Nowadays, breast cancer is considered one of the most threatening and common cancers for women due to the high rate of deaths that occurred yearly that reaches about 25% in all cancers. One of the most keys to decrease the mortal rate caused by breast cancer is its early detection. So, the research on developing computer-aided diagnosis systems (CADs) has been widely increased to improve the accuracy of breast cancer localization and classification. Generally, a proposed CAD is developed through four stages: data preparation and preprocessing, cancer detection, followed by its pathology classification. In this paper, the most recent proposed approaches to detect lesions in the breast mammograms and classify them are discussed with a comparative analysis to list the advantages and the disadvantages of most approaches. The main objective of this paper is to group the CADs with a performance evaluation and detailed analysis in order to furtherly develop others by avoiding the main weak points in the existing systems and to achieve high detection accuracy and classification performance at the same time.