Faculty Profiles - ODA, Masahiro

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 Reviewed Open Access

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

International Journal of Computer Assisted Radiology and Surgery 2025.4

Development of a YOLOv3-Based Model for Automated Detection of Thoracic Ossification of the Posterior Longitudinal Ligament and the Ligamentum Flavum on Plain Radiographs Reviewed 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 Vol. 14 ( 7 ) 2025.3

Bidirectional teaching between lightweight multi-view networks for intestine segmentation from CT volume Reviewed 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 Vol. 12 ( 2 ) 2025.3

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

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

International Journal of Computer Assisted Radiology and Surgery 2025.2

Multi-dimensional consistency learning between 2D Swin U-Net and 3D U-Net for intestine segmentation from CT volume Reviewed 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 2025.2

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

JAPANESE JOURNAL OF OPHTHALMOLOGY 2025.2

Language：English Publisher：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 Invited Reviewed

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 Vol. 15 2025.2

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 2025.2

Language：English Publisher：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 Reviewed

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 2025.2

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

Scientific Reports Vol. 15 ( 1462 ) 2025.1

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 Vol. 13406 2025

Publisher：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) Vol. 15371 LNCS page： 108 - 118 2025

Publisher：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) Vol. 15371 LNCS page： 3 - 13 2025

Publisher：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 Vol. 13408 2025

Publisher：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 Vol. 13408 2025

Publisher：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 Vol. 13408 2025

Publisher：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 Vol. 13410 2025

Publisher：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 Vol. 13406 2025

Publisher：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 Vol. 13406 2025

Publisher：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 Vol. 13407 2025

Publisher：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 Reviewed

Hayato Itoh, Masahiro Oda, Shinji Saiki, Koji Kamagata, Wataru Sako, Kei-ichi Ishikawa, Nobutaka Hattori, Shigeki Aoki, Kensaku Mori

International Journal of Computer Assisted Radiology and Surgery Vol. 19 ( 11 ) page： 2133 - 2142 2024.11

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 Vol. 69 ( 6 ) page： 787 - 792 2024.11

Language：Japanese Publisher：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 2024.9

Language：English Publisher：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 Vol. 98 ( 8 ) page： S1732 - S1732 2024.9

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 Vol. 14 ( 1 ) page： 15517 2024.7

Language：English Publisher：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 2024.5

Language：English Publisher：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 2024.4

Language：English Publisher：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 2024.3

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 2024.3

Use of 3D-printed model of liver by experts and novices Reviewed

Akihiro Maehigashi, Kazuhisa Miwa, Masahiro Oda, Yoshihiko Nakamura, Kensaku Mori, Tsuyoshi Igami

Current Psychology 2024.2

Publishing type：Research paper (scientific journal)

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 Reviewed

Yuzhang Liu, Yuichiro Hayashi, Masahiro Oda, Takayuki Kitasaka, Kensaku Mori

Healthcare Technology Letters 2024.1

Publishing type：Research paper (scientific journal)

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 2024.1

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

MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION - MICCAI 2024, PT VI Vol. 15006 page： 241 - 251 2024

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

Publisher：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) Vol. 14313 LNCS page： 135 - 144 2024

Publisher：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

Publisher：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

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 Vol. 12927 2024

Publisher：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 Vol. 12928 2024

Publisher：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

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 Vol. 12926 2024

Publisher：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

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 Vol. 12926 2024

Publisher：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

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) Vol. 15196 LNCS page： 32 - 41 2024

Publisher：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 Vol. 12927 2024

Publisher：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 Vol. 15006 page： 14 - 24 2024

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

Publisher：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 Vol. 15008 page： 612 - 622 2024

Publisher：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

Publisher：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. International journal

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

International journal of computer assisted radiology and surgery 2023.12

Language：English Publishing type：Research paper (scientific journal)

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 Vol. 2023 page： 8495937 2023.11

Language：English Publisher：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 Vol. 25 ( 3 ) page： 259 - 259 2023.11

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

JOURNAL OF CLINICAL MEDICINE Vol. 12 ( 15 ) 2023.8

Language：English Publisher：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 2023.7

Language：English Publisher：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 Vol. 42回 page： 153 - 154 2023.7

Language：Japanese Publisher：(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 Vol. 42回 page： 209 - 210 2023.7

Language：Japanese Publisher：(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 Vol. 97 ( 6 ) page： AB656 - AB656 2023.6

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 Vol. 18 ( 5 ) page： 795 - 805 2023.5

Language：English Publisher：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 Vol. 18 ( 5 ) page： 807 2023.5

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

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 2023.2

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 Vol. 12466 2023

Publisher：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 Vol. 12465 2023

Publisher：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

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 Vol. 12466 2023

Publisher：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 Vol. 12464 2023

Publisher：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 Vol. 12464 2023

Publisher：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

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

MEDICAL IMAGING 2023 Vol. 12464 2023

Language：English Publishing type：Research paper (international conference proceedings) Publisher：SPIE-INT SOC OPTICAL ENGINEERING

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

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 Vol. 12465 2023

Publisher：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

Oesophagus Achalasia Diagnosis from Esophagoscopy Based on a Serial Multi-scale Network Reviewed

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

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

KST-Mixer: kinematic spatio-temporal data mixer for colon shape estimation Reviewed International coauthorship

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

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

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 Vol. 12468 2023

Publisher：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) Vol. 13846 LNCS page： 185 - 202 2023

Publisher：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

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

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 Vol. Annual61 ( Abstract ) page： 127_2 - 127_2 2023

Language：Japanese Publisher：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

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) Vol. 14228 LNCS page： 429 - 439 2023

Publisher：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

ConDistFL: Conditional Distillation for Federated Learning from Partially Annotated Data

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) Vol. 14393 page： 311 - 321 2023

Publisher：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) Vol. 14220 LNCS page： 663 - 673 2023

Publisher：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

Anatomy aware-based 2.5D bronchoscope tracking for image-guided bronchoscopic navigation Reviewed

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 Vol. Annual61 ( Abstract ) page： 255_2 - 255_2 2023

Language：Japanese Publisher：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 Reviewed

