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/

Zendy Plus

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

Zendy Tools

Zendy Open

Reconstructing MRI from highly undersampled measurements is crucial foraccelerating medical imaging, but is challenging due to the ill-posedness ofthe inverse problem. While supervised deep learning (DL) approaches have shownremarkable success, they traditionally rely on fully-sampled ground truth (GT)images, which are expensive or impossible to obtain in real scenarios. Thisproblem has created a recent surge in interest in self-supervised learningmethods that do not require GT. Although recent methods are now fastapproaching "oracle" supervised performance, the lack of systematic comparisonand standard experimental setups are hindering targeted methodological researchand precluding widespread trustworthy industry adoption. We present SSIBench, amodular and flexible comparison framework to unify and thoroughly benchmarkSelf-Supervised Imaging methods (SSI) without GT. We evaluate 18 methods across4 realistic MRI scenarios on real data, showing a wide performance landscapewhose method ranking differs across scenarios and metrics, exposing the needfor further SSI research. Our insights also show how complementary methodscould be compounded for future improvements, exemplified by a novel loss wepropose, Multi-Operator Equivariant Imaging. To accelerate reproducibleresearch and lower the barrier to entry, we provide the extensible benchmarkand open-source reimplementations of all methods athttps://andrewwango.github.io/ssibench, allowing researchers to rapidly andfairly contribute and evaluate new methods on the standardised setup forpotential leaderboard ranking, or benchmark existing methods on customdatasets, forward operators, or models, unlocking the application of SSI toother valuable GT free domains such as 4D MRI and other nascent scientificimaging modalities.

Benchmarking Self-Supervised Learning Methods for Accelerated MRI Reconstruction