Development of beyond human-level AI for medical image diagnosis systems

2021 Fiscal Year Final Research Report

• Project/Area Number: 18K19892
• Research Category: Grant-in-Aid for Challenging Research (Exploratory)
• Allocation Type: Multi-year Fund
• Review Section: Medium-sized Section 90:Biomedical engineering and related fields
• Research Institution: Tohoku University
• Principal Investigator: HOMMA Noriyasu 東北大学, 医学系研究科, 教授 (30282023)
• Project Period (FY): 2018-06-29 – 2022-03-31
• Keywords: 計算機支援診断システム

Outline of Final Research Achievements

In this research, next generation medical image diagnosis systems have been developed by using a deep learning based artificial intelligence (AI) that is capable of super-performance beyond human experts. The AI-aided (AID) systems have been applied for breast cancer screening using mammography exmas, gastric cancer screening using fluoroscopic stomac exams, and drowning diagnosis using forensic imaging (autopsy imaging) of x-ray computed tomography. Experimental results showed that the AID systems were able to achieve the human experts' level performance for the tasks. Specifically, the AID system for mammographic diagnosis demonstrated superior performance beyond human experts and further more, the system made human experts' performance even better. These results clearly demonstrated the usefulness and effectiveness of the proposed AID systems in clinical use.

Free Research Field

知的医用システム

Academic Significance and Societal Importance of the Research Achievements

本研究で開発したAIDシステムは、臨床上十分有用な高性能を達成した。このような医師の診療業務を支援、さらには一部を代替可能な高性能AIの実用化により、医師の業務量低減や非専門領域に対する支援、さらには遠隔医療などを含めた効率化を実現することが可能になり、地方における医師不足に起因する医療提供の持続可能性や、都市部に比して専門医偏在に起因する医療の均てん化問題などの改善に繋がると期待される。