Comments on: “Endoscopic prediction of submucosal invasion in Barrett’s cancer with the use of artificial intelligence: a pilot study”

 

 Buy Article Permissions and Reprints

Ebigbo et al. recently trained an artificial intelligence (AI) system on the basis of deep artificial neural networks to differentiate between mucosal (T1a) and submucosal (T1b) Barrett’s cancer on white-light images, with sensitivity of 77 %, specificity of 64 %, and accuracy of 71 % [1]. Compared with five highly experienced endoscopists, there was no statistically significant difference. We congratulate the authors as the work has good application prospects and innovation, but it has certain limitations.

The indications for endoscopic therapy for T1b sm1 esophageal adenocarcinoma are controversial. Submucosal adenocarcinoma was once considered unsuitable for endoscopic therapy, but more recent published guidelines consider endoscopic therapy as an alternative to surgery for T1b sm1 tumors under certain conditions [2] [3] [4]. Whether tumors can be cured by endoscopic therapy depends not only on the depth of infiltration, but also on the lymphatic or vascular infiltration and degree of differentiation. According to such criteria for judgment, the final outcome of patients may also be different. Therefore, it is insufficient to differentiate between mucosal (T1a) and submucosal (T1b), as this alone cannot directly guide the choice of treatment strategy. This may limit application of the AI system. Secondly, the sample size of 230 lesion images is insufficient to train an AI system and the quality of images may influence the performance of the system. In endoscopy practice, endoscopists can judge a lesion through use of virtual chromoendoscopy and by observing dynamic characteristic such as the movement of the esophageal wall. Therefore, this study does not reflect the ideal situation in real-life settings. In addition, no normal images were included in the testing of the AI system and no P values were reported for the comparison of the Al system versus endoscopists.

Although this study has some shortcomings, this is the first step toward developing an AI system to aid in the prediction of submucosal invasion of Barrett’s cancer.

Publication History

Article published online:
27 January 2022

© 2022. Thieme. All rights reserved.

Georg Thieme Verlag KG
Rüdigerstraße 14, 70469 Stuttgart, Germany

• References

• 1 Ebigbo A, Mendel R, Rückert T. et al. Endoscopic prediction of submucosal invasion in Barrett’s cancer with the use of artificial intelligence: a pilot study. Endoscopy 2021; 53: 878-883
  Article in Thieme ConnectPubMedGoogle Scholar
• 2 Fitzgerald RC, di Pietro M, Ragunath K. et al. British Society of Gastroenterology guidelines on the diagnosis and management of Barrett’s oesophagus. Gut 2014; 63: 7-42
  CrossrefPubMedGoogle Scholar
• 3 Shaheen NJ, Falk GW, Iyer PG. et al. ACG clinical guideline: diagnosis and management of Barrett’s esophagus. Am J Gastroenterol 2016; 111: 30-50
  CrossrefPubMedGoogle Scholar
• 4 Weusten B, Bisschops R, Coron E. et al. Endoscopic management of Barrett’s esophagus: European Society of Gastrointestinal Endoscopy (ESGE) Position Statement. Endoscopy 2017; 49: 191-198
  Article in Thieme ConnectPubMedGoogle Scholar