Review on self supervised learning in medical image analysis

Nowadays, self-supervised learning is a popular method for analysing medical images since it annotates the unstructured data provided and uses these self-generated data labels as a foundation for subsequent model training rounds. This review paper is more focused on the recent research done in medical imaging using self-supervised learning, the major challenges of medical imaging, and the challenges of choosing the proper pretext task and data augmentation while using self-supervised learning. Finally, we address potential approaches for future studies and highlight concerns to consider while developing new self-supervised learning concepts and methodologies.