Although mammography has proven to be the most reliable and cost‑effective method for the early detection of breast cancer in a screening environment, mammographic interpretation is quite difficult, due to time required, the complexity of tissue patterns, and the low yield. All current CAD schemes use features extracted from images and abnorrnalities at the time when a positive reading was performed. Hence these systems are optimized for the detection of abnorrnalities at the same stage of development as the typical level identified by an experienced reader. In this project we propose to develop and test CAD schemes that focus on optimization of the detection in marnmograms that had been interpreted as negative but the patients were found to have breast cancers on a follow‑up (later) exarnination. This project will not only explore a new approach to detect small and subtle cancers, it may result in great benefit to the patients, since earlier detection of breast aulcer can substantially reduce mortality amd morbidity.
