Introduction: Breast cancer is the second type of cancer that threatens women's lives after lung cancer. To avoid negative effects brought about by breast cancer, it is crucial to detect the disease in early stages and protect from it. However, since abnormalities in breasts are intertwined with normal tissues, it is not an easy task to deduce from a simple naked eye examination that there is an apparent abnormality in the breast. To achieve this, computer aided diagnosis (CAD) system assists in improving radiologists' diagnosis. Methods: This paper presents a (CAD) system, which uses new contourlet transform (SFLCT) algorithm, principal component analysis (PCA) and least squares support vector machines (LS-SVM) to classify mammogram images. The SFLCT algorithm is used to feature extraction and PCA is for further dimensional reduction and feature selection. Mammogram images which are obtained from the mammographic image analysis society (MIAS) database are classified into categories as normal abnormal and benign-malignant. Result: Concerning the outcome of the implementation, the system achieved 100% classification accuracy for normal abnormal classification and 98.2457% for malignant-benign classification. Conclusion: It can be concluded that the system provides important support to CAD systems with very encouraging results.