Instantly Enhanced Image Selection for Mammogram Images Segmentation

Abstract

The proposed system is to design a CAD (Computer-aided diagnosis) system that would achieve higher precision of breast cancer masses diagnosis through image preprocessing step. Hybrid approach is proposed to filter and enhance mammogram images. Tow strong enhancement techniques are used, they are “contract limit adaptive histogram equalization” (CLAHE) and “Dynamic histogram equalization” DHE. When there are more than one technique nearly similar in effectiveness are applied on an image and their efficiency is varied at each enhancing case, the instant selection between those resulted images is the best-proposed way to passing the better image to the next step in the classification system; that is based on mathematical calculations. in order to achieve best results in terms of mean square error (MSE) and peak signal to noise ratio (PSNR), these calculations require an instantaneous collaboration between the implementing program (Matlab) and the memory that used for storing and comparing the calculated values. Instantly image selection provides an effective way to adopt the best enhanced image between the outputs of the applied techniques to be passing to segmentation stage. Seeded region growing (SRG) segmentation method and morphological operations are proposed for the purpose of eliminate labels and pectoral muscle part from the image. Extracting the suspected regions from the breast area is done using graph cut segmentation method.