Diagnosis of Gastrointestinal Diseases Using Modern CNN Techniques
Guduru Varalaxmi, Sahithi Reddy Baddam, Emmanuel Suhas Yalamarthi, K. Swaraja, K. Reddy Madhavi, C. N. Sujatha
Abstract
Endoscopy is used by the doctors in detecting the gastrointestinal diseases. Endoscopy is a procedure which helps in detection of many gastric diseases like polyps, cancer infections, celiac disease, Crohn’s disease, ulcerative colitis, diverticulitis, malabsorption and so on. As in recent days use of endoscopy is increasing drastically because of high increase in gastrointestinal related issues. As this method consumes more time and repeated trials on the patient, it may affect the patient. Therefore, detection of diseases using deep learning helps in detecting the gastric diseases without affecting the patients. By using modern Convolutional Neural Network (CNN) techniques we can acquire accurate results within no time, so that in case of emergencies the treatment can be started immediately without any delays. In this work ResNet50 is implemented which is the subset of Residual Network. RESNET (a neural network), uses the complementary clues of spectral and spatial features to improve the classification. The proposed framework attained an accuracy of 88.05%, sensitivity of 87.05% and specification of 92.33%.