2023 Fiscal Year Final Research Report
A fundamental research on curation and verification methods of learning data to promote AI-CAD development
| Project/Area Number |
21K07636
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 52040:Radiological sciences-related
|
|---|
| Research Institution | National Hospital Organization, Kyushu Medical Center (Clinical Institute) |
|---|
Principal Investigator |
Noguchi Tomoyuki 独立行政法人国立病院機構九州医療センター(臨床研究センター), その他部局等, 放射線部長 (40380448)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
松下 由実 国立研究開発法人国立国際医療研究センター, 臨床研究センター, 臨床研究統括部 室長 (50450599)
志多 由孝 国立研究開発法人国立国際医療研究センター, センター病院, 放射線診療部門・放射線管理室医長 (50774668)
山下 孝二 九州大学, 医学研究院, 助教 (80546565)
|
|---|
| Project Period (FY) |
2021-04-01 – 2024-03-31
|
| Keywords | Deep learning / 機械学習 / 深層学習 / 医用画像 / 人工知能 |
|---|
| Outline of Final Research Achievements |
AI technology that enables computers to perform "intelligent judgment" tasks typically carried out by humans has been attracting significant attention. Particularly, "deep learning AI," which autonomously learns and makes decisions, has made remarkable advancements in image recognition. However, the decision-making process of AI is often referred to as a "black box," and the commercialization of AI systems has progressed without sufficient foundational research and verification. This makes the systems vulnerable to errors. To mitigate this risk, it is crucial to conduct fundamental research to understand AI's decision-making processes. In this study, we promoted the following initiatives: 1) Curation of training data for medical information, 2) Development of objective and efficient performance validation methods for AI-CAD, and 3) Training medical AI generalists capable of supporting AI-CAD development from a healthcare perspective.
|
| Free Research Field |
放射線医学レギュラトリーサイエンス研究
|
| Academic Significance and Societal Importance of the Research Achievements |
本基礎研究では、医用画像などの医療情報について学習データをを収集・選別・調整する方法を深く探求するとともに、AI-CADの客観的かつ効率的に性能を適正に検証する方法を見出した。こうしたAI-CAD開発技術者との共同研究を通じ、医療側からAI-CAD開発を支援できる医療系AIジェネラリストの育成を推進した。今後は共同研究者らによる次世代AI開発が発展していくものと思われ、それを引き続き指導者として支援を目指す。
|
