CogGen: Cognitive-Load-Inspired Fully Unsupervised Deep Generative Modeling for Compressively Sampled MRI Reconstruction

Qingyong Zhu; Yumin Tan; Xiang Gu; Dong Liang

eess.IVFeb 20, 2026

CogGen: Cognitive-Load-Inspired Fully Unsupervised Deep Generative Modeling for Compressively Sampled MRI Reconstruction

Authors: Qingyong Zhu, Yumin Tan, Xiang Gu, Dong Liang

Organizations: Medical AI Research Centre, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, Guangdong, China · School of Biomedical Engineering, Southern Medical University, Guangzhou 510515, Guangdong, China · School of Mathematics and Statistics, Xi’an Jiaotong University, Xi’an 710049, Shaanxi, China · Paul C. Lauterbur Research Centre for Biomedical Imaging, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, Guangdong, China

Abstract

Fully unsupervised deep generative modeling (FU-DGM) offers significant potential for compressively sampled magnetic resonance imaging (CS-MRI) reconstruction. Representative FU-DGM formulations, such as deep image prior (DIP) and implicit neural representation (INR), employ architectural bias to induce a low-dimensional manifold in the image space that aligns with the forward observation. However, as the underlying inverse system is highly ill-posed, prolonged iterative fitting in FU-DGM typically leads to poor efficiency and noise amplification. In this paper, guided by the cognitive principle of easy-to-hard learning, we propose CogGen, an FU-DGM framework that reformulates CS-MRI reconstruction as a staged inversion problem. Specifically, CogGen implements an self-paced curriculum learning (SPCL)-driven progressive scheduling strategy through an MRI-aware dual-threshold weighting criterion, which adaptively regulates k-space measurement participation. The data-consistency residual thresholding evaluates the fitting reliability of the current generator, while the k-space radius thresholding controls stage-wise measurement exposure, thereby avoiding uniform fitting throughout optimization. Theoretically, our analysis shows that, when early stages favor easy-to-fit measurements, CogGen yields a reduced local sufficient-iteration bound and a smaller cumulative noise-amplification bound, explaining the improved convergence behavior and reconstruction fidelity of CogGen within a finite iteration budget. Numerical experiments demonstrate that both CogGen instantiations, CogGen-DIP and CogGen-INR, achieve superior performance over prevailing CS-MRI reconstruction techniques, including unsupervised and supervised pipelines.

Figures & tables

Fig. 1: Cognitive-load view of CogGen: measurements with lower load exhibit higher learning utility and are therefore selected earlier, whereas high-load measurements with reduced suitability are deferred to the further stages.

Fig. 2: Reconstruction results on Data $\#1$ at AF = 8. (a)-(j): DIP-TV, BM3D-FISTA, DISCUS, SSDU, aSeq-DIP, Hash-INR- $\ell_{2}$ - $\ell_{1}$ , PDAC, MoDL, CogGen-DIP and CogGen-INR. The corresponding error maps are presented in the bottom row, alongside the GT and downsampling mask displayed at the far left.

Fig. 3: Reconstruction results on Data $\#2$ at AF = 6. (a)-(j): DIP-TV, BM3D-FISTA, DISCUS, SSDU, aSeq-DIP, Hash-INR- $\ell_{2}$ - $\ell_{1}$ , PDAC, MoDL, CogGen-DIP and CogGen-INR. The corresponding error maps are presented in the bottom row, alongside the GT and downsampling mask displayed at the far left.

Fig. 4: Evolution of sample selection across five stages ( $C_{1}$ - $C_{5}$ ), showing the expansion from central k-space to the full acquired pattern. Top: Data $\#1$ (2D, $w=1$ ). Bottom: Data $\#2$ (1D, $w=0.8$ ). Here, $C_{5}$ denotes the final stage, which contains the original undersampled k-space measurements.

Methods	Data $\#1$ (2D, AF = 8)		Data $\#2$ (1D, AF = 6)
Methods	RLNE ${}_{\text{ROI}}$ (%)	PSNR ${}_{\text{ROI}}$ (dB)	RLNE ${}_{\text{ROI}}$ (%)	PSNR ${}_{\text{ROI}}$ (dB)
DIP-TV	9.84	38.17	6.87	30.65
BM3D-FISTA	9.60	38.45	6.28	31.42
DISCUS	8.59	38.91	5.66	32.32
SSDU	8.52	38.91	4.83	33.79
aSeq-DIP	8.41	39.24	5.42	32.77

TABLE I: Quantitative results (

\text{RLNE}_{\text{ROI}}

and

\text{PSNR}_{\text{ROI}}

) on Data

\#1

(2D, AF = 8) and Data

\#2

(1D, AF = 6). The top two performing entries are bolded.

Fig. 5: Performance gains from CogGen on Data $\#1$ (2D, AF = 8) and Data $\#2$ (1D, AF = 6). Integrating CogGen into DIP- and INR-based baselines notably boosts accuracy and convergence under the same iteration budget. (a)-(d): DIP-TV, Hash-INR- $\ell_{2}$ - $\ell_{1}$ , CogGen-DIP, and CogGen-INR.

