Detection of Mammographic Masses using FRFCM Optimized by PSO

Abstract

Since early detection of breast cancer can effectively reduce the mortality rate, hence, in an attempt, mass, a symptom of breast cancer which is difficult to identify due to its subtle nature, is targeted to locate it efficiently with the proposed detection scheme. This paper introduces FRFCM-PSO, a hybrid model of fast and robust fuzzy c-means clustering (FRFCM) and particle swarm optimization (PSO), for the localization of mammographic masses. FRFCM is an improvised version of FCM by employing morphological reconstruction and member-ship filters which alleviates the necessity of additional local spatial information which burdens the method with computational complexity. Moreover, the general limitation of clustering technique of initializing the center point has been mitigated by incorporating optimization method– PSO. Hence, the combinational approach yields a sensitivity of 96.6 % with 2.29 as false positives per image (FPs/I) when evaluated on the mini-MIAS dataset. Further, the FPs are reduced using feature extraction (LBP) and classification (Ensemble classifier) technique where an Az value of 0.846 is observed with an improvement of 74 % in FPs/I which is further compared with the similar competing scheme.