Improvement of Accuracy and Practical Application of Breast Ultrasound Diagnosis Assistance System Using Artificial Intelligence

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K08993
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 55010:General surgery and pediatric surgery-related
• Research Institution: Keio University
• Principal Investigator: Hayashida Tetsu 慶應義塾大学, 医学部(信濃町), 講師 (80327543)
• Co-Investigator(Kenkyū-buntansha): 永山 愛子 慶應義塾大学, 医学部(信濃町), 助教 (00573396) ; 高橋 麻衣子 慶應義塾大学, 医学部(信濃町), 助教 (50348661) ; 関 朋子 慶應義塾大学, 医学部(信濃町), 助教 (70528900)
• Project Period (FY): 2020-04-01 – 2023-03-31
• Keywords: 人工知能 / AI / 乳房超音波検査

Outline of Final Research Achievements

Artificial intelligence (AI) based on deep learning technology has made great progress and has started to be introduced into medicine. In this study, we conducted a multicenter collaborative study to establish AI diagnostic technology for breast ultrasound. We confirmed that the AI diagnosis system is accurate enough for practical use, with a sensitivity of 91.2% and specificity of 90.7%, and an AUC of 0.95 for the ROC curve based on the judgment threshold value. In addition, it was confirmed that the use of AI diagnosis by physicians as a reference for diagnosis contributed to an increase in the accuracy, and that there were no safety issues, such as an increase in missed cases, caused by the use of AI diagnosis as a reference.

Free Research Field

乳腺外科

Academic Significance and Societal Importance of the Research Achievements

AIの医療応用への試みは、画像診断・病理診断などを中心に行われているが、関連する学会をはじめとするアカデミアや臨床医が慎重にその適用や安全性を吟味する必要がある。我々はこのように実用に耐える精度のAI診断システムを構築し、これを医師が実際に使用することが、安全にかつ正診率の向上につながることを確認した。そのため、本研究はAI診断の実力と安全性を実際の臨床に即した形式で検証するものであり、そのような報告は世界的に見ても存在しないため、医療への今後のAI導入に対して、大きな貢献を与えると考えられる。