Development of a hemostasis support system for endoscopic surgery using artificial intelligence technology

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K08997
• 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: Kanagawa Cancer Center Research Institute
• Principal Investigator: Oshima Takashi 地方独立行政法人神奈川県立病院機構神奈川県立がんセンター(臨床研究所), その他部局等, 部長 (10448665)
• Co-Investigator(Kenkyū-buntansha): 篠原 尚 兵庫医科大学, 医学部, 教授 (70319549)
• Project Period (FY): 2020-04-01 – 2023-03-31
• Keywords: 人工知能技術 / 内視鏡外科手術 / 止血支援システム

Outline of Final Research Achievements

We constructed a hemostasis support system in which AI automatically recognizes bleeding in laparoscopic surgery and plays a repeat on the screen. As a result of verification in surgery, over- and under-detection of bleeding was observed. Since this was thought to be due to fluctuations in accuracy among endoscopic systems, we created teacher images from multiple types of endoscopic systems, and the image annotated by the physician to the evaluation image was used as the correct image. The Dice coefficient was used as the evaluation index.The Dice coefficient increased with image learning and reached 0.567. Since the region is generally matched to the naked eye when the Dice coefficient is 0.8 or higher, we will expand the training data with this as the target level and aim for effective use in actual clinical practice.

Free Research Field

外科学

Academic Significance and Societal Importance of the Research Achievements

内視鏡外科手術の躍進は目覚ましいが，内視鏡外科手術でより問題となるのは出血である。出血は、解剖構造の認識を著しく低下させ，正確，安全かつ迅速な手術の進行を妨げる。さらに，出血量が多くなると、開腹手術への移行を余儀なくさせるばかりでなく，術後合併症の増加を招き，長期生存を低下させ、医療費も増大させる。そこでわれわれは， AIにより、出血を自動認識し、外科医が必要とする出血情報を選別して，出血シーンの動画を腹腔鏡のモニターと同一のモニターの右上の画面上で自動逆再生することで，内視鏡外科手術における止血操作を正確且つ迅速に行う止血支援システムの開発がほぼ完成し、臨床での実用化を目指している。