Artificial intelligence for the detection of gastric precancerous conditions using image-enhanced endoscopy: What kind of abilities are required for application in real-world clinical practice?

An Optimal Artificial Intelligence System for Real-Time Endoscopic Prediction of Invasion Depth in Early Gastric Cancer

We previously constructed a VGG-16 based artificial intelligence (AI) model (image classifier [IC]) to predict the invasion depth in early gastric cancer (EGC) using endoscopic static images. However, images cannot capture the spatio-temporal information available during real-time endoscopy-the AI trained on static images could not estimate invasion depth accurately and reliably. Thus, we constructed a video classifier [VC] using videos for real-time depth prediction in EGC. We built a VC by attaching sequential layers to the last convolutional layer of IC v2, using video clips. We computed the standard deviation (SD) of output probabilities for a video clip and the sensitivities in the manner of frame units to observe consistency. The sensitivity, specificity, and accuracy of IC v2 for static images were 82.5%, 82.9%, and 82.7%, respectively. However, for video clips, the sensitivity, specificity, and accuracy of IC v2 were 33.6%, 85.5%, and 56.6%, respectively. The VC performed better analysis of the videos, with a sensitivity of 82.3%, a specificity of 85.8%, and an accuracy of 83.7%. Furthermore, the mean SD was lower for the VC than IC v2 (0.096 vs. 0.289). The AI model developed utilizing videos can predict invasion depth in EGC more precisely and consistently than image-trained models, and is more appropriate for real-world situations.

Construction of Stomach Cancer Lesion Detection Combined with Drug Therapy Based on Artificial Intelligence

The number of stomach cancer (SC) patients is increasing sharply every year, and gastroscope is a common method to check stomach-related diseases. A bulging lesion in the stomach is encountered during a gastroscopy. Due to the change in eating habits, the enhancement of health awareness, and the wide application of gastroscopy, the detection rate and cure rate of tumors have been significantly improved. This has certain clinical value for the early diagnosis and treatment of early SC. In this paper, based on the background of artificial intelligence, image segmentation technology is used to analyze and process the detection results of SC, so as to judge the effect of drug treatment. A total of 1408 gastric bulge lesions were investigated in 11023 patients during the one-year period 2019-2020. It also analyzed the age, lesion location, size, pathological type, and tumor detection results of 1408 patients. The experiment showed that among the 289 cases of submucosal bulging lesions, the detection rates of the young group, middle-aged group, and elderly group were 14.9% (43/289), 67.5% (195/289), and 17.6% (51/289), respectively. Among them, middle-aged people aged 41-65 have the highest detection rate. The incidence of gastric polyps was similar between different age groups. But with age, the rate of fundic gland polyps increases. The incidence of SC is not related to the age of the patient, but to its pathological type. The incidence of SC in middle-aged and elderly people is significantly higher than that in young people. SC is more common in the cardia, and gastrointestinal stromal tumors are most common with submucosal protrusion.