Breast cancer is the second most common cause of mortality for women. Early detection and classification of breast cancer is a crucial initial step in its therapy. Different screening methods like MRIs, ultrasounds, mammograms, computed tomography etc. are used to obtain breast images. Because of its capacity to process vast volumes of data, deep learning (DL), a branch of machine learning (ML), has demonstrated impressive outcomes in a number of domains, most notably the biomedical sector. However, the current deep learning-based breast categorization models have challenges due to the absence of substantial data collection. In order to expand the quantity of images, the proposed method uses a customized generative adversarial network (Cust-GAN) for data augmentation. Additionally, to enhance image quality and remove noise, employ adaptive bilateral filters with weight (ABFW) for image pre-processing. © 2025 Elsevier B.V., All rights reserved.