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

COMPUTER METHODS IN BIOMECHANICS AND BIOMEDICAL ENGINEERING-IMAGING AND VISUALIZATION 2022.12

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 2022.12

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

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

Scientific reports Vol. 12 ( 1 ) page： 20840 2022.12

Participating Report of MICCAI 2022 Invited

Oda Masahiro

Medical Imaging and Information Sciences Vol. 39 ( 4 ) page： 78 - 81 2022.12

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 Invited

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 Vol. 57 ( 10 ) page： 1189 - 1195 2022.10

A skeleton context-aware 3D fully convolutional network for abdominal artery segmentation. Reviewed

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

International journal of computer assisted radiology and surgery 2022.10

Class-wise confidence-aware active learning for laparoscopic images segmentation. Reviewed

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

International journal of computer assisted radiology and surgery 2022.10

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

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) Vol. 13594 LNCS page： 63 - 75 2022.10

Joint Multi Organ and Tumor Segmentation from Partial Labels Using Federated Learning Reviewed International coauthorship

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) Vol. 13573 LNCS page： 58 - 67 2022.10

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 Reviewed

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 Vol. 33 ( 10 ) page： 760 - 760 2022.10

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 Invited

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 Vol. 57 ( 10 ) page： 1189 - 1195 2022.10

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

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

MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION, MICCAI 2022, PT IV Vol. 13434 page： 467 - 477 2022.9

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

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

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY 2022.8

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 Vol. 41回 page： 124 - 125 2022.7

Language：Japanese Publishing type：Research paper (conference, symposium, etc.) Publisher：(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 Reviewed

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 Vol. 163 ( 1 ) page： 323 - 325.e3 2022.7

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

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) Vol. 2022-June page： 1863 - 1872 2022.6

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 Vol. 81 page： 1773 - 1774 2022.6

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

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 Vol. 115 ( 6 ) page： 890 - 897 2022.6

Publisher：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

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

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

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

Journal of Medical Imaging Vol. 9 ( 2 ) page： 024003 2022.3

A cascaded fully convolutional network framework for dilated pancreatic duct segmentation Reviewed International coauthorship

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

International Journal of Computer Assisted Radiology and Surgery Vol. 17 ( 2 ) page： 343 - 354 2022.2

Real-time Esophagus Achalasia Detection Method for Esophagoscopy Assistance Reviewed

MEDICAL IMAGING 2022: COMPUTER-AIDED DIAGNOSIS Vol. 12033 2022.2

Uncertainty meets 3D-spatial feature in colonoscopic polyp-size determination Reviewed

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 2022.1

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 Vol. 95 ( 1 ) page： 155 - 163 2022.1

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

Tao Hu, Hayato Itoh, Masahiro Oda, Yuichiro Hayashi, Zhongyang Lu, Shinji Saiki, Nobutaka Hattori, Koji Kamagata, Shigeki Aoki, Kanako K. Kumamaru, Toshiaki Akashi, Kensaku Mori

MICCAI (8) Vol. 13437 LNCS page： 736 - 744 2022

Spatially variant biases considered self-supervised depth estimation based on laparoscopic videos Reviewed

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

Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization 2021.12

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

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

Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization 2021.12

Attention-Guided Pancreatic Duct Segmentation from Abdominal CT Volumes Reviewed International coauthorship

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) Vol. 12969 LNCS page： 46 - 55 2021.11

Multi-task Federated Learning for Heterogeneous Pancreas Segmentation Reviewed International coauthorship

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) Vol. 12969 LNCS page： 101 - 110 2021.11

Intestine Segmentation with Small Computational Cost for Diagnosis Assistance of Ileus and Intestinal Obstruction Reviewed

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) Vol. 12969 LNCS page： 3 - 12 2021.11

COVID-19 Infection Segmentation from Chest CT Images Based on Scale Uncertainty Reviewed

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) Vol. 12969 LNCS page： 88 - 97 2021.11

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

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

International Journal of Computer Assisted Radiology and Surgery Vol. 17 ( 1 ) page： 97 - 105 2021.10

Graph Cuts Loss to Boost Model Accuracy and Generalizability for Medical Image Segmentation Reviewed

Zheng Zhou, Oda Masahiro, Mori Kensaku

2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION WORKSHOPS (ICCVW 2021) Vol. 2021-October page： 3297 - 3306 2021.10

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

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) Vol. 12965 LNCS page： 24 - 33 2021.9

Performance improvement of weakly supervised fully convolutional networks by skip connections for brain structure segmentation Reviewed

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

Medical Physics Vol. 48 ( 11 ) page： 7215 - 7227 2021.9

Binary polyp-size classification based on deep-learned spatial information Reviewed

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 Surgery Vol. 16 page： 1817 - 1828 2021.9

Depth-based branching level estimation for bronchoscopic navigation Reviewed

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 Vol. 16 ( 10 ) page： 1795 - 1804 2021.8

Deep learning-based medical image processing and their applications beyond image processing Invited

Oda Masahiro

Transactions of Japanese Society for Medical and Biological Engineering Vol. 59 ( 0 ) page： 256 - 256 2021.6

Authorship：Lead author Language：Japanese Publishing type：Research paper (conference, symposium, etc.) Publisher：Japanese Society for Medical and Biological Engineering

Deep learning became an essential method of medical image processing. Deep learning solved many tasks among medical image processing, including image classification, detection, and segmentation. Many deep learning applications were proposed for eye-fundus images and radiological images among medical image modalities. Practical performances were achieved from such applications. In this presentation, we review deep learning-based medical image processing methods and their applications from their methods and image modalities. Also, we introduce new applications of deep learning in medical image processing. Deep learning-based methods process many kinds of data as their explanatory and objective variables. They can be applied to medical data, including image, text, and sensor data. We introduce some examples of such applications. We review new deep learning applications in medical image processing from the examples.

