Consistency Models for Fast MRI Reconstruction Using Regularization by Denoising

Merve Gülle; Junno Yun; Yaşar Utku Alçalar; Mehmet Akçakaya

eess.IVAug 20, 2026

Consistency Models for Fast MRI Reconstruction Using Regularization by Denoising

Authors: Merve Gülle, Junno Yun, Yaşar Utku Alçalar, Mehmet Akçakaya

Organizations: Department of Electrical and Computer Engineering, University of Minnesota, MN, USA · Center for Magnetic Resonance Research, University of Minnesota, MN, USA

Abstract

Diffusion models (DMs) have emerged as powerful generative priors for MRI reconstruction with promising results. Yet DM-based methods require extensive iterative refinement, limiting their practical deployment. Consistency models (CMs) provide a compelling alternative, aiming to map out the diffusion trajectory in a single pass, enabling faster generation. In this work, we propose CM-RED, a novel MRI reconstruction method that integrates a pretrained CM into the regularization by denoising (RED) scheme. Our method builds on accelerated proximal gradient RED (RED-APG), and further incorporates controlled noise injection during the update steps to enhance generative diversity and accelerate convergence. Extensive experiments on the fastMRI knee and brain datasets demonstrate that CM-RED achieves high-quality reconstructions across multiple anatomies, contrast weights, acceleration factors, and undersampling patterns, using only 4 network function evaluations (NFEs). The proposed method consistently outperforms existing DM- and CM-based approaches in both quantitative metrics and visual fidelity, and exhibits strong robustness to hyperparameter variations, highlighting CM-RED as an efficient and effective generative framework for accelerated MRI reconstruction. The source code and pretrained models are publicly available at https://github.com/MerveGulle/CM-RED.

Figures & tables

Figure 1: Illustration of DM and CM trajectories on MRI data.

Figure 2: Illustration of CM-RED steps with $N\!=\!4$ , as described in Alg. 1 .

Acc.	$i_{N}$	$\gamma$	$\delta_{n}$	$\kappa$	$\rho$	$\mu^{(0)}$
4 $\times$	50	0.1	$\big(0.4,\,3.0,\,3.0,\,2.5\big)$	$\big(0.5,\,5.0\big)$	–2.0	0.9
8 $\times$	50	0.1	$\big(0.5,\,7.0,\,6.0,\,3.5\big)$	$\big(1.0,\,2.5\big)$	–3.0	0.9

Table 1: Complete set of hyperparameters for our four-step sampling setup, including noise schedule, regularization parameters, CM out scale parameter, and momentum coefficients.

Figure 3: Representative reconstructions of DM- and CM-based methods on fastMRI knee and brain data with equidistant $R=4$ and $R=8$ undersampling. The number of NFEs used by each method is shown in parentheses. Yellow boxes indicate magnified regions, and red arrows highlight residual artifacts. Best viewed when zoomed in.

Figure 4: Representative reconstructions for Gaussian undersampling. Number of NFEs for by each method is provided in parentheses. Yellow boxes indicate magnified regions, and red arrows highlight residual artifacts. Best viewed zoomed in.

Method	NFEs $\;\downarrow$	$\mathbf{R}$	Cor PD		Cor PD-FS		Axial T1-Pre		Axial T1-Post		Axial T2		Axial FLAIR
	NFEs $\;\downarrow$	$\mathbf{R}$	PSNR $\uparrow$	SSIM $\uparrow$	PSNR $\uparrow$	SSIM $\uparrow$	PSNR $\uparrow$	SSIM $\uparrow$	PSNR $\uparrow$	SSIM $\uparrow$	PSNR $\uparrow$	SSIM $\uparrow$	PSNR $\uparrow$	SSIM $\uparrow$
DPS	1000	$\times 4$	34.82	0.886	32.37	0.781	37.70	0.928	38.25	0.932	34.49	0.911	34.37	0.907
	1000	$\times 8$	31.79	0.832	30.30	0.727	33.75	0.896	33.89	0.899	30.14	0.868	30.23	0.862
DDS	100	$\times 4$	34.86	0.901	31.40	0.781	38.24	0.958	38.17	0.955	34.22	0.928	35.29	0.936
	100	$\times 8$	31.63	0.837	29.67	0.720	34.82	0.927	34.63	0.923	30.88	0.884	31.35	0.887
CM4IR	4	$\times 4$	29.96	0.766	24.32	0.528	36.83	0.918	37.35	0.917	31.43	0.856	34.47	0.915

Table 2: Quantitative comparison of reconstruction methods on fastMRI Knoll et al. (2020b) knee and brain datasets with

\mathbf{R}=4

and

8

using 1D equidistant undersampling. Best: BOLD , second-best: Underlined .

