The researchers are working on various types of abnormalities classification on mammogram images such as breast density, mass segmentation mass classification, and mico-calcification. The selection of a mammogram image dataset is an important step for developing advanced classification, segmentation, and lesion detection techniques using deep learning techniques. So in this manuscript, we summarise some of the public and private mammogram datasets such as MIAS Inbreast, KAU-BCMD, VinDr-Mammo, CBIS-DDSM, and many subsets of mass mammogram patch datasets are available in various repositories. The various properties of mammogram image datasets are discussed in detail of updations and compared them in the context of deep learning (DL) driven computer-aided diagnosis (CAD) for breast cancer.