Collected sources and patterns will appear here. Add from search, explore, or the patterns library.
UndersampledKSpace -> ReconstructedImage
Reconstruct clinical-grade magnetic resonance images from highly undersampled acquisitions using a foundation model trained across heterogeneous field strengths and scanner configurations.
Problem it solves
Reconstruction pipelines overfit to specific scanner manufacturers, acquisition sequences, and field strengths, requiring expensive site-specific fine-tuning.
Consumes
Emits
The real projects this mechanism was found in. Attribution is the point — this is how the best teams actually do it.