DOI： 10.11239/jsmbe.Annual59.256

Artificial intelligence-assisted colonic endocytoscopy for cancer recognition: a multicenter study Reviewed

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 page： E1004 - E1011 2021.6

Unsupervised colonoscopic depth estimation by domain translations with a Lambertian-reflection keeping auxiliary task Reviewed

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 Surgery Vol. 16 ( 6 ) page： 989 - 1001 2021.6

Deep learning system for automatic detection of bladder tumors in cystoscopic images Reviewed

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 Vol. 79 page： S1022 - S1023 2021.6

COVID-19診断支援AI開発における名古屋大学の取り組み Invited

小田昌宏, 鄭通, 林雄一郎, 森健策

Medical Imaging Technology Vol. 39 ( 1 ) page： 13 - 19 2021.2

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

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 Vol. 11597 2021.2

Automated Detection of Spinal Schwannomas Utilizing Deep Learning Based on Object Detection from MRI Reviewed

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 Vol. 46 ( 2 ) page： 95 - 100 2021.1

Current status and future perspective on artificial intelligence for lower endoscopy Reviewed

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 Vol. 33 ( 2 ) page： 273 - 284 2021.1

医療支援AIの現状と情報横断型AIへの期待 Invited

小田昌宏

日本整形外科学会雑誌 Vol. 95 ( 1 ) page： 3 - 8 2021.1

CURRENT STATUS AND FUTURE PERSPECTIVE ON ARTIFICIAL INTELLIGENCE FOR LOWER ENDOSCOPY Reviewed

MISAWA Masashi, MIYACHI Hideyuki, BABA Toshiyuki, ISHIDA Fumio, ITOH Hayato, ODA Masahiro, MORI Kensaku, KUDO Shin-ei, MORI Yuichi, MAEDA Yasuharu, OGAWA Yushi, ICHIMASA Katsuro, KUDO Toyoki, WAKAMURA Kunihiko, HAYASHI Takemasa

GASTROENTEROLOGICAL ENDOSCOPY Vol. 63 ( 7 ) page： 1402 - 1416 2021.1

Language：English Publishing type：Research paper (scientific journal) Publisher：Japan Gastroenterological Endoscopy Society

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

Object Detection method by deep learning for diagnosing bladder tumors in cystoscopic images

Report on the First JSCAS AI Challenge Invited

Journal of Japan Society of Computer Aided Surgery Vol. 23 ( 2 ) page： 74 - 75 2021

Mutaguchi Jun, Oda Masahiro, Kobayashi Satoshi, Inokuchi Junichi, Mori Kensaku, Eto Masatoshi

Transactions of Japanese Society for Medical and Biological Engineering Vol. 59 ( 0 ) page： 299 - 299 2021

Language：Japanese Publisher：Japanese Society for Medical and Biological Engineering

The overlooking bladder tumors during transurethral resection of bladder tumor (TURBT) causes high intravesical recurrence rate. The conventional White light imaging(WLI) and Narrow-band imaging (NBI) were subjective and poor reproducibility due to depending on doctor's experience and skills. Recently, image recognition using artificial intelligence (AI) has been applied for diagnostic imaging and achieving a good performance result, however the feasibility of cystoscopy with AI is still unknown. In this study, we constructed an object detection system based on the AI architecture. We prospectively obtained cystoscopic images from patients who underwent TURBT, and divided these images as the training data and test data. The sensitivity, specificity and Positive predictive value of WLI and NBI were 84.6%, 76.9% and 90.8%, and 78.1%, 97.1% and 95.8%, respectively. Our system has the possibility to improve detection rates of the bladder tumors during cystoscopy, and might be beneficial to reduce recurrence rate of bladder tumors.

DOI： 10.11239/jsmbe.Annual59.299

An Introduction to Deep Learing in Image Recognition（2）: 　Configuration of GPU Environments and Application to Image Segmentation Invited

HARA Takeshi, ODA Masahiro

Medical Imaging Technology Vol. 39 ( 3 ) page： 124 - 130 2021

Language：Japanese Publishing type：Research paper (scientific journal) Publisher：The Japanese Society of Medical Imaging Technology

In order to set up a practical experiment and research environment for deep learning, it is necessary to build an computing environment using GPU. This is because the calculation speed is several tens of times faster than the calculation using only the CPU. In this course, we will explain how to build a GPU environment and clarify how much the calculation time changes, using image segmentation as an example. At the same time, the method of image segmentation using U-Net model and the construction of the label data are also mentioned.

DOI： 10.11409/mit.39.124

Towards Explainable Artificial Intelligence: Introduction Invited

FUJIYOSHI Hironobu, ODA Masahiro

Medical Imaging Technology Vol. 39 ( 3 ) page： 97 - 98 2021

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

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 Vol. 15 ( 12 ) page： 2049 - 2059 2020.12

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

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 2020.11

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 Vol. 22 ( 4 ) page： 348 - 349 2020.11

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 Vol. 22 ( 4 ) page： 346 - 347 2020.11

【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 Vol. 38 ( 5 ) page： 213 - 216 2020.11

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 Vol. 22 ( 4 ) page： 306 - 307 2020.11

Artificial Intelligence System to Determine Risk of T1 Colorectal Cancer Metastasis to Lymph Node. Reviewed

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 2020.9

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

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

MEDICAL PHYSICS Vol. 47 ( 9 ) page： 3882 - 3891 2020.9

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 Vol. 39回 page： 29 - 29 2020.9

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 Vol. 39回 page： 569 - 572 2020.9

Language：Japanese Publisher：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). Reviewed

A visual SLAM-based bronchoscope tracking scheme for bronchoscopic navigation Reviewed

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 Vol. 15 ( 10 ) page： 1619 - 1630 2020.8

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 2020.7

A deformable model for navigated laparoscopic gastrectomy based on finite elemental method Reviewed

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 Vol. 29 ( 4 ) page： 210 - 216 2020.7

Artificial Intelligence-assisted System Improves Endoscopic Identification of Colorectal Neoplasms Reviewed

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 Vol. 18 ( 8 ) page： 1874 - + 2020.7

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

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 Vol. 91 ( 6 ) page： AB43 - AB43 2020.6

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 Vol. 36 ( 2 ) page： 77 - 82 2020.5

Language：Japanese Publisher：(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 Vol. 36 ( 2 ) page： 125 - 125 2020.5

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

CoRR Vol. abs/2004.03272 page： 454 - 459 2020.4

Language：English Publishing type：Research paper (scientific journal) Publisher：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

How Far Will Clinical Application of AI Applications Advance for Colorectal Cancer Diagnosis? Reviewed

