Comparative Study of Deep Learning Architectures for Accurate Colon Cancer Diagnosis
DOI : 10.46335/IJIES.2024.4.5.19
Authors : Jitendra P Patil , Tushar H Jaware, Ravindra D Badgujar, Mahesh B Dembrani
Abstract – This study addresses the pressing public health concern of colon cancer by examining performance of deep learning algorithms in its classification. Utilizing a diverse dataset and five key models—EfficientNet, ResNet, MobileNet, VGG16, and YOLOv5 small—the research aims to evaluate the suitability of these architectures and conduct a comparative analysis of their performance metrics. The primary objectives include contributing valuable insights to medical image analysis and aiding in the development of accurate colon cancer diagnostic tools. Through careful curation of a comprehensive dataset and meticulous preprocessing, the study ensures data quality for training and testing. Each deep learning model is configured with specific architectures, hyperparameters, and undergoes training with optimizers, fine-tuned learning rates, and data augmentation techniques. Quantitative evaluation metrics, provide a robust measurement of model performance.
