Validation of Artificial Intelligence Computer-Aided Detection on Colonic Neoplasm in Colonoscopy.

Background/Aims Detection of colon polyps during colonoscopy has importance of controlling quality regarding reducing over-all long term colorectal cancer risk. In this study, we aimed to validate previously developed artificial intelligence (AI) computer-aided detection (CADe) algorithm, called ALPHAON and to compare outcome with previous studies present AI outperforming and assisting endoscopists in detecting colon polyps. Methods We used the retrospective data of 500 still images, including 100 polyp images, and 400 normal images. Thereby we validated CADe algorithm measuring accuracy, sensitivity and specificity with result of receiver operating characteristic curves (ROC) and area under curve (AUC) in addition to comparing diagnostic performance status of 2 expert endoscopists, 2 trainees and 4 beginners with CADe algorithm. After a washing-out period of over 2 weeks, endoscopists performed polyp detection on the same dataset of the 500 endoscopic images again marked by ALPHAON. Results The CADe algorithm presented high validity on detecting colon polyps with accuracy (0.97, 95% CI:0.96 to 0.99), sensitivity (0.91, 95% CI:0.85 to 0.97), specificity (0.99, 95% CI:0.97 to 0.99) and AUC (0.967). ALPHAON outperformed expert in accuracy (0.97, 95% CI: 0.96 to 0.99), sensitivity (0.91, 95% CI: 0.85 to 0.97) and specificity (0.99, 95% CI: 0.97 to 0.99). After a washing-out period of over 2 weeks, overall validity significantly improved in both expert (accuracy 0.96 (0.94 to 0.97) p<0.001, sensitivity 0.84 (0.79 to 0.89) p<0.001, specificity 0.99 (0.97 to 0.99) p<0.001) and trainee (accuracy 0.95 (0.93 to 0.96) p<0.001, sensitivity 0.80 (0.74 to 0.86) p<0.001) with assistance of ALPHAON. Conclusion High validation performance state of CADe algorithm system was verified. ALPHAON has demonstrated enhancing validity for both expert and trainee which amplifies possibility of ALPHAON as being able endoscopic assistance. Prospective validation in multicenter would be required before application of CADe in clinical field for delicate endoscopic screening. [ABSTRACT FROM AUTHOR]