One of the most commonly diagnosed cancers at a global level is breast cancer having a significant contribution to cancer-related deaths among women. Breast cancer has a lower survival ratio so early detection is essential. A significant improvement has been observed in the early detection and diagnosis of breast cancer with the help of advanced deep learning algorithms. This work suggests a better method for classifying breast cancer using a breast cancer convolutional neural network (BC-CNN), designed to enhance diagnostic precision and trained on the breast cancer dataset. When working with the mentioned dataset, preprocessing techniques such as image scaling are applied to ensure uniformity and optimize the learning process. Using the preprocessed dataset, the BC-CNN uses a method to categorize breast cancer patients into four groups: normal, benign, invasive, and insitu. The primary objective of this methodology is to improve the accuracy and reliability of the diagnosis. Allowing timely medical interventions and more tailored treatment plans based on specific stages of the disease. Compared to traditional methods, including capsule networks, the BC-CNN demonstrates superior classification performance. Utilizing deep learning’s capabilities in extracting features and recognizing intricate patterns, the proposed method can potentially reduce false negatives and provide more reliable outcomes, thus refining the diagnosis process and aiding in the planning of effective treatments. The accuracy of the suggested method highlights its efficacy in clinical applications.