Microcalcifications detection using Fisher's linear discriminant and breast density

Breast cancer is one of the main causes of death in women. However, its early detection through microcalcifications identification is a powerful tool to save many lives. In this study, we present a supervised microcalcifications detection method based on Fisher's Linear Discriminant. Our method considers knowledge about breast density allowing it to identify microcalcifications even in difficult cases (when there is not high contrast between the microcalcification and the surrounding breast tissue). We evaluated our method with two mammograms databases for each of its phases: breast density classification, microcalcifications segmentation, and false-positive reduction, obtaining cumulative accuracy results around 90% for the microcalcifications detection task.