions from the trachea using the VOIs to segment each branch. A VOI is placed to envelop the branch currently being processed. Then a cavity enhancement filter is performed only inside the current VOI so that each branch is extracted. At the same time, we perform a leakage detection scheme to avoid any leakage regions inside the VOI. Next the gradient vector flow magnitude map and a tubular-likeness function are computed in each VOI. This assists the predictions of both the position and direction of the next child VOIs to detect the next child branches to continue the tracking algorithm. Finally, we unify all of the extracted airway regions to form a complete airway tree. We used a dataset that includes 50 standard-dose human chest CT volumes to evaluate our proposed algorithm. The average extraction rate was approximately 78.1% with a significantly decreased false positive rate compared to the previous method.

DOI： 10.11409/mit.36.133

A Multi-scale Pyramid of 3D Fully Convolutional Networks for Abdominal Multi-organ Segmentation 査読有り

Roth Holger R., Shen Chen, Oda Hirohisa, Sugino Takaaki, Oda Masahiro, Hayashi Yuichiro, Misawa Kazunari, Mori Kensaku

MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION - MICCAI 2018, PT IV 11073 巻頁： 417 - 425 2018年

医用画像処理におけるディープラーニング利用入門-序文- 招待有り

小田昌宏

Ｍｅｄｉｃａｌ　Ｉｍａｇｉｎｇ　Ｔｅｃｈｎｏｌｏｇｙ 36 巻 ( 2 ) 頁： 45 - 46 2018年

Towards Automated Colonoscopy Diagnosis: Binary Polyp Size Estimation via Unsupervised Depth Learning 査読有り

Itoh Hayato, Roth Holger R., Lu Le, Oda Masahiro, Misawa Masashi, Mori Yuichi, Kudo Shin-ei, Mori Kensaku

MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION - MICCAI 2018, PT II 11071 巻頁： 611 - 619 2018年

Towards dense volumetric pancreas segmentation in CT using 3D fully convolutional networks

Roth Holger, Oda Masahiro, Shimizu Natsuki, Oda Hirohisa, Hayashi Yuichiro, Kitasaka Takayuki, Fujiwara Michitaka, Misawa Kazunari, Mori Kensaku

MEDICAL IMAGING 2018: IMAGE PROCESSING 10574 巻 2018年

Unsupervised Pathology Image Segmentation Using Representation Learning with Spherical K-means 査読有り

Moriya Takayasu, Roth Holger H., Nakamura Shota, Oda Hirohisa, Nagara Kai, Oda Masahiro, Mori Kensaku

MEDICAL IMAGING 2018: DIGITAL PATHOLOGY 10581 巻 2018年

Unsupervised Segmentation of 3D Medical Images Based on Clustering and Deep Representation Learning 査読有り

Moriya Takayasu, Roth Holger R., Nakamura Shota, Oda Hirohisa, Nagara Kai, Oda Masahiro, Mori Kensaku

MEDICAL IMAGING 2018: BIOMEDICAL APPLICATIONS IN MOLECULAR, STRUCTURAL, AND FUNCTIONAL IMAGING 10578 巻 2018年

ディープラーニング活用の重要ポイント招待有り

小田昌宏

Ｍｅｄｉｃａｌ　Ｉｍａｇｉｎｇ　Ｔｅｃｈｎｏｌｏｇｙ 36 巻 ( 2 ) 頁： 72-75 2018年

担当区分：筆頭著者記述言語：日本語掲載種別：研究論文（研究会，シンポジウム資料等）

ディープラーニングは医用画像処理のさまざまなタスク解決に使用することができる便利な手法である．ディープラーニングを研究で活用し，その性能を引き出すために注意を払うべき点がいくつかある．その中で，ディープラーニングにおける多数の手動指定パラメーターの選択，データオーギュメンテーションを使用する上で注意する点，ディープラーニングのランダム要素が与える影響について解説する．

DOI： 10.11409/mit.36.72

Fully Convolutional Network-Based Eyeball Segmentation from Sparse Annotation for Eye Surgery Simulation Model 査読有り

