Review on feature extraction methods in computer-aided detection systems on mammogram images

Breast cancer is among the main causes of cancer-related deaths worldwide. Early detection and accurate diagnosis are essential in getting proper treatment. Mammographic imaging that is widely used in screening, often suffers from low contrast and subtle lesion visibility that becomes a challenge for manual interpretation. Feature extraction method plays a significant role in enhancing computer-aided diagnosis by isolating meaningful patterns from mammogram images. This review investigates various feature extraction approaches categorized into statistical, transform-based, keypoint-based, and learning-based methods. There is no single method consistently outperforming others, but combining method often enhances performance. This review highlights the diversity of approaches and concludes that the choice of method should consider dataset characteristics, computational cost, and clinical requirements, while future work should focus on hybridization, interpretability, and dataset standardization.