<![CDATA[Breast cancer is the leading cause of death for women worldwide, and it is predicted to be an important factor in public health. Therefore, early and accurate detection is crucial to enhancing survival rates. Recently, Magnetic Resonance Imaging (MRI) has become a superior option to biopsies due to its exceptional soft tissue imaging capabilities, making it highly effective for detecting and monitoring breast cancer. However, it requires a competent radiologist to perform the procedure. The researchers introduce an approach for breast cancer detection and classification that employs Graph Convolutional Networks (GCNs) to distinguish breast MRI images. The combination Dual-Tree Discrete Wavelet Transform (DTDWT) with GCNs enhances feature extraction, while the Gray-Level Co-Occurrence Matrix (GLCM) identifies texture patterns distinguishing normal, benign, and malignant tissues. The research also employs t-Distributed Stochastic Neighbor Embedding (t-SNE) for dimensionality reduction, improving pattern interpretation. This approach classifies the four breast cancer types using a dataset comprising 200 Dynamic Contrast-Enhanced (DCE)-MRI images from Radiopaedia, allocated as 160 training and 40 validation instances in categories including ductal carcinoma, lipoma, triplenegative breast cancer, and inflammatory breast cancer. A comparative analysis confirms the validity of the approach, which is the first to address these four categories in MRI. The experimental results indicate significant improvements, achieving an accuracy of 0.9821 in classifying breast tumors as benign or malignant, thereby establishing a new diagnostic standard.]]>
