Annual Report 2021
Division of Medical AI Research and Developmenty
Ryuji Hamamoto, Syuzo Kaneko, Kazuma Kobayashi, Masaaki Komatsu, Ken Asada, Satoshi Takahashi, Ken Takasawa, Masayoshi Yamada, Masamichi Takahashi, Kyoko Fujioka, Noriko Ikawa, Hiroko Kondo, Shigemi Yamada, Hidenori Machino, Amina Bolatkan, Norio Shinkai, Ai Dozen, Ryo Shimoyama, Akira Sakai, Rina Aoyama
Introduction
Our division has served as the core of medical AI research and development at the Tsukiji campus since 2016, when it launched a project to develop an integrated cancer medical system using artificial intelligence technology (AI) as part of JST's CREST project. In 2018, the project was added to the CREST project of the Cabinet Office's Public/Private R&D Investment Strategic Expansion Program (PRISM) "Development of Artificial Intelligence to Accelerate the Creation of New Drugs", and we are promoting research in close collaboration with staff from 20 departments of the National Cancer Center Hospital. In particular, we emphasize the following two points in our research activities: 1) Aiming for actual clinical application: promoting research for patients without falling into research for research's sake, and 2) Building an integrated database that includes a wealth of high-quality medical information.
Research activities
1. Development of new cancer diagnostic methods using AI technology
In collaboration with doctors from various departments of the National Cancer Center Hospital, AI software was developed to support endoscopic diagnosis, ultrasound diagnosis, and skin imaging diagnosis. In particular, the endoscopic diagnostic support AI we have developed is already in clinical application, having received regulatory approval in 2020 and met the CE Mark requirements. In FY2021, we successfully developed an AI that automatically and robustly predicts pathological diagnosis based on the revised Vienna Classification. The results indicate that the developed AI system could assist endoscopists in real time to avoid false positives during colonoscopy and improve differential diagnosis of colorectal cancer.
2. Development of a medical AI development support platform
In general, when developing AI technology to support diagnostic imaging, multiple tools are utilized to create a large number of data for the AI to train with, and after designing a learning model of what learning method to use, training is performed. Currently, a high level of engineering knowledge is required to carry out a series of development processes, as popular annotation tools are not optimized for medical images, and the subsequent learning process requires mastery of individual tools. Therefore, we have developed, in collaboration with Fujifilm, a medical AI development support platform, a research infrastructure system that enables physicians to develop AI technologies (Figure 1). By using the functions of this platform, it is no longer necessary to build the software and hardware environment for AI development and acquire the advanced engineering knowledge required for designing learning models, which have hitherto been required for research and development of image diagnosis support AI technology. It can also reduce the enormous amount of time physicians spend on processing (annotation) and management to create a large number of training data.
3. Construction of the world's largest integrated data base for lung cancer oriented toward AI analysis and development of a multi-omics analysis platform using AI technology
We are building the world's largest integrated lung cancer database oriented toward AI analysis, with more than 1,700 cases. In conjunction with the construction of the integrated database, a system was also built to efficiently collect medical information from electronic medical records and pathology systems, as well as medical image information from various modalities. Furthermore, the company has developed several platforms using AI technology to efficiently and accurately analyze large-scale omics data related to cancer.
Education
A total of five graduate students from Tokyo Medical and Dental University, Juntendo University, Keio University, and Showa University belonged to this division and were provided research guidance. Members of the division also actively participated in young researchers' seminars sponsored by the research institute.
Future Prospects
1. We will continue to promote research and development of novel cancer diagnosis support AI in collaboration with various departments and divisions of the National Cancer Center Hospital, aiming for clinical application.
2. We will promote research utilization and validation of the usefulness of this platform in multiple research themes within the National Cancer Center, and aim for social implementation in collaboration with FUJIFILM Corporation.
3. We will expand the integrated lung cancer database while also integrating it with the whole genome project, and continue to develop an omics analysis platform using machine learning and deep learning technologies.