Other Link： http://arxiv.org/pdf/2004.03272v1

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

JOURNAL OF THE ANUS RECTUM AND COLON Vol. 4 ( 2 ) page： 47 - 50 2020.4

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

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

TECHNIQUES AND INNOVATIONS IN GASTROINTESTINAL ENDOSCOPY Vol. 22 ( 2 ) page： 56 - 60 2020.4

Language：Japanese Publisher：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 Reviewed

Cardiac fiber tracking on super high-resolution CT images: a comparative study Reviewed

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 Vol. 7 ( 2 ) page： 026001 2020.3

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 Vol. 76回 page： 336 - 337 2020.3

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

MEDICAL IMAGE ANALYSIS Vol. 60 page： 101623 2020.2

Micro-computed tomography images of lung adenocarcinoma: detection of lepidic growth patterns Reviewed

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

NAGOYA JOURNAL OF MEDICAL SCIENCE Vol. 82 ( 1 ) page： 25 - 31 2020.2

Organ Segmentation From Full-size CT Images Using Memory-Efficient FCN Reviewed

Wang Chenglong, Oda Masahiro, Mori Kensaku

MEDICAL IMAGING 2020: COMPUTER-AIDED DIAGNOSIS Vol. 11314 2020

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

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 Vol. 11314 2020

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

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

MEDICAL IMAGING 2020: COMPUTER-AIDED DIAGNOSIS Vol. 11314 2020

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

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 Vol. 11314 2020

Deep Learning in Surgical Assistance Invited

Oda Masahiro

Journal of Japan Society of Computer Aided Surgery Vol. 22 ( 1 ) page： 54-58 2020

Clinical Application of EndoBRAIN, a Medical Device Adopting Artificial Intelligence: Challenges to Obtain Regulatory Approval and Insurance Reimbursement Invited

MORI Yuichi, KUDO Shin-ei, MISAWA Masashi, ITOH Hayato, ODA Masahiro, MORI Kensaku

Medical Imaging Technology Vol. 38 ( 5 ) page： 213-216 2020

Language：English Publishing type：Research paper (scientific journal)

EndoBRAIN (Cybernet System Corp.; Olympus Corp., Tokyo) is the computer-aided diagnostic system supportedby artificial intelligence (AI) that allows real-time identification of colorectal polyps. EndoBRAIN was the firstly approved medical device in the field of AI in Japan. This regulatory approval was achieved by the close collaboration among medical researchers, engineering researchers, industrial partners, and public funding bodies. This review article shows the overview of EndoBRAIN including its design, performance, and results in clinical trials. In addition, our approaches to regulatory approval and insurance reimbursement, which unfortunately has not been granted, are also introduced.

DOI： 10.11409/mit.38.213

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

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

Progress in Biomedical Optics and Imaging - Proceedings of SPIE Vol. 11313 2020

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

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 Vol. 11313 2020

Improved visual SLAM for bronchoscope tracking and registration with pre-operative CT images Reviewed

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 Vol. 11315 2020

Automated eye disease classification method from anterior eye image using anatomical structure focused image classification technique Reviewed

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

MEDICAL IMAGING 2020: COMPUTER-AIDED DIAGNOSIS Vol. 11314 2020

Abdominal artery segmentation method from CT volumes using fully convolutional neural network Reviewed

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

International Journal of Computer Assisted Radiology and Surgery Vol. 14 ( 12 ) page： 2069 - 2081 2019.12

Realistic Endoscopic Image Generation Method Using Virtual-to-real Image-domain Translation Reviewed

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

Healthcare Technology Letters Vol. 6 ( 6 ) page： 214 - 219 2019.12

Stable polyp-scene classification via subsampling and residual learning from an imbalanced large dataset Reviewed

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

HEALTHCARE TECHNOLOGY LETTERS Vol. 6 ( 6 ) page： 237 - 242 2019.12

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 2019.11

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 Vol. 21 ( 4 ) page： 332 - 333 2019.11

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 Reviewed

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 Vol. 21 ( 4 ) page： 318 - 319 2019.11

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 Vol. 21 ( 4 ) page： 232 - 233 2019.11

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 Vol. 21 ( 4 ) page： 229 - 230 2019.11

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 Vol. 21 ( 4 ) page： 226 - 227 2019.11

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 Vol. 21 ( 4 ) page： 321 - 321 2019.11

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 Vol. 35 ( 10 ) page： 1120 - 1128 2019.10

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

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 Vol. 77 page： 101642 2019.10

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 Vol. 61 ( Suppl.2 ) page： 2167 - 2167 2019.10

A view of three dimensional unit structures of alveoli in peripheral lung Reviewed

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

EUROPEAN RESPIRATORY JOURNAL Vol. 54 2019.9

Artificial intelligence and colonoscopy: Current status and future perspectives Reviewed

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

DIGESTIVE ENDOSCOPY Vol. 31 ( 4 ) page： 363 - 371 2019.7

Artificial intelligence and upper gastrointestinal endoscopy: Current status and future perspective Reviewed

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

DIGESTIVE ENDOSCOPY Vol. 31 ( 4 ) page： 378 - 388 2019.7

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 page： OP5-23 2019.7

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 page： OP5-11 2019.7

Polyp size classification in colorectal cancer using a Siamese network

page： OP2-14 2019.7

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 page： SY1-5 2019.7

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 page： OP3-20 2019.7

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 page： OP2-08 2019.7

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 page： OP3-12 2019.7

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 page： OP4-15 2019.7

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 page： OP5-14 2019.7

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 page： OP4-13 2019.7

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 page： OP1-24 2019.7

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 page： OP1-15 2019.7

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

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 Vol. 89 ( 6 ) page： AB646 - AB647 2019.6

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

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 Vol. 89 ( 6 ) page： AB89 - AB89 2019.6

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

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

GASTROINTESTINAL ENDOSCOPY Vol. 89 ( 6 ) page： AB76 - AB76 2019.6

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

GASTROINTESTINAL ENDOSCOPY Vol. 89 ( 6 ) page： AB646-AB647 2019.6

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 Vol. 14 ( 1 ) page： S24-25 2019.6

