Colorectal Cancer Prediction using Deep Learning

Abstract

Colorectal cancer (CRC) is one of the leading causes of cancer-related death worldwide. Early detection and diagnosis of CRC is critical for successful treatment and improved patient outcomes. The development of deep learning models for CRC prediction has the potential to revolutionize the way that CRC is identified and managed. The purpose of the study is to create a CRC prediction model that can be used in clinical settings to accurately identify colorectal cancer as early as possible. The experiments were conducted using the Fastai library, which provides a high-level interface for deep learning tasks. A pre-trained ResNet-34 model architecture was utilized, which has been proven effective for various image classification tasks. To optimize the model's performance, hyperparameter tuning experiments were conducted. Through the utilization of CNNs and other deep learning architectures, it has been demonstrated that histopathological images can be accurately analyzed to detect colorectal cancer. The evaluation metrics, including accuracy (95.6%), confusion matrix, and ROC curve, demonstrate the model's ability to accurately classify colorectal tissue samples. The findings align with existing literature on colorectal cancer prediction, validating the effectiveness of our proposed model. The study recommends that larger and more diverse datasets must be used to improve the accuracy and reliability of detection systems, larger since they will represent different populations and subtypes of cancer are needed. (Abstract)