After cardiovascular issues, breast cancer is the second most lethal disease. Breast imaging is essential for early detection to improve survival rates. Computer-aided diagnosis (CAD) systems assist radiologists with accurate decision-making, faster diagnosis, and effective treatment. This study uses four feature extraction approaches: kinetic, convolutional neural network deep learning features (CNN), gray level co-occurrence matrix (GLCM), and their combination to improve breast cancer detection, benefiting from advances in computer technology. This work focuses on developing Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) processing algorithms, incorporating advanced pre-processing, feature extraction, and feature selection to distinguish normal and abnormal breast scans. For feature selection, ANOVA was used for the feature's dimensionality reduction. Sets, effectively decreasing the number of descriptors per slice. Features were classified with SVM into normal and abnormal scans. Integrating machine learning-based CAD systems enhance diagnostic accuracy, achieving 97.5% accuracy when combining features in post-contrast 2, which may reduce human errors in the diagnosing procedure.