2022/04/14 by Mohammad Zalbagi Darestani, Jiayu Liu, Darestani, Mohammad Zalbagi +3 · 3 citations
Engineering · Medicine · Physics and Astronomy · #Advanced MRI Techniques and Applications #Atomic and Subatomic Physics Research #FOS: Electrical engineering #Image and Video Processing (eess.IV) #Sparse and Compressive Sensing Techniques #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2204.07204
openalex publication_date 2022/04/14 · openalex created_date 2022/04/19 · openalex updated_date 2026/07/28
Deep learning based image reconstruction methods outperform traditional methods. However, neural networks suffer from a performance drop when applied to images from a different distribution than the training images. For example, a model trained for reconstructing knees in accelerated magnetic resonance imaging (MRI) does not reconstruct brains well, even though the same network trained on brains reconstructs brains perfectly well. Thus there is a distribution shift performance gap for a given neural network, defined as the difference in performance when training on a distribution P and training on another distribution Q, and evaluating both models on Q. In this work, we propose a domain adaptation method for deep learning based compressive sensing that relies on self-supervision during training paired with test-time training at inference. We show that for four natural distribution shifts, this method essentially closes the distribution shift performance gap for state-of-the-art architectures for accelerated MRI.