Ulcer Recognition based on 6-Layers Deep Convolutional Neural Network

In medical imaging, Wireless Capsule Endoscopy (WCE) is an advanced technology for detecting gastrointestinal diseases such as ulcers, polyp, bleeding, and many more. In this work, a new technique based on the 6-Layers Convolutional Neural Network (CNN) model is proposed to identify ulcers. The proposed method follows the two-step process. In the first step, a region of interest (ROI) is detected from the original images by extracting statistical-based color features and mapped on the original image. Later, a third channel is selected from a mapped image and performs thresholding. After thresholding, regions props based infected region is detected as an ROI (Region of Interest) and set as input to the newly implemented 6-Layers Convolutional Neural Network (CNN) model. Afterward, cross entropy-based features are computed from the last layers and fed to the Softmax classifier for classification performance. The experimental process is performed on the privately collected dataset and achieved an accuracy of 96.4%.