Combination Of AI-Based Endoscopy and Biomarkers for Gastric Indefinite Dysplasia Diagnosis

Authors:
Yoshiyuki WATANABE^1,2,4, Ichiro ODA², Hiroyuki YAMAMOTO³, Seiji FUTAGAMI⁴, Tomohiro TADA⁵


Affliations:
¹Division of Gastroenterology, Department of Internal Medicine, St. Marianna University School of Medicine, Kawasaki, Japan
²Department of Internal Medicine, Kawasaki Rinko General Hospital, Kawasaki, Japan
³Department of Bioinformatics, St. Marianna University Graduate School of Medicine, Kawasaki, Japan
⁴Department of Internal Medicine, Division of Gastroenterology, Nippon Medical School, Tokyo, Japan
⁵Tada Tomohiro Institute of Gastroenterology and Proctology, Saitama, Japan


Corresponding email address:
Yoshiyuki WATANABE (ponponta@marianna-u.ac.jp)


Publication:
J Clin Lab Anal. 2022 Jan;36(1):e24122. doi: 10.1002/jcla.24122. Epub 2021 Nov 22.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8921718/

Background: GI Endoscopy and biopsy-based pathological findings is basically needed to diagnose early gastric cancer. However, the information of biopsy specimen is limited because of the topical procedure. Thus, pathology doctors diagnose as gastric indefinite for dysplasia (GIN).

Methods: We compared the accuracy of physician-performed endoscopy (trainee; n=3: specialists; n=3), AI-based endoscopy, and/or molecular markers (DNA methylation: BARHL2, MINT31, TET1, miR-148a, miR-124a-3, NKX6-1, mutation: TP53 and MSI) in diagnosing GIN lesions. We enrolled 24388 patients who underwent an endoscopy and 71 patients were diagnosed as GINs. Thirty-two endoscopic submucosal dissection (ESD) cases in 71 GINs and 32 endoscopically resected tissues were assessed by endoscopists, artificial intelligence (AI), and molecular markers to identify benign or malignant lesions.

Results: The endoscopy specialists group showed the highest accuracy in ROC curve (AUC: 0.931) followed by AI and miR148a DNA methylation (AUC: 0.825) than trainee endoscopists (AUC: 0.588).

Conclusion: AI with miR148s DNA methylation-based diagnosis is a potential modality to diagnose GIN.