A Novel Pectoral Removal Approach in Digital Mammogram Segmentation and USRR Feature Selection

Detection of micro calcification based on textural image segmentation and classification is the most effective early-diagnosis of breast cancer. The aim of segmentation is to extract a breast region by estimation of a breast skin-line and a pectoral muscle as well as removing radiographic artifacts and the background of the mammogram. The intensity value is taken using histogram function. After detecting the region intensity value from mammogram, the edge between pectoral region and breast region will be detected using Fuzzy Connected Component Labeling. Our proposed method worked good for separating pectoral region by eliminating cancer area. The raster scan method is used for fixing the pectoral removal area in the original image. After removal pectoral muscle from the mammogram, so that further processing is confined to the breast region alone. Two error measures were used to compare these three methods performance. One of the measures is Mean absolute error (MAE) and another one is hausdroff distance measure to find the distance between the binary pectoral region and binary breast region alone. We applied unsupervised feature selection of rough set based Relative Reduct algorithms and it demonstrates effectively remove the redundant features. The quality of the reduced data is measured by the classification performance and it is evaluated using WEKA classifier tool.