Abstract:
The breast cancer is a serious public health problem among women in the world. Mammogram breast X-ray is considered the low cost and most reliable method in early detection of breast cancer. In this thesis an approach is proposed to develop a Computer-Aided Diagnosis (CAD) system that can be very helpful for radiologist in diagnosing.
This work has tried to analyze the texture of mammography images taken from Mini MIAS data base and to find the values of various parameters of texture. Two features types, Haralick’s features based on spatial grey level dependency (SGLD) matrix and based wavelet coefficients are applied for classification of each Regions of Interest (ROIs).
The proposed method for detection of breast cancer on digital mammogram classified the normal breast tissues into three classes which: fat, glandular and dense and then into normal and abnormal classes. The features discriminating to detect abnormal from normal tissues was determined by stepwise linear discriminant analysis classifier (LDA).
This study investigates whether the texture could be used to discriminate among the various breast tissue types. The proposed method focuses on SGLD matrix as parameters for texture analysis which achieved the highest accuracy that 95.7% for classification of breast tissues on digital mammograms. This is an important step in the development of a CAD for mammograms analysis being developed.