Sugino Takaaki, Roth Holger R., Oda Masahiro, Mori Kensaku

International Workshop on Bio-Imaging and Visualization for Patient-Customized Simulations, BIVPCS 2018 11042 巻頁： 118-126 2018年

Machine learning-based colon deformation estimation method for colonoscope tracking 査読有り

Oda Masahiro, Kitasaka Takayuki, Furukawa Kazuhiro, Miyahara Ryoji, Hirooka Yoshiki, Goto Hidemi, Navab Nassir, Mori Kensaku

MEDICAL IMAGING 2018: IMAGE-GUIDED PROCEDURES, ROBOTIC INTERVENTIONS, AND MODELING 10576 巻 2018年

低ばね剛性機械構造を利用した積層造形臓器モデルの CNC 砥粒加工技術の開発

大庭孝之, 鈴木教和, 社本英二, 小田昌宏, 森健策, 竹内芳美

生産加工・工作機械部門講演会 : 生産と加工に関する学術講演会 2018.12 巻 ( 0 ) 頁： C20 2018年

On the influence of Dice loss function in multi-class organ segmentation of abdominal CT using 3D fully convolutional networks

Chen Shen, Holger R. Roth, Hirohisa Oda, Masahiro Oda, Yuichiro Hayashi, Kazunari Misawa, Kensaku Mori

MI2017-51 117 巻 ( 281 ) 頁： 15-20 2017年11月

Study on the Robustness of ORB-SLAM Based Outlier Elimination in Bronchoscope Tracking -- RANSAC + EPnP for Outlier Detection --

Cheng Wang, Masahiro Oda, Yuichiro Hayashi, Hirotoshi Honma, Hirotsugu Takabatake, Masaki Mori, Hiroshi Natori, Kensaku Mori

MI2017-58 117 巻 ( 281 ) 頁： 47-52 2017年11月

Automated Multi-Organ Segmentation in Abdominal CT with Hierarchical 3D Fully-Convolutional Networks

Holger R. Roth, Hirohisa Oda, MENG, Yuichiro Hayashi, Masahiro Oda, Natsuki Shimizu, Kensaku Mori, Michitaka Fujiwara, Kazunari Misawa

RSNA 2017 (Radiological Society of North America) Scientific Assembly and Annual Meeting PROGRAM IN BRIEF 頁： 267 2017年11月

Automated mediastinal lymph node detection from CT volumes based on intensity targeted radial structure tensor analysis 査読有り国際共著

Hirohisa Oda, Kanwal K. Bhatia, Masahiro Oda, Takayuki Kitasaka, Shingo Iwano, Hirotoshi Homma, Hirotsugu Takabatake, Masaki Mori, Hiroshi Natori, Julia A. Schnabel, Kensaku Mori

