A deep learning-based system for accurate detectionof anatomical landmarks in colon environment
Abstract
Colonoscopy is a standard imaging tool for examining the lower gastrointestinal tract of patients to capture lesionareas. However, if a lesion area is found during the colonoscopy process, it is difficult to record its location relative tothe colon for subsequent therapy or recheck without any reference landmark. Thus, automatic detection of biologicalanatomical landmarks is highly demanded to improve clinical efficiency. In this article, we propose a novel deeplearning-based approach to detect biological anatomical landmarks in colonoscopy videos. First, raw colonoscopyvideo sequences are pre-processed to reject interference frames. Second, a ResNet-101-based network is used todetectthreebiologicalanatomicallandmarksseparatelytoobtaintheintermediatedetectionresults. Third, toachievemore reliable localization, we propose to post-process the intermediate detection results by identifying the incorrectlypredicted frames based on their temporal distribution and reassigning them back to the correct class. Finally, theaverage detection accuracy reaches 99.75%. Meanwhile, the average intersection over union of 0.91 shows a highdegree of similarity between our predicted landmark periods and ground truth. The experimental results demonstratethat our proposed model can accurately detect and localize biological anatomical landmarks from colonoscopy videos
