A Novel Pectoral Muscle Segmentation Algorithm Based on Polyline Fitting and Elastic Thread Approaching

Mammography is one of the most effective detection methods of breast cancer. A novel algorithm was proposed to segment the pectoral muscle automatically. It firstly carried out a series of ROIs upon the neighborhood, and in each ROI, the optimal threshold was computed with the iterative thresholding. After combining these thresholds into an optimal threshold curve and computing the corresponding local mean square deviation curve, the segmentation threshold of the pectoral muscle could be determined. Then, zonal Hough transform and polyline fitting were applied to fit the boundary of the pectoral muscle, and finally an elastic thread algorithm was used to extract the pectoral muscle accurately. This algorithm overcame the defects of the conventional ones, and could get a generally high precision.