Computer-aided diagnosis: Application of wavelet transform to the detection of clustered microcalcifications in digital mammograms

Abstract

Computer methodologies are being developed to assist radiologists, as second readers, in the interpretation of mammograms. This could represent further amelioration by increasing diagnostic accuracy in the screening programs. We have developed a computerized scheme to detect clustered microcalcifications in digital mammograms, using 100 mammograms that were randomly selected from the mammographic screening program, currently undergoing at the Galicia Community (Spain). After the digitization process, the breast border was initially determined. A wavelet-based algorithm was employed to detect the clusters of microcalcifications. The sensitivity achieved was 79% at a false positive detection rate of 1.83.