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) Vol. 119 ( 51 ) page： 65-70 2019.5

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

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

MICROMACHINES Vol. 10 ( 2 ) 2019.2

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

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

GASTROINTESTINAL ENDOSCOPY Vol. 89 ( 2 ) page： 408 - 415 2019.2

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

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

Gastrointestinal endoscopy Vol. 89 ( 2 ) page： 408-415 2019.2

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 page： 109490B-1-7 2019.2

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) Vol. 118 ( 412 ) page： 143-148 2019.1

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 page： 0 2019.1

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) Vol. 118 ( 412 ) page： 65-70 2019.1

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) Vol. 118 ( 412 ) page： 97-102 2019.1

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) Vol. 118 ( 412 ) page： 191-196 2019.1

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) Vol. 118 ( 412 ) page： 7-12 2019.1

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 page： 82 2019.1

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) Vol. 118 ( 412 ) page： 109-114 2019.1

Automated Hand Eye Calibration in Laparoscope Holding Robot for Robot Assisted Surgery Reviewed

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

INTERNATIONAL WORKSHOP ON ADVANCED IMAGE TECHNOLOGY (IWAIT) 2019 Vol. 11049 2019

Applications of Generative Adversarial Networks in Medical Image Processing: Introduction Invited

ODA Masahiro

Medical Imaging Technology Vol. 37 ( 3 ) page： 123-124 2019

Artificial Intelligence/Deep Learning for Surgical Assistance Invited

Oda Masahiro

Journal of Japan Society of Computer Aided Surgery Vol. 21 ( 3 ) page： 143-146 2019

Radiological and Endoscope Image Analyses in Nagoya University Invited

ODA Masahiro, SHEN Chen, ODA Hirohisa, MORI Kensaku

Medical Imaging Technology Vol. 37 ( 2 ) page： 84-88 2019

Authorship：Lead author Language：English Publishing type：Research paper (conference, symposium, etc.)

This paper introduces our medical image analysis researches in the AMED project. Our research topics are radiological and endoscope image analyses. The radiological image analysis includes a blood vessel diagnosis assistance using non-contrasted abdominal CT volumes and an abdominal multi-organ segmentation from CT volumes. The endoscope image analysis includes observing area classifications of endoscope images taken in the stomach and colon. We explain approaches and results of these researches.

DOI： 10.11409/mit.37.84

Micro-CT in the Analysis of Formalin-Fixed Paraffin-Embedded Blocks of Resected Pancreatic Lesions Reviewed

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

PANCREAS Vol. 48 ( 10 ) page： 1523 - 1523 2019

Polyp-Size Classification with RGB-D features for Colonoscopy Reviewed

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

MEDICAL IMAGING 2019: COMPUTER-AIDED DIAGNOSIS Vol. 10950 2019

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

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 Vol. 10953 2019

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

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 Vol. 11840 page： 104-114 2019

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

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 Vol. 11769 page： 348-356 2019

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

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

MEDICAL IMAGING 2019: BIOMEDICAL APPLICATIONS IN MOLECULAR, STRUCTURAL, AND FUNCTIONAL IMAGING Vol. 10953 2019

Unsupervised Segmentation of Micro-CT Images of Lung Cancer Specimen Using Deep Generative Models Reviewed

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 Vol. 11769 page： 240-248 2019

Visual SLAM for bronchoscope tracking and bronchus reconstruction in bronchoscopic navigation Reviewed

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 Vol. 10951 2019

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 Vol. 10950 2019

Discriminative Feature Selection by Optimal Manifold Search for Neoplastic Image Recognition Reviewed

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

COMPUTER VISION - ECCV 2018 WORKSHOPS, PT IV Vol. 11132 page： 534-549 2019

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

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

INTERNATIONAL FORUM ON MEDICAL IMAGING IN ASIA 2019 Vol. 11050 2019

Improving V-Nets for multi-class abdominal organ segmentation Reviewed

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

MEDICAL IMAGING 2019: IMAGE PROCESSING Vol. 10949 2019

ディープラーニング実践の環境構築

小田昌宏

臨床画像 Vol. 35 page： 1120 - 1128 2019

Development of Colonoscope Tracking Method using Recurrent Neural Network for Colonoscopic Examination Navigation Reviewed

Masahiro Oda, Holger R. Roth, Takayuki Kitasaka, Kazuhiro Furukawa, Ryoji Miyahara, Yoshiki Hirooka, Nassir Navab, Kensaku Mori

Transactions of the Virtual Reality Society of Japan Vol. 23 ( 4 ) page： 249-252 2018.12

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 Vol. 20 ( 4 18(Ⅶ)-4 ) page： 322-323 2018.11

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 Vol. 20 ( 4 18(Ⅹ)-3 ) page： 348-349 2018.11

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 Vol. 20 ( 4 18(6)-10 ) page： 370 2018.11

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 Vol. 20 ( 4 18(Ⅸ)-4 ) page： 342-343 2018.11

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 page： CH218-ED-X 2018.11

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

Vol. 118 ( 286 ) page： 27-32 2018.11

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 Vol. 20 ( 4 18(ⅩⅠ)-10 ) page： 368-369 2018.11

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 Vol. 20 ( 4 18(Ⅹ)-10 ) page： 359 2018.11

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 Vol. 20 ( 4 18(Ⅵ)-9 ) page： 312-313 2018.11

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 Vol. 20 ( 4 18(Ⅲ)-4 ) page： 259-260 2018.11

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 Vol. 20 ( 4 18(Ⅱ)-8 ) page： 252-253 2018.11

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 Vol. 20 ( 4 18(Ⅲ)-5 ) page： 261-262 2018.11

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

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 Vol. 169 ( 6 ) page： 357 - 366 2018.9

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

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

EUROPEAN RESPIRATORY JOURNAL Vol. 52 2018.9

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

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 Vol. 88 ( 9 ) page： 882 - 885 2018.9

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 page： PS3-11 2018.8

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 page： PS2-17 2018.8

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 page： OP1-6 2018.7

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 page： OP1-8 2018.7

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) Vol. 118 ( 150 ) page： 11-16 2018.7