JOURNAL OF MEDICAL IMAGING 4 巻 ( 4 ) 頁： 044502 2017年10月

3D Fully Convolutional Networks と全連結条件付確率場による 3 次元 CT 画像からの多臓器自動抽出に関する検討

楊瀛, 小田昌宏, Roth Holger, 北坂孝幸, 三澤一成, 森健策

日本コンピュータ外科学会誌第26回日本コンピュータ外科学会大会特集号 19 巻 ( 4 ) 頁： 268-269 2017年10月

μCT 画像を用いた大変形を含む連続切片 HE 染色画像の 3 次元再構築

長柄快, Holger Roth, 中村彰太, 小田紘久, 守谷享泰, 小田昌宏, 森健策

日本コンピュータ外科学会誌第26回日本コンピュータ外科学会大会特集号 19 巻 ( 4 ) 頁： 363-364 2017年10月

レベルセット法を用いた腎臓皮質と髄質領域の分割

王成龍, 小田昌宏, 永山洵, 吉野能, 山本徳則, 森健策

日本コンピュータ外科学会誌第26回日本コンピュータ外科学会大会特集号 19 巻 ( 4 ) 頁： 266-267 2017年10月

マイクロ CT を用いた膵臓パラフィンブロック標本の解析

進藤幸治, 大内田研宙, Holger R. Roth, 小田紘久, 岩本千佳, 小田昌宏, 中村雅史, 森健策, 橋爪誠

日本コンピュータ外科学会誌第26回日本コンピュータ外科学会大会特集号 19 巻 ( 4 ) 頁： 244-245 2017年10月

サポートベクタマシンを用いたラジオミクスベースの消化管間質性腫瘍リスク評価システム

陳韜, 小田紘久, Holger R. Roth, 北坂孝幸, 小田昌宏, 李国新, 森健策

日本コンピュータ外科学会誌第26回日本コンピュータ外科学会大会特集号 19 巻 ( 4 ) 頁： 239-240 2017年10月

機械学習を用いた腹部動脈血管名自動命名における臓器情報利用方法に関する一考察

鉄村悠介, Holger Roth, 林雄一郎, 小田昌宏, 進藤幸治, 大内田研宙, 橋爪誠, 三澤一成, 森健策

日本コンピュータ外科学会誌第26回日本コンピュータ外科学会大会特集号 19 巻 ( 4 ) 頁： 361-362 2017年10月

超拡大大腸内視鏡画像を利用した病理自動診断〜腫瘍性病変に関する分類精度解析〜

伊東隼人, 森悠一, 三澤将史, 小田昌宏, 工藤進英, 森健策

日本コンピュータ外科学会誌第26回日本コンピュータ外科学会大会特集号 19 巻 ( 4 ) 頁： 319-320 2017年10月

複数フレームのステレオ内視鏡画像を用いた臓器表面形状復元に関する検討

柴田睦実, 林雄一郎, 小田昌宏, 三澤一成, 森健策

日本コンピュータ外科学会誌第26回日本コンピュータ外科学会大会特集号 19 巻 ( 4 ) 頁： 326-327 2017年10月

自動設計特徴量を用いた 3 次元腹部 CT 像における膵臓領域の位置推定

清水南月, Holger R. Roth, 小田昌宏, 三澤一成, 藤原道隆, 森健策

日本コンピュータ外科学会誌第26回日本コンピュータ外科学会大会特集号 19 巻 ( 4 ) 頁： 270-271 2017年10月

深層学習を用いたマイクロ CT 画像からの眼球構造自動抽出〜少量データ学習による解剖構造抽出性能の検証

杉野貴明, Holger R. Roth, 小田昌宏, 小俣誠二, 佐久間臣耶, 新井史人, 森健策

日本コンピュータ外科学会誌第26回日本コンピュータ外科学会大会特集号 19 巻 ( 4 ) 頁： 241-242 2017年10月

気管支鏡追跡における ORB-SLAM 適用に関する初期的検討

王成, 小田昌宏, 林雄一郎, 森健策

日本コンピュータ外科学会誌第26回日本コンピュータ外科学会大会特集号 19 巻 ( 4 ) 頁： 324-325 2017年10月

Virtual 3D microscope and magnified 3D print for naked eye analyses of alveoli and alveolar duct structures by Heitzman lung specimen with micro CT

Hiroshi Natori, Masaki Mori, Hirotsugu Takabatake, Hirotoshi Homma, ensaku Mori, Masahiro Oda, Hiroyuki Koba, Hiroki Takahashi

ERS International congress 2017 頁： Session 439 2017年9月

Accuracy of diagnosing invasive colorectal cancer using computer-aided endocytoscopy 査読有り

Kenichi Takeda, Shin-ei Kudo, Yuichi Mori, Masashi Misawa, Toyoki Kudo, Kunihiko Wakamura, Atsushi Katagiri, Toshiyuki Baba, Eiji Hidaka, Fumio Ishida, Haruhiro Inoue, Masahiro Oda, Kensaku Mori

ENDOSCOPY 49 巻 ( 8 ) 頁： 798 - 802 2017年8月

Multi-atlas pancreas segmentation: Atlas selection based on vessel structure 査読有り国際共著

Ken'ichi Karasawa, Masahiro Oda, Takayuki Kitasaka, Kazunari Misawa, Michitaka Fujiwara, Chengwen Chu, Guoyan Zheng, Daniel Rueckert, Kensaku Mori

MEDICAL IMAGE ANALYSIS 39 巻頁： 18 - 28 2017年7月

K-means法とJoint Unsupervised Learningによる3次元医用画像の教師なしセグメンテーション(Unsupervised 3D Medical Image Segmentation using K-means and Joint Unsupervised Learning) 査読有り

