Risk detection of breast cancer from MRI images using deep feature extraction and fine-tuning NCA feature selection

Breast cancer is the most common cancer among women and presents a serious health risk because of its complex nature. It results from the uncontrolled growth of abnormal cells in breast tissue, often appearing as lumps or visible changes. Early detection is crucial for lowering risk and improving treatment success. Advances in imaging, especially Magnetic Resonance Imaging (MRI), have improved diagnostic accuracy. When combined with Artificial Intelligence (AI), MRI becomes a powerful tool for risk assessment. AI models trained on large MRI datasets can detect tumors with high precision. This study aims to enhance breast cancer risk detection by combining Fine-Tuning Neighborhood Component Analysis (FTNCA) feature selection and DenseNet-201, NasNetMobile, ResNet-101, and Xception architectures. A dataset of 1,480 MRI images was categorized into benign and malignant cases, with the hybrid model reaching 99.77% accuracy. These findings emphasize the method's effectiveness in identifying cancer risk from breast MRI.

Department

Computer Science

Publisher

Taylor & Francis

Sponsor

na