English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Although deep learning (DL) has gained much popularity for acceleratedmagnetic resonance imaging (MRI), recent studies have shown that DL-based MRIreconstruction models could be oversensitive to tiny input perturbations (thatare called 'adversarial perturbations'), which cause unstable, low-qualityreconstructed images. This raises the question of how to design robust DLmethods for MRI reconstruction. To address this problem, we propose a novelimage reconstruction framework, termed SMOOTHED UNROLLING (SMUG), whichadvances a deep unrolling-based MRI reconstruction model using a randomizedsmoothing (RS)-based robust learning operation. RS, which improves thetolerance of a model against input noises, has been widely used in the designof adversarial defense for image classification. Yet, we find that theconventional design that applies RS to the entire DL process is ineffective forMRI reconstruction. We show that SMUG addresses the above issue by customizingthe RS operation based on the unrolling architecture of the DL-based MRIreconstruction model. Compared to the vanilla RS approach and several variantsof SMUG, we show that SMUG improves the robustness of MRI reconstruction withrespect to a diverse set of perturbation sources, including perturbations tothe input measurements, different measurement sampling rates, and differentunrolling steps. Code for SMUG will be available athttps://github.com/LGM70/SMUG.

SMUG: Towards robust MRI reconstruction by smoothed unrolling