CNN model applied on SNP protein sequences for intestinal cancer early detection

Abstract

One of the sharply increasing incidence rate diseases, is the intestinal cancer. Detecting and identifying this disease is very difficult since it is a type of internal cancer that may not cause symptoms. Early detection of bowel cancer saves effort, minimizes diagnostic time, and helps target the best treatment to increase the cure rate. Many deep learning systems are being produced to help physicians diagnose bowel cancer early. These systems use several types of data such as histopathological images. In this article, we propose a new method of classifying intestinal cancer readings and non-pathological lesions. This method is based on the use of a deep learning network CNN (convolutional neural network). As training data, we use the SNP protein sequences of APC gene which is located at q22.2 on the human chromosome 5. SNP sequences are transformed from textual form to scalogram images then presented on different color spaces to create three databases. In order to evaluate this technique, we use quantitative measures such as: accuracy, sensitivity, specificity and precision. The obtained values are respectively 96.19%, 97.42%, 85.46% and 98.32%. These results show the performance of the proposed method and confirm the relationship between small protein changes at the SNP level of APC gene and intestinal cancer.