Deep Learning-Based Diagnostic Framework for Colorectal Cancer Using Histopathological Images
Ashif Al Nayem Zeesan, Sadri Islam, Abrar Fahim, Md. Ahasanul Adib, Md. Touhidul Islam, Fardin Sabahat Khan · 2025
Colon cancer is one of the most fatal cancers globally among both males and females, highlighting the urgent need for effective early detection methods to improve survival rates. Histopathological image analysis plays a critical role in identifying malignancy by examining cellular patterns, but traditional diagnostic methods can be time-consuming and resource-intensive, necessitating more efficient solutions. This study focuses on the colon-specific subset of the LC25000 dataset, a collection of high-resolution histopathological images, to develop a deep learning-based diagnostic framework using the MobileNetV2 architecture. A unique preprocessing technique including key steps like noise reduction, grayscale conversion, and dimensional standardization was employed to enhance image quality for model training. For colon cancer detection, the model achieved an overall accuracy of 99.95% with 100% recall, precision, and F1-scores. This architecture is designed to reduce computational power and processing time, which makes it highly suitable for mass utilization and resource-limited conditions. A comparative performance analysis was done with existing works to highlight the model's effectiveness in detecting colon cancer. These findings present the impact of the proposed work to develop a more reliable solution for colon cancer diagnosis.