守谷享泰, Roth Holger R., 中村彰太, 小田紘久, 長柄快, 小田昌宏, 森健策

日本医用画像工学会大会予稿集 36回巻頁： 546 - 548 2017年7月

記述言語：英語出版者・発行元：日本医用画像工学会

本論文では,3次元医用画像向けの新しい教師なしセグメンテーションの手法を提案する.提案手法は2つの段階に分けられる.1つ目の段階では,JULEを用いた深層表現学習をおこなう.JULEは,CNNから出力される表現のクラスタリングと,クラスタラベルを教師信号としたCNNの更新を繰り返す手法である.2つ目の段階では,訓練済みのCNNから生成された深層表現に対し,K-means法を適用してセグメンテーションをおこなう.評価には肺がん標本のマイクロCT画像を用い,浸潤領域,非浸潤領域,正常領域という3つの領域に分けることを試みた.マイクロCT画像上で病理組織学的特徴に基づいたセグメンテーションをおこなうことは,将来の病理診断精度の向上につながる.セグメンテーション結果の定性評価から,深層表現が3次元医用画像のセグメンテーションに有用であることが示された.(著者抄録)

Accuracy of computer-aided diagnosis based on narrow-band imaging endocytoscopy for diagnosing colorectal lesions: comparison with experts 査読有り

3DU-Netによる3次元胸部CT像からのリンパ節検出

小田紘久, KanwalK.Bhatia, HolgerR.Roth, 小田昌宏, 北坂孝幸, 岩野信吾, 本間裕敏, 高畠博嗣, 森雅樹, 名取博, JuliaA.Schnabel, 森健策

第36回日本医用画像工学会大会予稿集頁： OP1-6 2017年7月

Improvement on Robustness of ORB-SLAM Based Surgical Navigation System by Building Submap

王成, 小田昌宏, 林雄一郎, 三澤一成, 森健策

第36回日本医用画像工学会大会予稿集頁： OP2-6 2017年7月

条件付き確率場による医用画像からの多臓器抽出におけるHigher Order Potential とボクセル連結構造の影響に関する考察

楊瀛, 小田昌宏, Roth Holger, 北坂孝幸, 三澤一成, 森健策

第36回日本医用画像工学会大会予稿集頁： OP16-4 2017年7月

マイクロ CT 画像情報を利用した特徴点対応付けに基づく顕微鏡画像の 3 次元再構築

長柄快, Holger R. ROTH, 中村彰太, 小田紘久, 守谷享泰, 小田昌宏, 森健策

第36回日本医用画像工学会大会予稿集頁： OP14-1 2017年7月

ステレオ内視鏡画像からの臓器形状復元手法における複数フレームの利用に関する初期的検討

柴田睦実, 林雄一郎, 小田昌宏, 三澤一成, 森健策

第36回日本医用画像工学会大会予稿集頁： OP2-8 2017年7月

Torso organ segmentation in CT using fine-tuned 3D fully convolutional networks

Holger ROTH, Ying YANG, Masahiro ODA, Hirohisa ODA, Yuichiro HAYASHI, Natsuki SHIMIZU, Takayuki KITASAKA, Michitaka FUJIWARA, Kazunari MISAWA, Kensaku MORI

第36回日本医用画像工学会大会予稿集頁： OP1-8 2017年7月

K-means 法と Joint Unsupervised Learning による3次元医用画像の教師なしセグメンテーション

守谷享泰, Holger R. Roth, 中村彰太, 小田紘久, 長柄快, 小田昌宏, 森健策

第36回日本医用画像工学会大会予稿集頁： OP16-5 2017年7月

機械学習を用いた腹部動脈血管名自動命名における肝動脈分岐情報利用方法に関する一考察

鉄村悠介, 張暁楠, Holger Roth, 林雄一郎, 小田昌宏, 三澤一成, 森健策

第36回日本医用画像工学会大会予稿集頁： OP6-1 2017年7月

血管情報を用いた経時リンパ節の自動対応付け手法に関する研究

舘高基, 小田昌宏, 中村嘉彦, 寶珠山裕, 三澤一成, 森健策

第36回日本医用画像工学会大会予稿集頁： OP15-4 2017年7月

Feature-based registration of micro CT volumes

Kai Nagara, Shota Nakamura, Hoiger R. Roth, Masahiro Oda, Hirotoshi Homma, Hirotsugu Takabatake, Masaki Mori, Hiroshi Natori, Kesaku Mori

