Development of AI-based real-time EUS image diagnostic system for early-stage gastric cancer

2023 Fiscal Year Final Research Report

• Project/Area Number: 22K18210
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 90130:Medical systems-related
• Research Institution: Osaka University
• Principal Investigator: Ryotaro Uema 大阪大学, 大学院医学系研究科, 特任助教(常勤) (30939161)
• Project Period (FY): 2022-04-01 – 2024-03-31
• Keywords: 早期胃癌 / 超音波内視鏡 / 人工知能 / ディープラーニング

Outline of Final Research Achievements

The aim of this study was to develop an accurate and reproducible diagnostic system by training computers on endoscopic ultrasonography (EUS) images of gastric cancer using deep learning, and to validate its utility in clinical practice. Using an algorithm that combines segmentation and image classification models, we demonstrated that the AI has diagnostic performance comparable to those of experts. In addition, the system's performance was validated on multi-institutional datasets, confirming that it achieves diagnostic performance on par with experts.

Free Research Field

消化器癌

Academic Significance and Societal Importance of the Research Achievements

本研究によって、従来は困難と考えられた早期胃癌の超音波内視鏡診断における人工知能をもちいた診断システムの有用性を示すことができた。本研究を通じて開発したシステムを用いることで、非熟練医であっても専門医に匹敵する診断を下せるようことが期待できる。これは診断精度の向上を通じて早期胃癌の治療成績の向上につながる可能性があり、社会的意義は大きいものと考える。