Clinicalstudy Of Artificial Intelligence Assisted Colonoscopy In Detectingcolorectal Polyps Clinical Study Of Artificial Intelligence Assisted Colonoscopy In Detecting Colorectal Polyps

Objective:Colorectal cancer is one of the most common cancers in the world.The incidence rate of colorectal cancer can be effectively reduced by colonoscopy and early detection of colorectal polyps at high risk.The study analyzed the information of colonoscopy procedure characteristics,colon polyp location,size,pathological classification and AI system error recognition,compared the outcome indicators such as polyp detection rate,and evaluated the effectiveness of colorectal polyp real-time artificial intelligence detection system in clinical research.Method:The subjects who met the inclusion and exclusion criteria were randomly divided into traditional enteroscopy group and AI enteroscopy group.The traditional colonoscopy group underwent ordinary colonoscopy,and the AI colonoscopy group underwent colonoscopy with real-time artificial intelligence detection system for colorectal polyps.When the artificial intelligence detection system automatically recognizes polyps or other intestinal mucosal abnormalities,a blue box will appear on the display for prompt,and the endoscopist will observe and identify the blue box area.After the completion of colonoscopy,record the required information,including the characteristics of colonoscopy procedure,the location,size and pathological classification of colonic polyps and the wrong identification of AI system.Analyze the detection rate of colorectal polyps,the average number of polyps detected in positive patients,the number of polyps detected in a single colonoscopy and the false alarm of artificial intelligence system.Results:A total of 610 subjects were included in the study,of which 316 underwent traditional colonoscopy and 294 underwent AI colonoscopy.(1)compared with the traditional colonoscopy group,the detection rate of polyps in AI colonoscopy group was significantly higher(37.34% vs 45.27%,P < 0.05).There was no significant difference between the average number of polyps detected in positive patients and the number of polyps detected in a single colonoscopy.(2)in the traditional enteroscopy group,61 polyps were detected in the morning and 88 polyps were detected in the afternoon;In AI enteroscopy group,76 polyps were detected in the morning and 91 polyps were detected in the afternoon,the difference was not statistically significant(2 = 0.669,P = 0.413).There was no significant difference in the number of polyps between traditional enteroscopy group and AI enteroscopy group in the first and fourth hours of the morning and the first and fourth hours of the afternoon(P > 0.05).⑶compared with the traditional colonoscopy group,there was no significant difference in the position and size of polyps detected in the AI colonoscopy group(P > 0.05).(4)compared with the traditional colonoscopy group,the number of insignificant(mucosal)polyps detected in the pathological classification of AI colonoscopy group increased(2 = 5.743,P = 0.017),and there was no significant difference in adenomatous polyps,proliferative polyps,inflammatory polyps and uncertainty(P > 0.05).(5)compared with the traditional enteroscopy group,there was no significant difference in the number of serrated adenoma,tubular adenoma,villous adenoma and tubular villous adenoma(2 = 3.930,P = 0.245).(6)the AI colonoscopy Group recorded 1116 false recognition times,with an average of 3.79 times per person,mostly bubbles and wrinkled walls.Conclusions:The real-time artificial intelligence detection system of colorectal polyps can indeed improve the detection rate of colorectal polyps to a certain extent.The polyp system detected by endoscopists in AI enteroscopy group was not missed.During the study period,there was no adverse event caused by its system.The research shows that the real-time artificial intelligence detection system of colorectal polyps is safe and effective in polyp detection.