The University of British Columbia

An algorithm for segmentation of digital mammograms is developed based on the fuzzy set idea. The algorithm is intended to be part of a soft tissue abnormality (such as a mass) detection scheme. The algorithm divides a mammogram into different regions according to its intensity information. For those images which contain a lesion with brighter intensity than other patterns present in the mammogram, the algorithm can successfully detect the border of the lesion. The algorithm with four desired levels of segmentation was implemented and applied on a set of 20 mammographic images, each containing a malignant mass, and the algorithm was successful in segmenting each image into 4 regions. It also successfully detected the boundaries of those masses appearing brighter than other regions in each of the mammograms.