| Project/Area Number |
17K10374
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Radiation science
|
|---|
| Research Institution | Jikei University School of Medicine |
|---|
Principal Investigator |
Nakata Norio 東京慈恵会医科大学, 医学部, 准教授 (80237297)
|
|---|
| Project Period (FY) |
2017-04-01 – 2021-03-31
|
| Project Status |
Completed (Fiscal Year 2020)
|
| Budget Amount *help |
¥3,900,000 (Direct Cost: ¥3,000,000、Indirect Cost: ¥900,000)
Fiscal Year 2019: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
Fiscal Year 2018: ¥650,000 (Direct Cost: ¥500,000、Indirect Cost: ¥150,000)
Fiscal Year 2017: ¥2,080,000 (Direct Cost: ¥1,600,000、Indirect Cost: ¥480,000)
|
|---|
| Keywords | 人工知能 / 機械学習 / 深層学習 / 超音波診断 / 乳腺 / 乳癌 / 画像診断 / コンピュータ支援診断 / ソフトコンピューティング |
|---|
| Outline of Final Research Achievements |
In deep learning, first of all, as we proceeded with the analysis of moving images, there was a need to learn with still images, and we conducted research on deep learning for image classification. The development of an algorithm to differentiate between benign and malignant mammary masses in still ultrasound images prior to contrast was almost completed, and the results of the determination of benign and malignant mammary masses for malignant tumors were 85.6% fit rate, 92.7% recall rate, 84.7% specificity, 89.1% F value (F1 Score), 88.7% overall accuracy, and 96.0% AUC on the ROC curve In addition, the overall accuracy was 88.7% and the AUC of the ROC curve was 96.0%. Next, we conducted research on deep learning to diagnose ultrasound images in moving images. We were able to develop an algorithm to detect and track mammary masses in moving images, and were able to differentiate benign and malignant mammary masses in moving images.
|
| Academic Significance and Societal Importance of the Research Achievements |
乳癌広がり診断に用いられMRI 検査と比較して、超音波検査は、安価で、手術中も手軽に使用できるなどの利点がある。超音波検査が客観性の点でMRIなどに劣っている点を克服するため画像定量化に基づく診断基準の作成と、これを支援するコンピュータ支援診断（CAD）システムの必要性が高まっている。すなわち、超音波検査をする検査者（超音波検査士や医師）の技量や診断能力の個人差により、その診断成績に優劣が生じるという問題がある。本研究の成果により、乳腺腫瘤の診断のうち乳癌の超音波診断のCADシステム開発のための基礎的研究が進み、近い将来これらのシステムが超音波診断装置に搭載されることが期待されるようになった。
|
