Ensemble of Handcrafted and Learned Features for Colorectal Cancer Classification

Colorectal cancer (CRC) remains one of the most common and lethal malignancies worldwide. The current gold standard for CRC diagnosis relies on histopathological analysis, a time-consuming process subject to inter-observer variability and dependent on expert experience. While convolutional neural networks (CNNs) have achieved remarkable success in medical image analysis, they often require large annotated datasets and lack interpretability. Traditional handcrafted texture descriptors, on the other hand, provide domain-specific insights but may fall short in capturing complex patterns. To address these limitations, we propose a novel ensemble approach that integrates handcrafted texture descriptors with deep learning-based features extracted from CNNs. Our method leverages the complementary strengths of both feature types, resulting in a more robust and discriminative feature space. Experimental evaluations demonstrate that our ensemble approach outperforms state-of-the-art methods across various metrics, achieving an accuracy of 99.20% by combining color textures with deep learning features. This study underscores the potential of integrating traditional and modern techniques to advance medical image analysis, presenting a significant step forward in automated CRC classification and fostering advancements in medical computing and image processing.