Comparative Analysis of CNNs and Transformers for Binary Classification of Breast Cancer Mammograms

This paper focuses on a binary classification task for mammography breast cancer detection, aiming to distinguish between cancer-positive and cancer-negative cases. Our work involves exploring various existing models in deep learning such as Convolutional Neural Networks (CNN) models and Transformer models that can cater to better detection of breast cancer. Comparisons among these models are obtained concerning their performance using evaluation metrics such as accuracy, recall, precision, and F1-score. VGG-19 and CaiT models proved to perform better with 0.6893 and 0.6085 F1-scores, respectively. Additionally, we dealt with the challenges encountered in the breast cancer detection dataset considered for the task. By examining these methods and their results, our objective is to offer perspectives on enhancing the effectiveness and precision of breast cancer detection techniques.