Evaluating Machine Learning Techniques for Breast Cancer Detection: A Comprehensive Review
Owen Kresse, Youssef Kamel Kamel Hassan Rezk, Alaaelddin Ibrahim Said, Rana HossamEldin Mousa, Matheus Sampaio Ibrahim, Ahmed Fathy Sweed, Tomas Pegorari, Pablo Nakasato, Ainhoa Osa-Sanchez, Itxasne Del Barrio, Francesc Serra Crespí, Keltse Santisteban Ortiz, Naiara Melián Eguia, Mohamed Elsharkawy, Ibrahim Abdelhalim, Begonya García-Zapirain, Ayman S El-Baz · MEJ Mansoura Engineering Journal · 2025
Breast cancer is considered one of the most common types of cancer among women. significant amount of effort done in early detection to increase survival chance since early detection is a challenging task especially in certain breast cancer conditions or using inefficient imaging modalities AI demonstrated significant potential in breast cancer detection algorithms including convolutional neural networks (CNNs) and Transformers, which have achieved highly accurate results but they have some limitations, such as the large amounts of data required for training as CNNs rely on local features, while Transformers focus on global features However, recent research has proposed hybrid models or modified attention as well as preprocessing techniques and transfer learning to solve this problem resulting in improved outcomes more challenges lie in data availability as there are limited datasets containing sufficient samples to train models effectively and the interpretability of models, as they are often treated as black boxes. Explainable AI is an active area of research in this paper we imaging modalities, disease statistics, publicly available datasets, and experimental AI models in detecting breast cancer including recent models.