The accurate diagnosis of breast cancer through mammography significantly depends on image quality. In this study, a new deep learning model is developed to enhance the quality of mammogram images. Our approach involved modifying the established visual geometry group (VGG) neural network architecture to specifically address the challenges to denoise the mammogram images. Unlike the standard VGG network, our model includes additional convolutional layers and employs advanced noise reduction techniques designed for mammography. These modifications enable the model to effectively reduce image noise while preserving essential diagnostic details. The proposed VGG inspired CNN model had achieved a high PSNR of 79, surpassing existing techniques and showing potential for improved breast cancer diagnosis.