International Journal of Computer Assisted Radiology and Surgery 12 巻頁： S201-S203 2017年6月

Multi-organ segmentation in abdominal CT using 3D fully convolutional networks

Holger R. Roth, Masahiro Oda, Yuichiro Hayashi, Hirohisa Oda, Kensaku Mori

International Journal of Computer Assisted Radiology and Surgery 12 巻頁： S55-S57 2017年6月

Automatic anatomical labeling of arteries and veins using conditional random fields 査読有り

Takayuki Kitasaka, Mitsuru Kagajo, Yukitaka Nimura, Yuichiro Hayashi, Masahiro Oda, Kazunari Misawa, Kensaku Mori

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY 12 巻 ( 6 ) 頁： 1041 - 1048 2017年6月

Masashi Misawa, Shin-ei Kudo, Yuichi Mori, Kenichi Takeda, Yasuharu Maeda, Shinichi Kataoka, Hiroki Nakamura, Toyoki Kudo, Kunihiko Wakamura, Takemasa Hayashi, Atsushi Katagiri, Toshiyuki Baba, Fumio Ishida, Haruhiro Inoue, Yukitaka Nimura, Masahiro Oda, Kensaku Mori

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY 12 巻 ( 5 ) 頁： 757-766 2017年6月

Diagnostic Ability of Automated Diagnosis System Using Endocytoscopy for Invasive Colorectal Cancer 査読有り

Takeda Kenichi, Kudo Shinei, Mori Yuichi, Kataoka Shinichi, Yasuharu Maeda, Ogawa Yushi, Nakamura Hiroki, Misawa Masashi, Kudo Toyoki, Wakamura Kunihiko, Hayashi Takemasa, Katagiri Atsushi, Baba Toshiyuki, Hidaka Eiji, Ishida Fumio, Inoue Haruhiro, Oda Masahiro, Mori Kensaku

GASTROINTESTINAL ENDOSCOPY 85 巻 ( 5 ) 頁： AB408-AB408 2017年5月

Artificial Intelligence for Endocytoscopy Provides Fully Automated Diagnosis of Histological Remission in Ulcerativ E. Coli Tis 査読有り

Yasuharu Maeda, Kudo Shinei, Mori Yuichi, Misawa Masashi, Wakamura Kunihiko, Hayashi Seiko, Ogata Noriyuki, Takeda Kenichi, Kudo Toyoki, Hayashi Takemasa, Katagiri Atsushi, Ishida Fumio, Ohtsuka Kazuo, Oda Masahiro, Mori Kensaku

GASTROINTESTINAL ENDOSCOPY 85 巻 ( 5 ) 頁： AB248-AB248 2017年5月

Computer-Aided Diagnosis Based on Endocytoscopy With Narrow-Band Imaging Allows Accurate Diagnosis of Diminutive Colorectal Lesions 査読有り

Misawa Masashi, Kudo Shinei, Mori Yuichi, Takeda Kenichi, Kataoka Shinichi, Nakamura Hiroki, Maeda Yasuharu, Ogawa Yushi, Yamauchi Akihiro, Igarashi Kenta, Hayashi Takemasa, Kudo Toyoki, Wakamura Kunihiko, Katagiri Atsushi, Baba Toshiyuki, Ishida Fumio, Oda Masahiro, Mori Kensaku

GASTROINTESTINAL ENDOSCOPY 85 巻 ( 5 ) 頁： AB57-AB57 2017年5月

Can Artificial Intelligence Correctly Diagnose Sessile Serrated Adenomas/Polyps? 査読有り

Mori Yuichi, Kudo Shinei, Ogawa Yushi, Misawa Masashi, Takeda Kenichi, Kudo Toyoki, Wakamura Kunihiko, Hayashi Takemasa, Ichimasa Katsuro, Maeda Yasuharu, Toyoshima Naoya, Nakamura Hiroki, Katagiri Atsushi, Baba Toshiyuki, Ishida Fumio, Oda Masahiro, Mori Kensaku, Inoue Haruhiro

