Colorectal Cancer Diagnosis with Deep Learning Models

Abstract

The third most common disease in the world, colorectal cancer, frequently has the highest death rate. Surgery is a viable treatment option, but after five years, thirty to forty percent of patients have recurrence. Many people who have effectively treated their colorectal cancer also develop metastatic illness. Early detection is crucial since colorectal cancer has a high fatality rate. Deep learning techniques make colorectal cancer screening timelier and more cost-effective by enabling early and quicker identification of the disease. A collection of cell pictures was employed in the study to detect colorectal cancer. To demonstrate the capability of deep learning approaches, we used Convolutional Neural Networks (CNN), AlexNet, VGG-16, ResNet models and our proposed model as Hybrid CNN-LSTM. The accuracy and loss rates provided by the propos models were compared. The highest accuracy rate performance was observed with from the Hybrid CNN-LSTM model. The highest loss rate performance was observed with from the CNN model.