Efficient Deep Learning Approach for Breast Cancer Detection

Breast cancer detection saves lives by enabling early diagnosis, timely intervention, and treatment. Early detection and treatment reduce patients' mortality rates and reduce the overall burden of the disease on affected individuals and healthcare systems. A scientific research challenge for breast cancer detection is achieving as high accuracy as possible. Machine learning is providing greater hope for achieving high accuracy. This paper proposes a high-accuracy breast cancer detection method utilizing feature correlation and a reconfigurable neural network (RNN). The proposed method has been comprehensively tested using breast cancer detection datasets and is compared against Support Vector Machine (SVM) and K-Nearest Neighbors (KNN) methods. The results demonstrate the superior performance of the proposed method, achieving an accuracy of 97.2% and sensitivity of 97.8 %, surpassing known traditional methods. The proposed method is implemented using VHDL on Xilinx Virtex-7 FPGA and incurs a lower power consumption of 0.984 W with acceptable resource utilization.