Joint Supervised and Self-Supervised Training with Acquisition-Robust Techniques for Accelerated 4D Flow MRI Reconstruction

Mengyuan Xue; Bochun Mei

eess.IVSep 29, 2026

Joint Supervised and Self-Supervised Training with Acquisition-Robust Techniques for Accelerated 4D Flow MRI Reconstruction

Authors: Mengyuan Xue, Bochun Mei

Organizations: Department of Radiology and Biomedical Imaging, University of California, San Francisco, San Francisco, CA, USA

Abstract

4D flow MRI measures time-resolved, three-directional blood velocity but requires long acquisition times, and its diagnostic signal is carried by the phase difference \emph{between} velocity encodings, not by image magnitude. Recent work has developed a per-encoding variational network to address image reconstruction in this field. In this work, we incorporate a joint supervised and self-supervised training regime and utilize both magnitude and velocity data during supervision. At the same time, we add multiple acquisition-robust and conditioning strategies based on the acceleration factors. On the CMRx4DFlow~2026 aortic dataset (1.5 and 3T), our model lowers RelErr by $38$ -- $50\%$ and AngErr by $7.0$ -- $8.7^\circ$ against a training-matched baseline across $R{=}10$ -- $50$ , improving on every held-out subject at every acceleration. Our model also shows strong generalization ability to transfer on out-of-distribution data by employing the joint training scheme, with an increase of $7.2\%$ in SSIM and decrease of $38\%$ and $31\%$ in AngErr and RelErr respectively.

Figures & tables

Panel of Figure 1 — Figure 1: Method overview. Each velocity encoding $k_{j}$ passes through the same network $f_{\theta}$ with $8$ unrolled stages in its own forward.

Figure 2: Qualitative reconstruction comparisons of four approaches at $R{=}30$ for a held-out subject of a peak-systole frame. ROI of mask was labeled on each panel.

Figure 3: Local held-out accuracy vs. acceleration

R

of different approaches. Faint points are values on individual held-out subjects; solid lines their per-

R

means.

Generalization (frozen model, zero-shot)	SSIM $\uparrow$	nRMSE $\downarrow$	RelErr $\downarrow$	AngErr ( ∘ ) $\downarrow$
Cross-site, proposed	0.942	0.084	0.467	28.98
Supervised twin (no joint objective)	0.938	0.086	0.488	29.39
FlowVN baseline	0.879	0.123	0.677	46.67
Cross-organ mean , proposed	0.966	0.052	0.398	29.51
Carotid	0.961	0.055	0.247	18.83
Cerebrovascular	0.950	0.046	0.470	32.81

Table 1: Zero-shot generalization of different approaches on two out-of-distribution datasets: cross-site (an unseen site and scanner model) and cross-organ (non-aortic, categorized by organ).

Configuration	SSIM $\uparrow$	RelErr $\downarrow$	AngErr mean $\downarrow$	AngErr R50 $\downarrow$
Baseline + bounded momentum	0.951	0.552	36.25 ∘	43.87 ∘
+ acceleration conditioning	0.955	0.509	34.00 ∘	43.96 ∘
+ acquisition-robust techniques	0.956	0.507	34.18 ∘	43.40 ∘
+ dual-domain $\mathcal{L}_{\mathrm{ksp}}$ ( 6 )	0.946	0.501	33.04 ∘	43.50 ∘
+ JSSL and velocity losses	0.959	0.376	28.18 ∘	35.01 ∘

Table 2: Ablation studies. Step sizes, optimizer and scheduler are fixed across configurations. Runs are single-seed (seed 42).

Explore similar work

Self-Auditing Residual Drifting for Pathology-Preserving Accelerated Knee MRI
Jul 2, 2026Qing Lyu, Jianxu Wang, Mohammad Kawas +2Uncertainty Quantification Image Reconstruction
Department of Radiology & Biomedical Imaging, Yale School of Medicine, New Haven, CT 06510 USA · Department of Biomedical Engineering, Rensselaer Polytechnic Institute, Troy, NY 12180 USA
MPFlow: Multi-modal Posterior-Guided Flow Matching for Zero-Shot MRI Reconstruction
Mar 4, 2026Seunghoi Kim, Chen Jin, Henry F. J. Tregidgo +2Cross-Modal Alignment Multimodal Pretraining
University College London, UK · Centre for AI, DS&AI, AstraZeneca, UK
Next-Acceleration-Scale Prediction for Autoregressive MRI Reconstruction
May 19, 2026Yilmaz Korkmaz, Vishal M. PatelVisual Autoregressive Models Computational Imaging
Johns Hopkins University, Baltimore MD 21218, USA

Jul 2, 2026, eess.IVJ/K move · Enter open · S save

Self-Auditing Residual Drifting for Pathology-Preserving Accelerated Knee MRI

Qing Lyu, Jianxu Wang, Mohammad Kawas, Ge Wang+1

Department of Radiology & Biomedical Imaging, Yale School of Medicine, New Haven, CT 06510 USA · Department of Biomedical Engineering, Rensselaer Polytechnic Institute, Troy, NY 12180 USA

Uncertainty Quantification Image Reconstruction Accelerated MRI Reconstruction

Open paper AskPDF arXiv

Accelerated magnetic resonance imaging reduces acquisition time, but reconstruction from undersampled k-space can blur diagnostically relevant structures or introduce failures that are not captured by global image metrics. We propose SA-RDM-DC, a Self-Auditing Residual generative Drifting Model with Data Consistency for accelerated knee MRI. The method adapts the newly proposed generative drifting paradigm to accelerated MRI by training a physics-conditioned drift field from the zero-filled reconstruction toward the fully sampled residual correction. It predicts image- and missing-k-space residual corrections, enforces data consistency with acquired k-space, uses frequency-aware and residual drifting supervision to recover fine detail, and produces dense error maps and slice-level risk scores in the same inference pass. We evaluate SA-RDM-DC on multi-coil fastMRI knee data at acceleration factors of 4, 8, and 12, with fastMRI+ pathology annotations for region-level and classifier-based task preservation, and on SKM-TEA for zero-shot and fine-tuned protocol-shift evaluation. Compared with zero-filled reconstruction, UNet-image-SENSE, DC-UNet, Score-Diffusion, ELF-Diff, SENSE-VarNet, and MoDL baselines, SA-RDM-DC achieves the highest SSIM across fastMRI acceleration factors while retaining subsecond per-slice inference and avoiding the long sampling time of iterative diffusion baselines. In pathology-aware analysis, SA-RDM-DC preserves lesion-region structural fidelity and reduces meniscus prediction instability. Its self-auditing scores strongly identify high-error reconstructions on fastMRI and partially transfer as a selective-review signal under SKM-TEA protocol shift. These results support reconstruction evaluation that jointly considers image fidelity, pathology preservation, runtime, and case-specific reliability.

Abstract

Explore similar work

Self-Auditing Residual Drifting for Pathology-Preserving Accelerated Knee MRI

MPFlow: Multi-modal Posterior-Guided Flow Matching for Zero-Shot MRI Reconstruction

Next-Acceleration-Scale Prediction for Autoregressive MRI Reconstruction