Binary and multi-classification models for breast cancer diagnosis using automated deep learning and mammography images with different augmentation cases

Mammography is a very efficient medical imaging procedure that is used to detect and diagnose breast cancer. However, the use of mammography for the early detection and identification of cancer is very complicated and represents a considerable workload for radiologists. Machine learning (ML) can help address these challenges by providing accurate, automated diagnosis, but traditional ML methods are complex and resource-intensive. Google AutoML Vision offers a simplified approach, enabling healthcare professionals with minimal programming skills to develop effective diagnostic models. The aim of this study was to evaluate the ability of automated deep learning using mammography images using Google AutoML with different augmentation cases. In this work, two models were created: one for binary classification and another for multi- classification. The binary classification model includes two scenarios: non-cancerous and malignant, while the multi-classification approach includes three scenarios: normal, benign and malignant. The average accuracy of the two classifications was evaluated and compared. The average accuracy of the binary and multi-classification models was 77.98% and 79.29%, respectively. These results suggest that Google AutoML can simplify the use of ML models in the clinical setting and provide a reliable diagnostic tool that can reduce the workload of radiologists. This study shows that AutoML has the potential to streamline diagnostic workflows in healthcare and make machine learning more accessible and effective in medical practise.