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 page： OP13-6 2018.7

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 page： OP4-2 2018.7

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 page： OP14-2 2018.7

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 page： OP1-7 2018.7

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 page： OP14-4 2018.7

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 Vol. 13 ( 1 ) page： s140-142 2018.6

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 Vol. 13 ( 1 ) page： s97-98 2018.6

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 Vol. 13 ( 1 ) page： s104-106 2018.6

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 Vol. 13 ( 1 ) page： s11-12 2018.6

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 Vol. 13 ( 1 ) page： s21-22 2018.6

Artificial Intelligence-Assisted Polyp Detection for Colonoscopy: Initial Experience Reviewed

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 Vol. 154 ( 8 ) page： 2027 - + 2018.6

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

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 Vol. 87 ( 6 ) page： AB304-AB304 2018.6

An application of cascaded 3D fully convolutional networks for medical image segmentation Reviewed

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 Vol. 66 page： 90 - 99 2018.6

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

GASTROINTESTINAL ENDOSCOPY Vol. 87 ( 6 ) page： AB304-AB304 2018.6

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) Vol. 117 ( 518 ) page： 75-80 2018.3

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) Vol. 117 ( 518 ) page： 7-12 2018.3

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

page： 87-105 2018.3

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) Vol. 117 ( 518 ) page： 81-86 2018.3

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) Vol. 117 ( 518 ) page： 117-122 2018.3

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) Vol. 117 ( 518 ) page： 51-56 2018.3

Develop and Validate a Finite Element Method Model for Deformation Matching of Laparoscopic Gastrectomy Navigation Reviewed

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 Vol. 10576 2018

Micro Anatomical Structure Imaging by Desktop Micro-focus X-ray CT Scanner Invited

MORI Kensaku, NAKAMURA Shota, AKITA Toshiaki, ODA Hirohisa, Holger ROTH R., ODA Masahiro

Medical Imaging Technology Vol. 36 ( 3 ) page： 127-132 2018

Language：Japanese Publishing type：Research paper (conference, symposium, etc.)

This paper shows examples of micro anatomical structure imaging using a desktop micro-focus X-ray CT scanner. Currently clinical X-ray CT scanners can take CT images in 0.5mm to 1mm per voxel in resolution. Such scanner can take anatomical structures close to image resolution of scanners. Micro-focus X-ray CT (micro CT) scanner can take CT images of 1μm per voxel to 50μm per voxel in resolution. Such scanner enables us to explore micro anatomical structures. This paper shows several examples of micro CT images of lung and heart specimens with discussion on future direction.

DOI： 10.11409/mit.36.127

Guidance of Deep Learning for Medical Image Processing: Introduction Invited

ODA Masahiro

Medical Imaging Technology Vol. 36 ( 2 ) page： 45 - 46 2018

Towards Automated Colonoscopy Diagnosis: Binary Polyp Size Estimation via Unsupervised Depth Learning Reviewed

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 Vol. 11071 page： 611 - 619 2018

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 Vol. 10574 2018

Unsupervised Pathology Image Segmentation Using Representation Learning with Spherical K-means Reviewed

Moriya Takayasu, Roth Holger H., Nakamura Shota, Oda Hirohisa, Nagara Kai, Oda Masahiro, Mori Kensaku

MEDICAL IMAGING 2018: DIGITAL PATHOLOGY Vol. 10581 2018

Unsupervised Segmentation of 3D Medical Images Based on Clustering and Deep Representation Learning Reviewed

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 Vol. 10578 2018

Important Points to Take Advantage of Deep Learning Invited

ODA Masahiro

Medical Imaging Technology Vol. 36 ( 2 ) page： 72-75 2018

Authorship：Lead author Language：Japanese Publishing type：Research paper (conference, symposium, etc.)

Deep learning can be applied to many tasks in medical image processing. There are some points to be careful in research use of deep learning to take advantage of its performance. Among such points, three topics including manual specification of many parameters in deep learning, pitfalls of data augmentation, and how randomly selected parameters in deep learning affect its performance, are discussed here.

DOI： 10.11409/mit.36.72

A Multi-scale Pyramid of 3D Fully Convolutional Networks for Abdominal Multi-organ Segmentation Reviewed

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 Vol. 11073 page： 417 - 425 2018

Fully Convolutional Network-Based Eyeball Segmentation from Sparse Annotation for Eye Surgery Simulation Model Reviewed

Sugino Takaaki, Roth Holger R., Oda Masahiro, Mori Kensaku

International Workshop on Bio-Imaging and Visualization for Patient-Customized Simulations, BIVPCS 2018 Vol. 11042 page： 118-126 2018

Fine Segmentation of Tiny Blood Vessel Based on Fully-Connected Conditional Random Field Reviewed

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

MEDICAL IMAGING 2018: IMAGE PROCESSING Vol. 10574 2018

Dense volumetric detection and segmentation of mediastinal lymph nodes in chest CT images Reviewed

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 Vol. 10575 2018

Deep Learning and Its Application to Medical Image Segmentation Invited

ROTH Holger R., SHEN Chen, ODA Hirohisa, ODA Masahiro, HAYASHI Yuichiro, MISAWA Kazunari, MORI Kensaku

Medical Imaging Technology Vol. 36 ( 2 ) page： 63-71 2018

Language：English Publishing type：Research paper (conference, symposium, etc.)

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 Reviewed

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

MEDICAL IMAGING 2018: COMPUTER-AIDED DIAGNOSIS Vol. 10575 2018

BESNet: Boundary-Enhanced Segmentation of Cells in Histopathological Images Reviewed

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 Vol. 11071 page： 228-236 2018

Automatic segmentation of eyeball structures from micro-CT images based on sparse annotation Reviewed

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 Vol. 10578 2018

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 Vol. 36 ( 3 ) page： 133-146 2018

Language：English Publishing type：Research paper (scientific journal)

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

Machine learning-based colon deformation estimation method for colonoscope tracking Reviewed

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 Vol. 10576 2018

Development of a CNC abrasive machining technique for 3D-printed organ models by utilizing a flexible mechanical structure

OHBA Takayuki, SUZUKI Norikazu, SHAMOTO Eiji, ODA Masahiro, MORI Kensaku, TAKEUCHI Yoshimi

The Proceedings of The Manufacturing & Machine Tool Conference Vol. 2018.12 ( 0 ) page： C20 2018

Language：Japanese Publisher：The Japan Society of Mechanical Engineers

This paper presents a novel surface finishing technology for free-formed surfaces of three-dimensional (3D) printed resin organ models. Demand for the medical applications of three-dimensional layered fabrication technology such as organ models is increasing. Surface integrity deteriorations due to laminated micro steps formed on the surface is a significant problem. In general, material removal of layered micro step is performed manually, hence automation technology is desired. In the present study, we proposed a quasi-constant machining force grinding / polishing by use of 3-axis CNC machine tools. The workpiece is fixed to the flexible mechanical structure, whose translational flexibilities are designed to be identical regardless of contact directions and low spring rigidity. With the developed machining system, the liver model processing was demonstrated. Experimental results verified the feasibility of highly-efficient abrasive machining.

