Fast MRI for All: Bridging Access Gaps by Training without Raw Data

TL;DR

This work tackles the barrier to deploying physics-driven DL MRI reconstructions in settings lacking access to raw k-space data. It introduces CUPID, a compressibility-inspired unsupervised learning framework that combines a weighted $\ell_1$-based compressibility loss with a parallel-imaging fidelity term constructed via perturbation equivariance, enabling training solely on routine clinical DICOM images. CUPID achieves reconstruction quality on par with fully supervised and self-supervised methods that require raw data, while outperforming conventional CS and diffusion priors, and it demonstrates effective zero-shot and prospective performance. By removing the raw-data bottleneck, CUPID offers a practical pathway to broader access to fast MRI in rural and under-resourced populations, potentially reducing wait times and improving diagnostic throughput.

Abstract

Physics-driven deep learning (PD-DL) approaches have become popular for improved reconstruction of fast magnetic resonance imaging (MRI) scans. Though PD-DL offers higher acceleration rates than existing clinical fast MRI techniques, their use has been limited outside specialized MRI centers. A key challenge is generalization to rare pathologies or different populations, noted in multiple studies, with fine-tuning on target populations suggested for improvement. However, current approaches for PD-DL training require access to raw k-space measurements, which is typically only available at specialized MRI centers that have research agreements for such data access. This is especially an issue for rural and under-resourced areas, where commercial MRI scanners only provide access to a final reconstructed image. To tackle these challenges, we propose Compressibility-inspired Unsupervised Learning via Parallel Imaging Fidelity (CUPID) for high-quality PD-DL training using only routine clinical reconstructed images exported from an MRI scanner. CUPID evaluates output quality with a compressibility-based approach while ensuring that the output stays consistent with the clinical parallel imaging reconstruction through well-designed perturbations. Our results show CUPID achieves similar quality to established PD-DL training that requires k-space data while outperforming compressed sensing (CS) and diffusion-based generative methods. We further demonstrate its effectiveness in a zero-shot training setup for retrospectively and prospectively sub-sampled acquisitions, attesting to its minimal training burden. As an approach that radically deviates from existing strategies, CUPID presents an opportunity to provide broader access to fast MRI for remote and rural populations in an attempt to reduce the obstacles associated with this expensive imaging modality.

Figures (14)

• Figure 1: Many regions lack direct MRI access or rely on local/mobile units: (a) Over half of MRI services in Minnesota are in non-urban areas burdorf2022comparing, and (b) these scanners often lack vendor agreements for raw data access, limiting AI fine-tuning.
• Figure 2: Our Compressibility-inspired Unsupervised Learning via Parallel Imaging Fidelity (CUPID) method trains PD-DL models in an unsupervised and/or zero-shot manner without requiring any raw k-space data. The network is unrolled for $T$ units, with each unit consisting of regularizer (R) and data fidelity (DF). The proposed loss function comprises two terms: (a) a reweighted $\ell_1$ component that assesses the compressibility of the network's output; (b) a fidelity term that ensures the output stays consistent with parallel imaging reconstructions via carefully designed perturbations, thereby preventing the network from producing a sparse all-zeros output.
• Figure 3: Representative slices reconstructed at an acceleration factor of $R = 4$ using equidistant undersampling from coronal PD and coronal PD-FS knee MRI, as well as axial T2-weighted brain MRI. The baseline CG-SENSE, CS, EI-trained PD-DL, and DDS suffer from residual artifacts highlighted by red arrows. PD-DL trained with CUPID improves upon them while delivering reconstruction quality comparable to supervised and SSDU-trained PD-DL.
• Figure 4: Representative subject-specific/zero-shot learning results for various algorithms on coronal PD and coronal PD-FS knee MRI for retrospective $R=4$ equidistant undersampling, along with results from database-trained diffusion models on knee data. Baseline CG-SENSE, CS, ScoreMRI and DDS suffer from residual artifacts (red arrows) and blurring. PD-DL with CUPID loss successfully removes these artifacts, and functions in a similar manner to ZS-SSDU.
• Figure 5: Prospective acceleration results for various methods that operate on parallel imaging-reconstructed DICOMs exported from the scanner. (a) Vendor-provided PI DICOM ($R=4$), (b) CS, (c) DDS, and (d) CUPID.