A Review of Predictive and Contrastive Self-supervised Learning for Medical Images

arXiv:2302.05043v2 Announce Type: replace
Abstract: Over the last decade, supervised deep learning on manually annotated big data has been progressing significantly on computer vision tasks. But the application of deep learning in medical image analysis was limited by the scarcity of high-quality annotated medical imaging data. An emerging solution is self-supervised learning (SSL), among which contrastive SSL is the most successful approach to rivalling or outperforming supervised learning. This review investigates several state-of-the-art contrastive SSL algorithms originally on natural images …

abstract analysis application arxiv big big data computer computer vision cs.cv data deep learning image images imaging medical medical imaging predictive quality review self-supervised learning solution supervised learning tasks type vision