DOI： 10.1299/jsmemmt.2018.12.c20

CiNii Research

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 Vol. 117 ( 281 ) page： 15-20 2017.11

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 Vol. 117 ( 281 ) page： 47-52 2017.11

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 page： 267 2017.11

Automated mediastinal lymph node detection from CT volumes based on intensity targeted radial structure tensor analysis Reviewed International coauthorship

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 Vol. 4 ( 4 ) page： 044502 2017.10

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 Vol. 19 ( 4 ) page： 268-269 2017.10

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 Vol. 19 ( 4 ) page： 239-240 2017.10

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 Vol. 19 ( 4 ) page： 363-364 2017.10

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 Vol. 19 ( 4 ) page： 244-245 2017.10

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 Vol. 19 ( 4 ) page： 319-320 2017.10

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 Vol. 19 ( 4 ) page： 326-327 2017.10

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 Vol. 19 ( 4 ) page： 270-271 2017.10

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 Vol. 19 ( 4 ) page： 241-242 2017.10

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 Vol. 19 ( 4 ) page： 324-325 2017.10

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 Vol. 19 ( 4 ) page： 361-362 2017.10

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 Vol. 19 ( 4 ) page： 266-267 2017.10

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 page： Session 439 2017.9

Accuracy of diagnosing invasive colorectal cancer using computer-aided endocytoscopy Reviewed

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 Vol. 49 ( 8 ) page： 798 - 802 2017.8

Multi-atlas pancreas segmentation: Atlas selection based on vessel structure Reviewed International coauthorship

Ken'ichi Karasawa, Masahiro Oda, Takayuki Kitasaka, Kazunari Misawa, Michitaka Fujiwara, Chengwen Chu, Guoyan Zheng, Daniel Rueckert, Kensaku Mori

MEDICAL IMAGE ANALYSIS Vol. 39 page： 18 - 28 2017.7

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) Reviewed

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 Vol. 36回 page： 546 - 548 2017.7

Language：English Publisher：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)

Automatic anatomical labeling of arteries and veins using conditional random fields Reviewed

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 page： OP1-6 2017.7

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 page： OP2-8 2017.7

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

page： OP1-8 2017.7

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 page： OP16-5 2017.7

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 page： OP2-6 2017.7

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 page： OP14-1 2017.7

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 page： OP15-4 2017.7

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 page： OP6-1 2017.7

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 page： OP16-4 2017.7

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 Vol. 12 page： S201-S203 2017.6

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 Vol. 12 page： S55-S57 2017.6

Takayuki Kitasaka, Mitsuru Kagajo, Yukitaka Nimura, Yuichiro Hayashi, Masahiro Oda, Kazunari Misawa, Kensaku Mori

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY Vol. 12 ( 6 ) page： 1041 - 1048 2017.6

Accuracy of computer-aided diagnosis based on narrow-band imaging endocytoscopy for diagnosing colorectal lesions: comparison with experts Reviewed

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 Vol. 12 ( 5 ) page： 757-766 2017.6

Diagnostic Ability of Automated Diagnosis System Using Endocytoscopy for Invasive Colorectal Cancer Reviewed

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 Vol. 85 ( 5 ) page： AB408-AB408 2017.5

Artificial Intelligence for Endocytoscopy Provides Fully Automated Diagnosis of Histological Remission in Ulcerativ E. Coli Tis Reviewed

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 Vol. 85 ( 5 ) page： AB248-AB248 2017.5

Computer-Aided Diagnosis Based on Endocytoscopy With Narrow-Band Imaging Allows Accurate Diagnosis of Diminutive Colorectal Lesions Reviewed

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 Vol. 85 ( 5 ) page： AB57-AB57 2017.5

Can Artificial Intelligence Correctly Diagnose Sessile Serrated Adenomas/Polyps? Reviewed

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 Vol. 85 ( 5 ) page： AB510-AB510 2017.5

Diagnostic Ability of Automated Diagnosis System Using Endocytoscopy for Invasive Colorectal Cancer Reviewed

GASTROINTESTINAL ENDOSCOPY Vol. 85 ( 5 ) page： AB408-AB408 2017.5

Automatic segmentation of airway tree based on local intensity filter and machine learning technique in 3D chest CT volume Reviewed

Qier Meng, Takayuki Kitasaka, Yukitaka Nimura, Masahiro Oda, Junji Ueno, Kensaku Mori

INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY Vol. 12 ( 2 ) page： 245 - 261 2017.2

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) page： 266-268 2017.1

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) page： 54-56 2017.1

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 Vol. 116 ( 393 ) page： 191-194 2017.1

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) Vol. 116 ( 393 ) page： 23-28 2017.1

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) Vol. 116 ( 393 ) page： 133-136 2017.1

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 Vol. 116 ( 393 ) page： 157-162 2017.1

Robust colonoscope tracking method for colon deformations utilizing coarse-to-fine correspondence findings Reviewed International coauthorship

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 Vol. 12 ( 1 ) page： 39-50 2017.1

Extracellular matrix directions estimation of the heart on micro-focus X-ray CT volumes Reviewed