GASTROINTESTINAL ENDOSCOPY 85 巻 ( 5 ) 頁： AB510-AB510 2017年5月

Diagnostic Ability of Automated Diagnosis System Using Endocytoscopy for Invasive Colorectal Cancer 査読有り

GASTROINTESTINAL ENDOSCOPY 85 巻 ( 5 ) 頁： AB408-AB408 2017年5月

Automatic segmentation of airway tree based on local intensity filter and machine learning technique in 3D chest CT volume 査読有り

Qier Meng, Takayuki Kitasaka, Yukitaka Nimura, Masahiro Oda, Junji Ueno, Kensaku Mori

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY 12 巻 ( 2 ) 頁： 245 - 261 2017年2月

GPU Implementation of SLIC Supervoxel Oversegmentation

Hirohisa Oda, Kanwal K. Bhatia, Masahiro Oda, Takayuki Kitasaka, Shingo Iwano, Julia A. Schnabel, Kensaku Mori

International Forum on Medical Imaging in Asia (IFMIA) 頁： 266-268 2017年1月

Multi-scale Image Fusion Between Pre-operative Clinical CT and X-ray microtomography of Lung Pathology

Holger Roth, Kai Nagara, Hirohisa Oda, Masahiro Oda, Tomoshi Sugiyama, Shota Nakamura, Kensaku Mori

International Forum on Medical Imaging in Asia (IFMIA) 頁： 54-56 2017年1月

Deep-Learning-Based Segmentation for the Head Sectioned Images of the Visible Korean Project

Mohammad Eshghi, Holger R. Roth, Hirohisa Oda, Masahiro Oda, Min Suk Chung, Kensaku Mori

MI2016-119 116 巻 ( 393 ) 頁： 191-194 2017年1月

Structured Random Forestを用いた3次元腹部CT像からのリンパ節自動検出

寳珠山裕, Holger Roth, 小田昌宏, 中村嘉彦, 三澤一成, 藤原道隆, 森健策

電子情報通信学会技術研究報告(MI) 116 巻 ( 393 ) 頁： 23-28 2017年1月

MISTライブラリのためのGPUプログラミング

小田紘久, 小田昌宏, 北坂孝幸, 森健策

電子情報通信学会技術研究報告(MI) 116 巻 ( 393 ) 頁： 133-136 2017年1月

Influence of Voxel-Connection Structure in Organ Segmentation Based on Conditional Random Field

Ying Yang, Masahiro Oda, Kazunari Misawa, Daniel Rueckert, Kensaku Mori

MI2016-112 116 巻 ( 393 ) 頁： 157-162 2017年1月

Robust colonoscope tracking method for colon deformations utilizing coarse-to-fine correspondence findings 査読有り国際共著

Masahiro Oda, Hiroaki Kondo, Takayuki Kitasaka, Kazuhiro Furukawa, Ryoji Miyahara ,Yoshiki Hirooka, Hidemi Goto, Nassir Navab, Kensaku Mori

International Journal of Computer Assisted Radiology and Surgery 12 巻 ( 1 ) 頁： 39-50 2017年1月

Extracellular matrix directions estimation of the heart on micro-focus X-ray CT volumes 査読有り