Method	NFEs $\;\downarrow$	$\mathbf{R}$	Cor PD		Cor PD-FS		Axial T1-Pre		Axial T1-Post		Axial T2		Axial FLAIR
	NFEs $\;\downarrow$	$\mathbf{R}$	PSNR $\uparrow$	SSIM $\uparrow$	PSNR $\uparrow$	SSIM $\uparrow$	PSNR $\uparrow$	SSIM $\uparrow$	PSNR $\uparrow$	SSIM $\uparrow$	PSNR $\uparrow$	SSIM $\uparrow$	PSNR $\uparrow$	SSIM $\uparrow$
DPS	1000	$\times 4$	35.19	0.894	32.81	0.795	35.36	0.915	35.88	0.918	33.37	0.903	32.92	0.901
	1000	$\times 8$	33.64	0.862	31.65	0.757	33.87	0.906	34.16	0.909	32.22	0.893	31.92	0.887
DDS	100	$\times 4$	35.10	0.909	31.94	0.796	37.77	0.956	38.00	0.956	34.15	0.928	35.47	0.940
	100	$\times 8$	33.14	0.871	30.47	0.745	35.63	0.939	35.68	0.936	32.19	0.903	33.14	0.912
CM4IR	4	$\times 4$	32.71	0.839	26.58	0.598	36.62	0.927	37.63	0.934	32.31	0.879	33.33	0.900

Table 3: Quantitative comparison of reconstruction methods on fastMRI Knoll et al. (2020b) knee and brain datasets with

\mathbf{R}=4

and

8

using 1D Gaussian undersampling. Best: BOLD , second-best: Underlined .

Panel of Figure 5 — Figure 5: Hyperparameter sensitivity analysis for CM-RED with $R=8$ Gaussian undersampling. Violin plots show the per-slice difference in PSNR (left) and SSIM (right) between the nominal configuration and reconstructions obtained with randomly perturbed hyperparameters.

Noise Inj.	Momentum	Cor PD Knee	Ax T2 Brain
✗	✗	35.59 / 0.917	34.96 / 0.936
✗	✓	36.09 / 0.923	35.25 / 0.937
✓	✗	35.71 / 0.917	35.23 / 0.937
✓	✓	36.58 / 0.927	36.13 / 0.945

Table 4: PSNR/SSIM on coronal PD knee and axial T2 brain R=4 equidistant undersampling with/without momentum and noise injection using 4 NFEs.

Panel of Figure 6 — Figure 6: Effect of noise injection and momentum on CM-RED convergence for fastMRI axial T1-post brain data with R=8 equidistant undersampling. PSNR and SSIM are reported between 4 to 30 NFEs.

Explore similar work

Accelerating Stroke MRI with Diffusion Probabilistic Models through Large-Scale Pre-training and Target-Specific Fine-Tuning
Mar 13, 2026Yamin Arefeen, Sidharth Kumar, Steven Warach +2Accelerated MRI Reconstruction Magnetic Resonance Imaging
Chandra Family Department of Electrical and Computer Engineering The University of Texas at Austin Austin, TX, USA · Department of Imaging Physics MD Anderson Cancer Center Houston, TX, USA · Dell Medical School Department of Neurology The University of Texas at Austin Austin, TX, USA +3
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
Phase-map synthesis from magnitude-only MR images using conditional score-based diffusion models with application in training of accelerated MRI reconstruction models
May 2, 2026M. Berk Sahin, Dilek Yalcinkaya, Abolfazl Hashemi +1Accelerated MRI Reconstruction Magnetic Resonance Imaging
Purdue University

Mar 13, 2026J/K move · Enter open · S save

Accelerating Stroke MRI with Diffusion Probabilistic Models through Large-Scale Pre-training and Target-Specific Fine-Tuning

Yamin Arefeen, Sidharth Kumar, Steven Warach, Hamidreza Saber+1

Chandra Family Department of Electrical and Computer Engineering The University of Texas at Austin Austin, TX, USA · Department of Imaging Physics MD Anderson Cancer Center Houston, TX, USA · Dell Medical School Department of Neurology The University of Texas at Austin Austin, TX, USA+3

Accelerated MRI Reconstruction Magnetic Resonance Imaging Diffusion Model Fine-Tuning Diffusion Models Neuroimaging

Open paper AskPDF arXiv

Purpose: To develop a data-efficient strategy for accelerated MRI reconstruction with Diffusion Probabilistic Generative Models (DPMs) that enables faster scan times in clinical stroke MRI when only limited fully-sampled data samples are available. Methods: Our simple training strategy, inspired by the foundation model paradigm, first trains a DPM on a large, diverse collection of publicly available brain MRI data in fastMRI and then fine-tunes on a small dataset from the target application using carefully selected learning rates and fine-tuning durations. The approach is evaluated on controlled fastMRI experiments and on clinical stroke MRI data with a blinded clinical reader study. Results: DPMs pre-trained on approximately 4000 subjects with non-FLAIR contrasts and fine-tuned on FLAIR data from only 20 target subjects achieve reconstruction performance comparable to models trained with substantially more target-domain FLAIR data across multiple acceleration factors. Experiments reveal that moderate fine-tuning with a reduced learning rate yields improved performance, while insufficient or excessive fine-tuning degrades reconstruction quality. When applied to clinical stroke MRI, a blinded reader study involving two neuroradiologists indicates that images reconstructed using the proposed approach from

2 \times

accelerated data are non-inferior to standard-of-care in terms of image quality and structural delineation. Conclusion: Large-scale pre-training combined with targeted fine-tuning enables DPM-based MRI reconstruction in data-constrained, accelerated clinical stroke MRI. The proposed approach substantially reduces the need for large application-specific datasets while maintaining clinically acceptable image quality, supporting the use of foundation-inspired diffusion models for accelerated MRI in targeted applications.

Abstract

Explore similar work

Accelerating Stroke MRI with Diffusion Probabilistic Models through Large-Scale Pre-training and Target-Specific Fine-Tuning

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

Phase-map synthesis from magnitude-only MR images using conditional score-based diffusion models with application in training of accelerated MRI reconstruction models