Deep J-Sense: Accelerated MRI Reconstruction via Unrolled Alternating Optimization

TL;DR

Deep J-Sense is a deep learning method for MRI reconstruction that unrolls alternating optimization to simultaneously refine image and coil sensitivity estimates, improving robustness and performance across different scan conditions.

Contribution

It introduces a novel unrolled alternating minimization approach that refines both image and coil sensitivity maps, enhancing robustness over prior methods that depend on calibration data.

Findings

Improves MRI reconstruction quality on knee fastMRI data.

Increases robustness to varying acceleration factors.

Reduces dependence on calibration data.

Abstract

Accelerated multi-coil magnetic resonance imaging reconstruction has seen a substantial recent improvement combining compressed sensing with deep learning. However, most of these methods rely on estimates of the coil sensitivity profiles, or on calibration data for estimating model parameters. Prior work has shown that these methods degrade in performance when the quality of these estimators are poor or when the scan parameters differ from the training conditions. Here we introduce Deep J-Sense as a deep learning approach that builds on unrolled alternating minimization and increases robustness: our algorithm refines both the magnetization (image) kernel and the coil sensitivity maps. Experimental results on a subset of the knee fastMRI dataset show that this increases reconstruction performance and provides a significant degree of robustness to varying acceleration factors and calibration region sizes.