Development of a fine tuning model of radiology with deep learning

• Project/Area Number: 18K15597
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 52040:Radiological sciences-related
• Research Institution: Osaka City University
• Principal Investigator: Tsutsumi Shinichi 大阪市立大学, 健康科学イノベーションセンター, 特別研究員 (60647866)
• Project Period (FY): 2018-04-01 – 2022-03-31
• Project Status: Completed (Fiscal Year 2021)
• Budget Amount: ¥4,160,000 (Direct Cost: ¥3,200,000、Indirect Cost: ¥960,000); Fiscal Year 2021: ¥130,000 (Direct Cost: ¥100,000、Indirect Cost: ¥30,000); Fiscal Year 2020: ¥260,000 (Direct Cost: ¥200,000、Indirect Cost: ¥60,000); Fiscal Year 2019: ¥130,000 (Direct Cost: ¥100,000、Indirect Cost: ¥30,000); Fiscal Year 2018: ¥3,640,000 (Direct Cost: ¥2,800,000、Indirect Cost: ¥840,000)
• Keywords: 人工知能 / AI / 乳癌 / マンモグラフィ / 説明可能なAI / 可視化 / 病理 / 深層学習 / Deep Learning / コンピュータ支援診断 / 機械学習

Outline of Final Research Achievements

We are pleased to present the results of a study in which pathological classification of breast cancer, which is usually determined by invasive pathological methods, could be performed based solely on mammograms. This is the greatest achievement because it was created using a large amount of mammography data in the context of this project funded by this KAKENHI.

Academic Significance and Societal Importance of the Research Achievements

今回の研究は、AIの判断根拠を可視化することで、AIを説明に挑戦した。この点ではAIと医師の架け橋となりうるような研究である。また、このモデルはGitHub上でオープンソース(https://github.com/ pathology-mammography)で公開しており、すべての研究者が本モデルを参考することができ、比較やさらなる発展を望むことができる。

Research Products

• [Journal Article] Visualizing “featureless” regions on mammograms classified as invasive ductal carcinomas by a deep learning algorithm: the promise of AI support in radiology (2020)
  • Author(s): Ueda Daiju、Yamamoto Akira、Takashima Tsutomu、Onoda Naoyoshi、Noda Satoru、Kashiwagi Shinichiro、Morisaki Tamami、Tsutsumi Shinichi、Honjo Takashi、Shimazaki Akitoshi、Goto Takuya、Miki Yukio
  • Journal Title: Japanese Journal of Radiology Volume: 39 Issue: 4 Pages: 333-340
  • DOI: 10.1007/s11604-020-01070-9
  • NAID: 210000161498
  • Peer Reviewed / Open Access
• [Presentation] マンモグラフィへのAIの応用 (2022)
  • Author(s): 植田大樹
  • Organizer: Breast Cancer Web Seminar
  • Invited
• [Presentation] マンモグラフィへのAIの応用 (2022)
  • Author(s): 植田大樹
  • Organizer: 第31回日本乳癌画像研究会
  • Invited