Enhancement Techniques on Deep Learning-based Mammography Classification for Breast Cancer Detection

Breast cancer is a significant health concern worldwide, and early detection is crucial in improving patient outcomes. In this study, we aimed to develop an advanced mammography classification system using deep learning (DL) techniques and image processing methods to enhance the accuracy of breast cancer detection. Several DL models, including InceptionResNetV2, InceptionV3, DenseNet169, MobileNetV2, VGG16, ResNet101, VGG19, and ResNet50, were evaluated, along with image processing methods such as Adaptive Contrast Smoothing, gamma sharpening, and Adaptive Gamma Sharpening. Our research demonstrated that implementing image processing methods, particularly Adaptive Gamma Sharpening, significantly improved the classification accuracy across multiple models. Notably, when combined with the Adaptive Gamma Sharpening technique, the VGG16 model achieved the highest accuracy of 98.07%. The evaluation using the Area Under the Receiver Operating Characteristic curve (AUC-ROC) metric revealed that the Adaptive Gamma Sharpening method effectively discriminated between cancerous and non-cancerous cases, with the highest AUC-ROC of 9S.06% observed in the VGG16 model. By leveraging the combination of DL techniques and image processing methods, this research successfully enhanced the performance of mammography classification models in detecting breast cancer. These findings support the potential of image processing methods, particularly Adaptive Gamma Sharpening, in improving the accuracy and reliability of breast cancer detection.