Breast cancer is the second-leading cause of cancer worldwide. Researchers have utilized machine learning classifiers on FPGA to increase the effectiveness of breast cancer diagnosis. The proposed approach features an optimized real-time breast cancer detection system utilizing a Support Vector Machine (SVM) linear classifier for point of care. To minimize hardware complexity, this method incorporates a wrapper-based feature selection technique, identifying the most significant features, and utilizes an approximated adder design. The system achieves an accuracy of 96% while employing a reduced feature subset. FPGA implementation demonstrates a 14% reduction in hardware requirements compared to existing methods. For a single-chip solution in 45nm technology, the system achieves an area of 63,903 μm2, power consumption of 43 mW, and a delay of 6.6 ns. Overall, the proposed system integrates high accuracy with reduced hardware requirements, enabling efficient real-time diagnostics in resource-constrained environments with a target of point-of-care diagnostic tools.