Advances in Medical Image Analysis with Vision Transformers: A Comprehensive Review

Understand

The remarkable performance of the Transformer architecture in natural language processing has recently also triggered broad interest in Computer Vision.

• Among other merits, Transformers are witnessed as capable of learning long-range dependencies and spatial correlations, which is a clear advantage over convolutional neural networks (CNNs), which have been the de facto standard in Computer Vision problems so far.
• Thus, Transformers have become an integral part of modern medical image analysis.
• In this review, we provide an encyclopedic review of the applications of Transformers in medical imaging.