Real-Time Evaluation of Helicobacter pylori Infection by Convolution Neural Network During White-Light Endoscopy: A Prospective, Multicenter Study (With Video)

Abstract

Objectives: Convolutional neural network (CNN) during endoscopy may facilitate evaluation of Helicobacter pylori infection without obtaining gastric biopsies. The aim of the study was to evaluate the diagnosis accuracy of a computer-aided decision support system for H. pylori infection (CADSS-HP) based on CNN under white-light endoscopy. Methods: Archived video recordings of upper endoscopy with white-light examinations performed at Sir Run Run Shaw Hospital (January 2019-September 2020) were utilized to develop CADSS-HP. Patients receiving endoscopy were prospectively enrolled (August 2021-August 2022) from three centers to calculate the diagnostic property. Accuracy of CADSS-HP for H. pylori infection was also compared to endoscopic impression, urea breath test (URT) and histopathology. H. pylori infection was defined by positive test on histopathology and/or URT. Results: Video recordings of 599 patients who received endoscopy were utilized to develop CADSS-HP. Subsequently, 456 patients participated in the prospective evaluation including 189 (41.4%) with H. pylori infection. With a threshold of 0.5, CADSS-HP achieved an area under the curve (AUC) of 0.95 (95% confidence interval [CI], 0.93-0.97) with sensitivity and specificity of 91.5% (95%CI 86.4%-94.9%) and 88.8% (95%CI 84.2%-92.2%), respectively. CADSS-HP demonstrated higher sensitivity (91.5% vs 78.3%; mean difference=13.2%, 95%CI 5.7%-20.7%) and accuracy (89.9% vs 83.8%, mean difference=6.1%, 95%CI 1.6%-10.7%) compared to endoscopic diagnosis by endoscopists. Sensitivity of CADSS-HP in diagnosing H. pylori was comparable to URT (91.5% vs 95.2%; mean difference=3.7%, 95%CI -1.8%-9.4%), better than histopathology (91.5% vs 82.0%; mean difference=9.5%, 95%CI 2.3%-16.8%). Conclusions: CADSS-HP achieved high sensitivity in the diagnosis of H. pylori infection in the real-time test, outperforming endoscopic diagnosis by endoscopists and comparable to URT. Clinicaltrials.gov; ChiCTR2000030724.

Artificial Intelligence in Healthcare: 2023 Year in Review

“…Imaging publications continue to focus on various imaging modalities including interpretation of X-rays, CT scans, MRIs and ultrasounds[10–12]. Majority of this research work utilizes deep learning, more specifically and not unexpectedly convolutional neural networks[13,14]. There were a few publications which focus on optimizing image processing and clinician workflows[15].…”

Machine Learning-Based Prediction of Helicobacter pylori Infection Study in Adults

“…Takeshi et al [ 27 ] developed a universal endoscopic machine learning method to diagnose H. pylori infection, with 87.6% accuracy. Shen et al [ 28 ] evaluated a computer-aided decision-support system for H. pylori infection based on a convolutional neural network under endoscopy. It achieved high sensitivity in the diagnosis of H. pylori infection, outperforming endoscopic diagnosis by endoscopists, and was comparable with rapid urease testing.…”

The Use of Artificial Intelligence in Gastroenterology: A Glimpse Into the Present

“…When I think about AI, I am struck by its potential to enhance both the care we deliver and the additional time we may save while providing that care. In this special issue, you'll find examples of how AI may be able to improve diagnostic accuracy, whether it's computer-aided diagnosis of in situ colon polyps ( 1 ), real-time detection of Helicobacter pylori infection ( 2 ), or determining the invasion depth of esophageal squamous cell carcinoma ( 3 ). AI can also improve patient access to GI diagnostics, as some technologies may become more easily utilized by primary care providers or allied health professionals.…”