Oda Hirohisa, Oda Masahiro, Kitasaka Takayuki, Akita Toshiaki, Mori Kensaku

MEDICAL IMAGING 2017: BIOMEDICAL APPLICATIONS IN MOLECULAR, STRUCTURAL, AND FUNCTIONAL IMAGING Vol. 10137 2017

Evaluation of a cardiac model using different 3D printers and materials Reviewed

Murata Masato, Hirayama Koudai, Takenouchi Sinsaku, Oda Masahiro, Mori Kensaku, Niki Kiyomi

Transactions of Japanese Society for Medical and Biological Engineering Vol. 55 ( 4 ) page： 273-273 2017

Language：Japanese Publishing type：Research paper (scientific journal)

[Purpose] Models of the diseased heart produced by 3D printers are expected to provide useful information for surgical procedures. However, the precision of commercially available machines is uncertain. Therefore, we assessed the 3D models by two different methods. [Method] Using DICOM data derived from x-ray CT images of a heart, 3D models were produced by a fused deposition modeling method (FDM) and an ink-jet method. Polylactic resin (PLA) and silicone rubber (AR-G1L) were used as materials for the FDM and ink-jet methods, respectively. [Result] The model created using the FDM was produced at a reduced size due to limitations of the machine, which caused the coronary artery to collapse. It was also difficult to cut. The model created using the ink-jet method was produced at full scale and quite flexible. [Conclusion] The heart model created using the ink-jet method with silicone rubber provides excellent information for surgical treatment.

DOI： 10.11239/jsmbe.55Annual.273

Automated endoscopic image analysis based on machine learning

Mori Kensaku, Oda Masahiro, Misawa Masashi, Mori Yuichi, Kudo Shinei

Transactions of Japanese Society for Medical and Biological Engineering Vol. 55 ( 4 ) page： 344-344 2017

Language：Japanese Publishing type：Research paper (conference, symposium, etc.)

This presentation introduces automated endoscopic image diagnosis method based on machine learning. Especially we discuss effectiveness of machine learning in automated diagnosis of endoscopic images by showing automated pathological type diagnosis of colonic polyps from endocytoscopy images as examples. Machine learning research staring from perceptron or statistical pattern recognition has started in many years ago. Progress of high performance computing has enabled us to train neural network having very complex network architecture. Machine learning techniques are considered to help physicians to diagnose endoscopic images, where very higher skills are required for diagnosis. This presentation will show several automated diagnosis methods for endoscopic images. The first method classifies pathological types of colonic polyps by using hand-crafted feature values. SVM is utilized for classification. The second example is the method using CNN for automated classification. We will show these method from the technological viewpoint. Also we will discuss about training data generation.

DOI： 10.11239/jsmbe.55Annual.344

Motion Vector for Outlier Elimination in Feature Matching and Its Application in SLAM Based Laparoscopic Tracking Reviewed

Wang Cheng, Oda Masahiro, Hayashi Yuichiro, Misawa Kazunari, Roth Holger, Mori Kensaku

COMPUTER ASSISTED AND ROBOTIC ENDOSCOPY AND CLINICAL IMAGE-BASED PROCEDURES Vol. 10550 page： 60-69 2017

Micro-CT Guided 3D Reconstruction of Histological Images Reviewed

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

PATCH-BASED TECHNIQUES IN MEDICAL IMAGING (PATCH-MI 2017) Vol. 10530 page： 93-101 2017

Airway extraction from 3D chest CT volumes based on iterative extension of VOI enhanced by cavity enhancement filter Reviewed

Meng Qier, Kitasaka Takayuki, Oda Masahiro, Mori Kensaku

SPIE MEDICAL IMAGING 2017: COMPUTER-AIDED DIAGNOSIS Vol. 10134 2017

Hessian-Assisted Supervoxel: Structure-Oriented Voxel Clustering and Application to Mediastinal Lymph Node Detection from CT Volumes Reviewed International coauthorship

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 Vol. 10134 2017

Tracking and segmentation of the airways in chest CT using a fully convolutional network Reviewed

Qier Meng, Holger R. Roth, Takayuki Kitasaka, Masahiro Oda, Junji Ueno, Kensaku Mori

MICCAI 2017 Vol. 10434 LNCS page： 198 - 207 2017

TBS: Tensor-based supervoxels for unfolding the heart Reviewed International coauthorship

Hirohisa Oda, Holger R. Roth, Kanwal K. Bhatia, Masahiro Oda, Takayuki Kitasaka, Toshiaki Akita, Julia A. Schnabel, Kensaku Mori

MICCAI 2017 Vol. 10433 LNCS page： 681 - 689 2017

Comparison of the Deep-Learning-Based Automated Segmentation Methods for the Head Sectioned Images of the Virtual Korean Human Project Reviewed

Eshghi Mohammad, Roth Holger R., Oda Masahiro, Chung Min Suk, Mori Kensaku

PROCEEDINGS OF THE FIFTEENTH IAPR INTERNATIONAL CONFERENCE ON MACHINE VISION APPLICATIONS - MVA2017 page： 290 - 293 2017

Automatic Segmentation of Head Anatomical Structures from Sparsely-annotated Images Reviewed

Sugino Takaaki, Roth Holger R., Eshghi Mohammad, Oda Masahiro, Chung Min Suk, Mori Kensaku

2017 IEEE INTERNATIONAL CONFERENCE ON CYBORG AND BIONIC SYSTEMS (CBS) page： 145 - 149 2017

3D FCN Feature Driven Regression Forest-Based Pancreas Localization and Segmentation Reviewed International coauthorship

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 Vol. 10553 page： 222 - 230 2017

Extraction of membrane structure in eyeball from MR volumes Reviewed

Oda Masahiro, Kin Taichi, Mori Kensaku

SPIE MEDICAL IMAGING 2017: BIOMEDICAL APPLICATIONS IN MOLECULAR, STRUCTURAL, AND FUNCTIONAL IMAGING Vol. 10137 2017

Influence of using 3D images and 3D-printed objects on the formation of spatial mental models of experts and novices

MAEHIGASHI Akihiro, MIWA Kazuhisa, ODA Masahiro, NAKAMURA Yoshihiko, MORI Kensaku, IGAMI Tsuyoshi

JSAI Technical Report, SIG-ALST Vol. 79 ( 0 ) 2017

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 page： 2669 - 2674 2017

Language：Japanese Publisher：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.