Optimized Prediction and Classification of Celiac Disease in Biopsy Images using Transfer Learning

A biopsy image analysis is necessary for a timely and accurate diagnosis of celiac disease (CD), a chronic autoimmune disease. This research introduces a refined method for predicting and categorizing CD by integrating transfer learning with advanced optimization strategies to improve diagnostic accuracy. The proposed approach follows a structured optimization sequence, beginning with Enhanced Cuttlefish Optimization (ECO) to minimize noise, followed by Improved Whale Optimization (IWO) to extract key texture and shaperelated features, and concluding with Modified Crow Search (MCS) for selecting the most relevant features. For classification, these refined characteristics are subsequently fed into a convolutional neural network (CNN) that uses transfer learning. According to experimental results, this approach performs better than traditional methods in terms of accuracy, precision, recall, and F1-score, making it a solid framework for clinical applications involving early CD identification and classification.