2022 Fiscal Year Final Research Report
Establishment of AI diagnostic technology for hematologic malignancy
| Project/Area Number |
21K19462
|
|---|
| Research Category |
Grant-in-Aid for Challenging Research (Exploratory)
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
Medium-sized Section 52:General internal medicine and related fields
|
|---|
| Research Institution | The University of Tokyo |
|---|
Principal Investigator |
Kato Motohiro 東京大学, 医学部附属病院, 教授 (40708690)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
出口 隆生 国立研究開発法人国立成育医療研究センター, 小児がんセンター, 診療部長 (70345990)
|
|---|
| Project Period (FY) |
2021-07-09 – 2023-03-31
|
| Keywords | 癌 / 白血病 |
|---|
| Outline of Final Research Achievements |
This study attempted to establish AI diagnostic support technology for precise diagnosis by having AI learn diagnostic information obtained by surface marker analysis results.
The AI achieved a 99% correct diagnosis classification rate. The AI's decision time was shorter than that of a physician's diagnosis.
The presence of sentinel cytogenetic abnormalities in leukemia cells could be estimated from the patterns of surface markers with a 76% to 90% agreement rate. Furthermore, we were able to confirm which surface markers are important for each immunodiagnosis and genomic aberration classification to be an explainable AI.
|
| Free Research Field |
小児科学
|
| Academic Significance and Societal Importance of the Research Achievements |
白血病の治療には、白血病を正確に診断し、それぞれの分類に最適な治療骨格と強度を選択することが重要である。しかし、白血病の診断技術は専門性が高く、その精度が診療の内容に直結し、治療成績の向上や臨床研究の質の担保に必要である。本研究では、専門医の診断をAIで支援させることの実行可能性を確認した。開発した診断システムの性能をさらに向上させ、一般化し診療実装に向けるために臨床研究を開始する予定である。
|