Oda Hirohisa, Oda Masahiro, Kitasaka Takayuki, Akita Toshiaki, Mori Kensaku

MEDICAL IMAGING 2017: BIOMEDICAL APPLICATIONS IN MOLECULAR, STRUCTURAL, AND FUNCTIONAL IMAGING 10137 巻 2017年

異なった3Dプリンタと素材を用いた心臓モデルの作成と有用性の検討査読有り

村田雅登, 平山貢大, 武内新作, 小田昌宏, 森建策, 仁木清美

生体医工学 55 巻 ( 4 ) 頁： 273-273 2017年

記述言語：日本語掲載種別：研究論文（学術雑誌）

[目的] 3Dプリンタは手術支援の一つとしての期待が高まっているが、現在の汎用機で、どの位正確に病変を表すことが可能であるかに関して不明である。そこで本研究では、異なった3Dプリンタ造形法と素材を用いて心臓モデルを作成し有用性を検討した。［方法］CT画像から抽出した心臓データをstlファイルに変換することにより、３Ｄプリンタによる心臓模型をつくった。その際、二種類の3Dプリンタにおける造形法を試した。一つ目は材料にPLAを使用した熱溶解積層法、二つ目は材料にAR-G1Lを使用したインクジェット法である。[結果] 熱溶解積層法を用いて作成したものは、原寸大では印刷できず75％に縮小し印刷した。その結果、血管部分の樹脂が崩れてしまった、また、サイズが小さくなってしまったことに加え、モデル全体が単色であることから、血管がどこにあるかということが一目で分かりにくく、さらに材質が固く力を加えた際に壊れやすいという問題があった。インクジェット方式を用いて作成したものは熱溶解積層法とは異なり原寸大での作成が可能であった。病変部位を正確に表現できた。さらにゴム素材を使用可能でき塑性に富んでいた。[結論]インクジェット方式を採用し作成した心臓模型は塑性に優れ有用である。

DOI： 10.11239/jsmbe.55Annual.273

3Dイメージと3D印刷されたオブジェクトの利用が熟達者と初心者の空間的メンタルモデル形成に及ぼす影響

機械学習を用いた内視鏡画像自動診断

Mori Kensaku, Oda Masahiro, Misawa Masashi, Mori Yuichi, Kudo Shinei

生体医工学 55 巻 ( 4 ) 頁： 344-344 2017年

記述言語：日本語掲載種別：研究論文（研究会，シンポジウム資料等）

本講演では、機械学習を利用した内視鏡画像の自動診断手法について紹介する。特に、超拡大内視鏡画像を用いた大腸ポリープの類型判別の自動診断手法を例に挙げながら、内視鏡画像診断における機械学習の有用性について議論したい。パーセプトロン、統計的パターン認識に始まる機械学習は長年研究が行われてきたが、GPGPUなどに代表される手軽なハイパフォーマンスコンピューティング技術の発展により、非常に複雑なアーキテクチャを持つニューラルネットワークを用いたパターン認識などが可能となった。一方、内視鏡画像の診断には高度な技術が必要とされ、医師間の差による診断のブレも発生する。そこで、機械学習を用いた内視鏡画像の自動診断方法について２，３紹介する。一つ目の方法は、Hand-crafted特徴量（微分特徴量）などを求め、その特徴量により大腸ポリープの組織型をSVM (Support Vector Machine)による自動分類するものである。二つ目の方法は、CNN (Convolutional Neural Network)を用いて内視鏡画像の自動分類を行うものである。これらの手法について、技術的な側面に焦点をあて解説を行う。機械学習において重要な学習データ生成法についても合わせて議論したい。

DOI： 10.11239/jsmbe.55Annual.344

Motion Vector for Outlier Elimination in Feature Matching and Its Application in SLAM Based Laparoscopic Tracking 査読有り

Wang Cheng, Oda Masahiro, Hayashi Yuichiro, Misawa Kazunari, Roth Holger, Mori Kensaku

COMPUTER ASSISTED AND ROBOTIC ENDOSCOPY AND CLINICAL IMAGE-BASED PROCEDURES 10550 巻頁： 60-69 2017年

Airway extraction from 3D chest CT volumes based on iterative extension of VOI enhanced by cavity enhancement filter 査読有り

Meng Qier, Kitasaka Takayuki, Oda Masahiro, Mori Kensaku

SPIE MEDICAL IMAGING 2017: COMPUTER-AIDED DIAGNOSIS 10134 巻 2017年

Micro-CT Guided 3D Reconstruction of Histological Images 査読有り

Nagara Kai, Roth Holger R., Nakamura Shota, Oda Hirohisa, Moriya Takayasu, Oda Masahiro, Mori Kensaku

PATCH-BASED TECHNIQUES IN MEDICAL IMAGING (PATCH-MI 2017) 10530 巻頁： 93-101 2017年

Hessian-Assisted Supervoxel: Structure-Oriented Voxel Clustering and Application to Mediastinal Lymph Node Detection from CT Volumes 査読有り国際共著

