Detection of Subtle Microcalcifications in Digital Mammograms

Abstract

The work reported in this paper aims to contribute to systems which provide computer assistance to radiologists reading screening mammograms. The goal of such a system is not to replace the radiologist but to augment the information available on which decisions are based. Accordingly, the emphasis is on aspects of reading mammograms which pose difficulties for the radiologist. In particular, this paper is concerned with detecting microcalcifications which are of low contrast, small size and which are approximately spherical in shape. Of course these are not the only type of calcification, but we conjecture that in order to improve the performance of detection schemes, separate methods must be tuned for detecting different classes of calcification.