Pectoral muscle segmentation in mammograms based on homogenous texture and intensity deviation

In this paper, a novel method is proposed to segment the pectoral muscle in mammograms. First two anatomical features of the pectoral muscle, homogeneous texture and high intensity deviation are employed to identify the initial pectoral muscle edge. Then Kalman filter is used to refine the ragged initial edge. The proposed method is tested on Mammographic Image Analysis Society Mini-Mammographic (mini-MIAS) database and Digital Database for Screening Mammography (DDSM) database. The acceptable rate is 90.06% and 92% for the mini-MIAS database and the DDSM database, respectively.

► High intensity deviation is combined with homogeneous texture to increase the robustness of pectoral muscle segmentation. ► An accelerated displacement curve is established to model the edge of the pectoral muscle. ► Kalman state filter is introduced to refine the initial edge, leading to high precision segmentation.