Oda Hirosha, Bhatia Kanwal K., Oda Masahiro, Kitasaka Takayuki, Iwano Shingo, Homma Hirotoshi, Takabatake Hirotsugu, Mori Masaki, Natori Hiroshi, Schnabel Julia A., Mori Kensaku

MEDICAL IMAGING 2017: COMPUTER-AIDED DIAGNOSIS 10134 巻 2017年

3D FCN Feature Driven Regression Forest-Based Pancreas Localization and Segmentation 査読有り国際共著

Masahiro Oda, Natsuki Shimizu, Holger R. Roth, Ken’ichi Karasawa, Takayuki Kitasaka, Kazunari Misawa, Michitaka Fujiwara, Daniel Rueckert, Kensaku Mori

MICCAI 2017, 3rd Workshop on Deep Learning in Medical Image Analysis 10553 巻頁： 222 - 230 2017年

Extraction of membrane structure in eyeball from MR volumes 査読有り

Oda Masahiro, Kin Taichi, Mori Kensaku

SPIE MEDICAL IMAGING 2017: BIOMEDICAL APPLICATIONS IN MOLECULAR, STRUCTURAL, AND FUNCTIONAL IMAGING 10137 巻 2017年

Tracking and segmentation of the airways in chest CT using a fully convolutional network 査読有り

Qier Meng, Holger R. Roth, Takayuki Kitasaka, Masahiro Oda, Junji Ueno, Kensaku Mori

MICCAI 2017 10434 LNCS 巻頁： 198 - 207 2017年

TBS: Tensor-based supervoxels for unfolding the heart 査読有り国際共著

Hirohisa Oda, Holger R. Roth, Kanwal K. Bhatia, Masahiro Oda, Takayuki Kitasaka, Toshiaki Akita, Julia A. Schnabel, Kensaku Mori

MICCAI 2017 10433 LNCS 巻頁： 681 - 689 2017年

Comparison of the Deep-Learning-Based Automated Segmentation Methods for the Head Sectioned Images of the Virtual Korean Human Project 査読有り

Eshghi Mohammad, Roth Holger R., Oda Masahiro, Chung Min Suk, Mori Kensaku

PROCEEDINGS OF THE FIFTEENTH IAPR INTERNATIONAL CONFERENCE ON MACHINE VISION APPLICATIONS - MVA2017 頁： 290 - 293 2017年

Automatic Segmentation of Head Anatomical Structures from Sparsely-annotated Images 査読有り

Sugino Takaaki, Roth Holger R., Eshghi Mohammad, Oda Masahiro, Chung Min Suk, Mori Kensaku

2017 IEEE INTERNATIONAL CONFERENCE ON CYBORG AND BIONIC SYSTEMS (CBS) 頁： 145 - 149 2017年

前東晃礼, 三輪和久, 小田昌宏, 中村嘉彦, 森建策, 伊神剛

人工知能学会研究会資料先進的学習科学と工学研究会 79 巻 ( 0 ) 2017年

Influence of using 3D images and 3D-printed objects on spatial reasoning of experts and novices

Maehigashi A., Miwa K., Oda M., Nakamura Y., Mori K., Igami T.

CogSci 2017 - Proceedings of the 39th Annual Meeting of the Cognitive Science Society: Computational Foundations of Cognition 頁： 2669 - 2674 2017年

記述言語：日本語出版者・発行元：CogSci 2017 - Proceedings of the 39th Annual Meeting of the Cognitive Science Society: Computational Foundations of Cognition

This study focuses on the infuence of a three-dimensional (3D) graphic image and a 3D-printed object on the spatial reasoning of experts and novices in the medical field. The spatial reasoning task of this study required doctors specializing in digestive surgery to infer cross sections of a liver with a 3D image and a 3D-printed object in a situation where liver resection surgery was simulated. The task performance was compared with that of university students who conducted the same task in Maehigashi et al. (2016). The results of the analysis indicated that the doctors showed the same task performance when using the 3D image and the 3D-printed object. However, the university students learned faster and inferred the inside of a liver structure more accurately with the 3D-printed object than with the 3D image, and they performed equally to the professional doctors. Our results are then discussed in relation to previous studies.