Microcalcification Detection in Mammograms Using Interval Type-2 Fuzzy Logic System

Abstract

Breast cancer is an important deleterious disease. Mortality rate from this cancer is effectively high and rapidly increasing. The detection at the earlier state can help to reduce the mortality rate. In this paper, we develop a system that helps radiologists to detect microcalcification in mammograms. In particular, we apply the interval type-2 fuzzy logic system with four features, i.e., B-descriptor, D-descriptor, average intensity inside boundary, and intensity difference between inside and outside boundaries. We also compare the result with the result from a type-1 Mamdani fuzzy inference system with the same set of features. The result from the type-1 fuzzy logic system yields 87.95% correct classification with 11.33 false positives per image whereas interval type-2 fuzzy logic system provides 90.36% correct classification with only 4.73 false positives per image.