Breast cancer is the most common cancer which is affecting most of today’s women. As an invasive ductal carcinoma, this cancerous cell may also initiate their origin in the glandular tissue, becoming critical for diagnosis. Early diagnosis of breast cancer can increase the lifetime of women. Mammogram X-ray images are greater support for effective detection of the same. Tumor mass identification is the major challenge when mammogram images are used in a trained model. Moreover, these cancer cells can originate in different breast regions, which is the uncontrolled characteristic of breast cancer. Due to this the density of tumor mass may vary at surrounding regions, so the mammogram data presents clear and unclear mass clouds. Machine learning (ML) models misread the features lead to critical health issue. Deep CNN models overcome this issue and improve the accuracy rate in detection of mass cells. This review presents different ML and DL models used for preprocessing, feature extraction, and classification. This study describes various preprocessing techniques used for removal of artifacts and different segmentation techniques used and also presents comparison of ML and DL models.