List of papers published in 2021
Journal
1. Kobayashi Kato M, Asami Y, Takayanagi D, Matsuda M, Shimada Y, Hiranuma K, Kuno I, Komatsu M, Hamamoto R, Matsumoto K, Ishikawa M, Kohno T, Kato T, Shiraishi K, Yoshida H. Clinical impact of genetic alterations of CTNNB1 in patients with grade 3 endometrial endometrioid carcinoma. Cancer science, 113:1712-1721, 2022
2. Machino H, Kaneko S, Komatsu M, Ikawa N, Asada K, Nakato R, Shozu K, Dozen A, Sone K, Yoshida H, Kato T, Oda K, Osuga Y, Fujii T, von Keudell G, Saloura V, Hamamoto R. The metabolic stress-activated checkpoint LKB1-MARK3 axis acts as a tumor suppressor in high-grade serous ovarian carcinoma. Communications biology, 5:39, 2022
3. Ryu TY, Kim K, Han TS, Lee MO, Lee J, Choi J, Jung KB, Jeong EJ, An DM, Jung CR, Lim JH, Jung J, Park K, Lee MS, Kim MY, Oh SJ, Hur K, Hamamoto R, Park DS, Kim DS, Son MY, Cho HS. Human gut-microbiome-derived propionate coordinates proteasomal degradation via HECTD2 upregulation to target EHMT2 in colorectal cancer. The ISME journal, 16:1205-1221, 2022
4. Bolatkan A, Asada K, Kaneko S, Suvarna K, Ikawa N, Machino H, Komatsu M, Shiina S, Hamamoto R. Downregulation of METTL6 mitigates cell progression, migration, invasion and adhesion in hepatocellular carcinoma by inhibiting cell adhesion molecules. International journal of oncology, 60:2022
5. Sakai A, Komatsu M, Komatsu R, Matsuoka R, Yasutomi S, Dozen A, Shozu K, Arakaki T, Machino H, Asada K, Kaneko S, Sekizawa A, Hamamoto R. Medical Professional Enhancement Using Explainable Artificial Intelligence in Fetal Cardiac Ultrasound Screening. Biomedicines, 10:2022
6. Tanosaki T, Mikami Y, Shindou H, Suzuki T, Hashidate-Yoshida T, Hosoki K, Kagawa S, Miyata J, Kabata H, Masaki K, Hamamoto R, Kage H, Miyashita N, Makita K, Matsuzaki H, Suzuki Y, Mitani A, Nagase T, Shimizu T, Fukunaga K. Lysophosphatidylcholine Acyltransferase 1 Deficiency Promotes Pulmonary Emphysema via Apoptosis of Alveolar Epithelial Cells. Inflammation, 2022
7. Kawaguchi RK, Takahashi M, Miyake M, Kinoshita M, Takahashi S, Ichimura K, Hamamoto R, Narita Y, Sese J. Assessing Versatile Machine Learning Models for Glioma Radiogenomic Studies across Hospitals. Cancers, 13:2021
8. Takahashi S, Takahashi M, Tanaka S, Takayanagi S, Takami H, Yamazawa E, Nambu S, Miyake M, Satomi K, Ichimura K, Narita Y, Hamamoto R. A New Era of Neuro-Oncology Research Pioneered by Multi-Omics Analysis and Machine Learning. Biomolecules, 11:2021
9. Kobayashi K, Miyake M, Takahashi M, Hamamoto R. Observing deep radiomics for the classification of glioma grades. Scientific reports, 11:10942, 2021
10. Yamada M, Saito Y, Yamada S, Kondo H, Hamamoto R. Detection of flat colorectal neoplasia by artificial intelligence: A systematic review. Best practice & research. Clinical gastroenterology, 52-53:101745, 2021
11. Kaneko S, Mitsuyama T, Shiraishi K, Ikawa N, Shozu K, Dozen A, Machino H, Asada K, Komatsu M, Kukita A, Sone K, Yoshida H, Motoi N, Hayami S, Yoneoka Y, Kato T, Kohno T, Natsume T, Keudell GV, Saloura V, Yamaue H, Hamamoto R. Genome-Wide Chromatin Analysis of FFPE Tissues Using a Dual-Arm Robot with Clinical Potential. Cancers, 13:2021
12. Kobayashi K, Hataya R, Kurose Y, Miyake M, Takahashi M, Nakagawa A, Harada T, Hamamoto R. Decomposing normal and abnormal features of medical images for content-based image retrieval of glioma imaging. Medical image analysis, 74:102227, 2021
13. Kuno I, Takayanagi D, Asami Y, Murakami N, Matsuda M, Shimada Y, Hirose S, Kato MK, Komatsu M, Hamamoto R, Okuma K, Kohno T, Itami J, Yoshida H, Shiraishi K, Kato T. TP53 mutants and non-HPV16/18 genotypes are poor prognostic factors for concurrent chemoradiotherapy in locally advanced cervical cancer. Scientific reports, 11:19261, 2021
14. Asada K, Takasawa K, Machino H, Takahashi S, Shinkai N, Bolatkan A, Kobayashi K, Komatsu M, Kaneko S, Okamoto K, Hamamoto R. Single-Cell Analysis Using Machine Learning Techniques and Its Application to Medical Research. Biomedicines, 9:2021
15. Asada K, Kaneko S, Takasawa K, Machino H, Takahashi S, Shinkai N, Shimoyama R, Komatsu M, Hamamoto R. Integrated Analysis of Whole Genome and Epigenome Data Using Machine Learning Technology: Toward the Establishment of Precision Oncology. Frontiers in oncology, 11:666937, 2021
16. Lee J, Kim K, Ryu TY, Jung CR, Lee MS, Lim JH, Park K, Kim DS, Son MY, Hamamoto R, Cho HS. EHMT1 knockdown induces apoptosis and cell cycle arrest in lung cancer cells by increasing CDKN1A expression. Molecular oncology, 15:2989-3002, 2021
17. Komatsu M, Sakai A, Dozen A, Shozu K, Yasutomi S, Machino H, Asada K, Kaneko S, Hamamoto R. Towards Clinical Application of Artificial Intelligence in Ultrasound Imaging. Biomedicines, 9:2021
18. Kaneko S, Takasawa K, Asada K, Shinkai N, Bolatkan A, Yamada M, Takahashi S, Machino H, Kobayashi K, Komatsu M, Hamamoto R. Epigenetic Mechanisms Underlying COVID-19 Pathogenesis. Biomedicines, 9:2021
19. Asada K, Komatsu M, Shimoyama R, Takasawa K, Shinkai N, Sakai A, Bolatkan A, Yamada M, Takahashi S, Machino H, Kobayashi K, Kaneko S, Hamamoto R. Application of Artificial Intelligence in COVID-19 Diagnosis and Therapeutics. Journal of personalized medicine, 11:2021