Early breast cancer diagnosis is crucial for effective therapy. Breast cancer has several subtypes, making it difficult to diagnose, especially in low-resource settings. Deep learning models, particularly Convolutional Neural Networks (CNNs), can help diagnose and categorise breast cancer early. This thesis presents a novel approach to breast cancer diagnosis using deep-near-infrared (CNN) and image clustering. Our objective is to build a cost-effective and efficient deep learning model for breast cancer detection and categorisation to help doctors make fast and accurate diagnoses. The model will use picture clustering to find comparable mammography images from a dataset. After that, deep convolutional neural networks (CNNs) will classify images as benign or cancerous. The precision, sensitivity, specificity, y, and F1 score will be used to evaluate the suggested model. The suggested model will be evaluated using a variety of performance measures and compared to cutting-edge methods. This thesis will show how deep convolutional neural networks can improve the accuracy and reproducibility of breast cancer diagnosis. These findings will be based on the thesis. This study could change the diagnosis of breast cancer and patient survival.