Research on physician education using eye measurement data and development of next-generation AI with a physician's point of view

2022 Fiscal Year Final Research Report

• Project/Area Number: 21K16751
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 56030:Urology-related
• Research Institution: University of Tsukuba
• Principal Investigator: ikeda Atsushi 筑波大学, 医学医療系, 講師 (50789146)
• Project Period (FY): 2021-04-01 – 2023-03-31
• Keywords: 膀胱内視鏡検査 / 膀胱癌 / 人工知能 / 視線計測 / 経尿道的膀胱腫瘍切除術

Outline of Final Research Achievements

Cystoscopy is an essential procedure in bladder cancer treatment, but its diagnostic accuracy varies depending on the physicians' knowledge and experience. In this study, we had physicians with different backgrounds observe the same case of cystoscopy using a non-contact eye tracking and eye measurement system for monitors to obtain the viewpoint coordinates and dwell time of the physicians during their observation. It was calculated and compared. It was confirmed that experienced urologists efficiently observed a wide screen area in real-time and early stages of the examination. Visualizing the observer's gaze is expected to be used to evaluate skill and for educational purposes toward improving skill.

Free Research Field

泌尿器科

Academic Significance and Societal Importance of the Research Achievements

視線計測は、医師の技量を評価しスキルアップのための課題を明らかにできる可能性があることから、消化器領域の手術においての報告例が散見される。泌尿器科領域においても膀胱内視鏡検査や経尿道膀胱腫瘍切除（TURBT）における経験の異なる医師らの視線経路の計測は、術者の技量を客観的に評価することにつながる。先行研究で開発している人工知能（AI）を利用した膀胱内視鏡検査支援システムの学習用の教師データとして、経験豊富な医師の視線という情報を加えることができれば、画像しか学んでいなかった現代のAIを次のステージへと導く可能性がある。結果として、AIのさらなる診断精度の向上と最適化につながると考えられる。