Automated Detection of Clustered Microcalcifications on Digitized Mammograms

Abstract

We have been developing an automated detection scheme for clustered microcalcifications on digital mammograms and reported the methods in several papers [1]–[4]. These schemes show a good performance in detection, but there is a problem that many false-positive candidates (ten and more) appear in some specific images. Therefore, an improvement of the elimination step of false positives is required. To achieve this, we have developed new methods of discrimination of the candidates and elimination of the false positives.