Diagnosis of Esophagitis Based on Deep Learning

Abstract

With the advent of the information age, medical image classification technology has become a hot research topic in the field of computer vision. Based on the Tensorflow deep learning framework, a ResNet model of esophagitis disease recognition from gastrointestinal mirror image is established in this paper. Further, the random erasing data augmentation algorithm RE is applied to optimizing the images, which are then used in the network. As a result, the accuracy of the optimized network over the Kvasir data set can reach 97%.