胃内視鏡検診におけるAIの活用

Artificial Intelligence Application in Endoscopic Screening for Gastric Cancer

要旨●胃内視鏡検診の普及に伴い，検査の質を高い精度で均てん化させ，どのように精度管理を行っていくかが大きな課題となっている．一方，人工知能（AI）を用いた画像認識は，機械学習とdeep learningという革新的技術により飛躍的に発展した．胃内視鏡診断でもAIによる内視鏡観察部位診断，胃癌の存在診断・質的診断の研究が進められおり，医師と同等レベルの精度が報告されている．現在，AIによる二次読影支援および検査時のリアルタイム診断が臨床導入されようとしている．胃内視鏡検診の精度管理には，人間の力だけではなくAIの支援が有用であり，AIの導入により均てん化された質の高い内視鏡検査が期待される．

　Ensuring an equalized examination quality with high accuracy and controlling the examination accuracy are major challenges with the spread of endoscopic screening for gastric cancer. Meanwhile, image recognition using AI（artificial intelligence）has made dramatic progress with innovative machine learning and deep learning technologies. AI has been studied to diagnose endoscopic observation sites and presence, as well as qualitative gastric cancer diagnosis in gastrointestinal endoscopic diagnosis, and its accuracy has been reported as equivalent to that of physicians. Currently, AI is about to be clinically introduced for double-checking after examination and real-time diagnosis during examination. Not only human power but also AI support is useful for quality control of endoscopic screening examinations. Additionally, the introduction of AI is expected to lead to more uniform and higher-quality endoscopic examinations.