Classification of Gastrointestinal Diseases from Endoscopic Images Using Convolutional Neural Network with Transfer Learning

Artificial intelligence (AI) has the potential to revolutionize the field of gastrointestinal disease diagnosis by enabling the development of accurate and efficient automated systems. This study comprehensively investigates endoscopic image classification for gastrointestinal diseases using deep learning models. Gastrointestinal diseases are a significant healthcare concern worldwide, necessitating accurate diagnostic tools. The proposed architecture relies on the convolutional layers of pre-trained CNN networks (DenseNet121, EfficientNetB5, Xception, and ResNet50) for feature extraction with a newly designed fully connected layer and output layer to facilitate reliable intestinal disease classification. The primary goal is to improve clinical decision support to diagnose gastrointestinal diseases. The paper demonstrates the efficiency of these models in precisely categorizing gastrointestinal diseases through extensive evaluation using the KVASIR Dataset, which includes 4000 different types of intestinal images. The experimental results show that EfficientNetB5 achieved the highest testing accuracy of 92.58% and outperformed all networks trained using the proposed framework.