Colon Cancer Classification from Histopathological Images Using Convolutional Neural Networks

Abstract

Colon cancer is one of the major causes of cancer-related deaths, and early diagnosis is essential for effective treatment. Traditional histopathological image analysis relies on expert pathologists and is often time-consuming and subjective. This paper proposes an automated colon cancer classification system using Convolutional Neural Networks (CNNs) to improve diagnostic efficiency and accuracy. Histopathological images from the LS25000 colon dataset are pre-processed using resizing, normalization, and data augmentation techniques such as rotation and flipping. The CNN model automatically extracts relevant features and classifies images into normal, benign, and malignant categories. Model performance is evaluated using standard metrics including accuracy, precision, recall, F1-score, and confusion matrix. In addition, Grad-CAM visualization is employed to highlight important image regions influencing the model’s predictions, enhancing interpretability. Experimental results demonstrate that the proposed approach provides accurate and consistent classification, making it a reliable decision-support tool for assisting pathologists in early colon cancer detection.