Dr Tomohiro TADA AI Medical Services Inc Japan Link to biosketch

Abstract

In recent years, AI’s image diagnostic capability has surpassed human beings due to three factors: deep learning (CNN: convolutional neural network), high-performance computer (GPU), and a large amount of digitized image data. In application of AI to gastrointestinal endoscopy, Japanese endoscopists have shown their presence by presenting the world’s first achievements one after another.

AI application to detect gastric cancer (GC) was first reported by Hirasawa et al. [1] marking 98.6% sensitivity for cancers with diameter of 6 mm or more. Ishioka et al. applied the system to videos, and the results were similar to those with still images, 94.1% sensitivity [2]. Furthermore, according to Ikenoyama et al., using still images, AI’s sensitivity to detect GC was significantly higher than 67 endoscopists with a difference of 26.5% (58.4% vs 31.9%) [3]. Horiuchi et al [4] and Ueyama et al [5] reported AI’s high accuracy to diagnose GC based on magnifying endoscopy with narrow-band imaging (ME-NBI). Also, Horiuchi et al. reported that the sensitivity of AI was 87.4% for ME-NBI videos in diagnosing GC which is equivalent to experts [6]. Namikawa et al. developed an AI to diagnose gastric ulcer (GU), which could distinguish GC from GU by 99% sensitivity [7]. Nagao et al. reported that AI could predict the invasion depth of GC with 94.5% sensitivity [8].

In this presentation, I will look back at the past achievements about the endoscopic AI for GC and introduce you to the latest status of our development. In addition, I will explain the future prospects of endoscopic AI which will prevail around the world.

References
1. Hirasawa T et al. Gastric Cancer. 2018 ;21:653-660.
2. Ishioka M et al. Dig Endosc. 2019 Mar;31(2):e34-e35.
3. Ikenoyama Y et al. Dig Endosc. 2020;10.1111/den.13688.
4. Horiuchi Y et al. Dig Dis Sci. 2020;65(5):1355-1363.
5. Ueyama H et al. J Gastroenterol Hepatol. 2020;10.1111/jgh.15190.
6. Horiuchi Y et al. Gastrointest Endosc. 2020;92(4):856-865.
7. Namikawa K et al. Endoscopy 2020 ;10.1055/a-1194-8771.
8. Nagao S et al. Gastrointest Endosc. 2020;92(4):866-873.