Current Evidence and Future Perspective of Accuracy of Artificial Intelligence Application for Early Gastric Cancer Diagnosis With Endoscopy: A Systematic and Meta-Analysis

Abstract

Background & Aims: Gastric cancer is the common malignancies from cancer worldwide. Endoscopy is currently the most effective method to detect early gastric cancer (EGC). However, endoscopy is not infallible and EGC can be missed during endoscopy. Artificial intelligence (AI)-assisted endoscopic diagnosis is a recent hot spot of research. We aimed to quantify the diagnostic value of AI-assisted endoscopy in diagnosing EGC. Method: The PubMed, MEDLINE, Embase and the Cochrane Library Databases were searched for articles on AI-assisted endoscopy application in EGC diagnosis. The pooled sensitivity, specificity, and area under the curve (AUC) were calculated, and the endoscopists' diagnostic value was evaluated for comparison. The subgroup was set according to endoscopy modality, and number of training images. A funnel plot was delineated to estimate the publication bias. Result: 16 studies were included in this study. We indicated that the application of AI in endoscopic detection of EGC achieved an AUC of 0.96 (95% CI, 0.94-0.97), a sensitivity of 86% (95% CI, 77-92%), and a specificity of 93% (95% CI, 89-96%). In AI-assisted EGC depth diagnosis, the AUC was 0.82(95% CI, 0.78-0.85), and the pooled sensitivity and specificity was 0.72(95% CI, 0.58-0.82) and 0.79(95% CI, 0.56-0.92). The funnel plot showed no publication bias. Conclusion: The AI applications for EGC diagnosis seemed to be more accurate than the endoscopists. AI assisted EGC diagnosis was more accurate than experts. More prospective studies are needed to make AI-aided EGC diagnosis universal in clinical practice.

Keywords: artificial intelligence; deep learning; early gastric cancer; endoscopy; machine learning.

Figure 1

Artificial intelligence methods in medical imaging. Artificial intelligence (AI) methods for a typical classification task were shown. Two classical methods comprise traditional machine learning (A) and deep learning (B). Conv, Convolutional layer; Pool, Pooling layer; FC, receiver operating characteristic curve; EGC,: Early gastric cancer.

Figure 2

The forest plot of pooled sensitivity and specificity of AI detection on EGC. The pooled sensitivity was 86% (95% CI, 77–92%) and specificity was 93% (95% CI, 89–96%).

Figure 3

The forest plot of pooled sensitivity and specificity of AI distinction depth on EGC. The pooled sensitivity was 72% (95% CI, 58–82%) and specificity was 79% (95% CI, 56–92%).

Figure 4

Area under the receiver operating characteristic curve (A). The AUC of the AI-assisted endoscopy diagnose in the EGC detection was 0.96 (95% CI, 0.94–0.97). (B) The AUC of the AI-assisted endoscopy diagnose in the EGC depth distinction was 0.82 (95% CI, 0.78–0.85).

Figure 5

Result of subgroup analysis. (A) The pooled sensitivity and specificity of number of images in training process showed when the images were more than 10,000, the diagnostic value would be better. (B) The pooled sensitivity and specificity of AI detection, expert endoscopist, and non-expert endoscopist showed AI detection and expert endoscopist judgement were significantly more accurate than non-expert endoscopist. (C) The pooled sensitivity and specificity of original images extracted by NBI and WLE showed NBI image applied performed better.