Abstract:Introduction:We aim to develop and validate a deep learning algorithm for detection of Barrett s neoplasia.Methods:We collected 132 HD WL images from 46 lesions of histologically confirmed Barrett s n...Introduction:We aim to develop and validate a deep learning algorithm for detection of Barrett s neoplasia.Methods:We collected 132 HD WL images from 46 lesions of histologically confirmed Barrett s neoplasia.Images were annotated and reviewed by two experts.119 images of non-dysplastic Barrett s were collected from 20 patients as control.Images were divided into training, validation and testing datasets.We used SegNet segmentation architecture.We collected metrics on processing speed, sensitivity, specificity and accuracy.Results:Image Processing speed was 33ms/image.The algorithm acheived sensitivity of 93%, specificity of 78% and global accuracy of 83%.Conclusions:We developed and validated an early AI algorithm with high sensitivity and reasonable specificity when compared with PIVI criteria.The ultra short image processing time would suggest this algorithm may be suitable for real time detection of Barrett s neoplasia.Read More