Cost Efficient Mammogram Segmentation and Classification with NeuroMem® Chip for Breast Cancer Detection

In this paper, a Computer Aided Diagnosis system to detect and classify anomalies on mammograms is proposed. A segmentation method for anomaly extraction has been proposed using the NeuroMem® Chip NM500 which integrates physical neural networks, up to 83% of the anomalies were detected. We configured two subnetworks for the mammogram classification step the accuracy reached 87%.