Fig. 6: Comparison between MRI-Aware and Random Scheduling on Data $\#3$ at an AF of 5. (a)-(c): DIP-TV, SRS-DIP, CogGen-DIP; (d)-(f): Hash-INR- $\ell_{2}$ - $\ell_{1}$ , SRS-INR, CogGen-INR. The corresponding error maps are presented in the bottom row, alongside the GT and downsampling mask displayed at the far left.

Methods	Data $\#3$ (1D, AF = 5)
Methods	RLNE ROI (%)	PSNR ROI (dB)
DIP-TV	8.13	37.58
SRS-DIP	7.96	37.76
CogGen-DIP	6.04	40.13
Hash-INR- $\ell_{2}$ - $\ell_{1}$	7.01	38.86
SRS-INR	6.95	38.95

TABLE II: Quantitative evaluation of MRI-aware vs. random scheduling on Data

\#3

at AF = 5.

Fig. 7: Influence of stage count on Data $\#4$ at AF = 10. Performance peaks at $C_{4}$ and decreases when additional stages are introduced, highlighting the importance of appropriate stage-count selection for reconstruction performance.

Fig. 8: Single-mode vs. dual-mode thresholding (Data $\#4$ , AF = 10). Dual-mode scheduling enhances detail recovery and reconstruction fidelity. (a-d): Hash-INR- $\ell_{2}$ - $\ell_{1}$ , CogGen CL -INR, CogGen SPL -INR and CogGen-INR.

Methods	Data $\#4$ (2D, AF = 10)
Methods	RLNE ROI (%)	PSNR ROI (dB)
Hash-INR- $\ell_{2}$ - $\ell_{1}$	8.89	39.10
CogGen CL -INR	8.41	39.59
CogGen SPL -INR	8.29	39.71
CogGen-INR	8.14	39.88

TABLE III: Quantitative comparison for two thresholding schemes on Data

\#4

(AF = 10).

Accelerated MRI Reconstruction

Accelerated MRI reconstruction recovers diagnostic images from undersampled k-space data to reduce scan time while preserving anatomical and quantitative information.

See all 85 papers

Stochastic Finite-Transition Learning and Exploration for Accelerated MRI Reconstruction
Oct 9, 2026Muhammad Usama Mirza, Esin Öztürk Işık, Tolga ÇukurAccelerated MRI Reconstruction Generative Modeling
Dept. of Electrical-Electronics Engineering and National Magnetic Resonance Research Center (UMRAM), Bilkent University, Ankara, Turkey, 06800 · Institute of Biomedical Engineering, Bogazici University, Istanbul, Turkey, 34684
Consistency Models for Fast MRI Reconstruction Using Regularization by Denoising
Aug 20, 2026Merve Gülle, Junno Yun, Yaşar Utku Alçalar +1Accelerated MRI Reconstruction Diffusion-Based Inverse Problems
Department of Electrical and Computer Engineering, University of Minnesota, MN, USA · Center for Magnetic Resonance Research, University of Minnesota, MN, USA
Information-Theoretic Optimization for Task-Adapted Compressed Sensing Magnetic Resonance Imaging
Apr 14, 2026Xinyu Peng, Ziyang Zheng, Wenrui Dai +5Magnetic Resonance Imaging Compressed Sensing
Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai 200240, China. · Department of Electronic Engineering, Shanghai Jiao Tong University, Shanghai 200240, China. · Department of Computer Science, City University of Hong Kong, Hong Kong SAR, China. +1

Oct 9, 2026J/K move · Enter open · S save

Stochastic Finite-Transition Learning and Exploration for Accelerated MRI Reconstruction

Muhammad Usama Mirza, Esin Öztürk Işık, Tolga Çukur

Dept. of Electrical-Electronics Engineering and National Magnetic Resonance Research Center (UMRAM), Bilkent University, Ankara, Turkey, 06800 · Institute of Biomedical Engineering, Bogazici University, Istanbul, Turkey, 34684

Accelerated MRI Reconstruction Generative Modeling

Open paper AskPDF arXiv

Generative MRI reconstruction maps an initial source distribution to the target image distribution along a learned recovery trajectory. Common generative methods, such as diffusion and flow, typically learn fine-grained local dynamics whose integration defines this global mapping, hence requiring many network evaluations at inference. Existing few-step sampling approaches can lower computational burden by approximating large transformations from local dynamics, albeit at the expense of reconstruction fidelity. FLEX-MRI instead partitions recovery into a small number of finite intervals between retained states defined by progressive changes in k-space support, and directly learns the reconstruction change across each interval. Because such finite transitions can admit multiple plausible outcomes, FLEX-MRI combines support-relaxed finite-transition learning with stochastic exploration. During training, randomized k-space sampling masks define finite-transition targets with varying support, while the prediction loss remains unrestricted by each individual mask. At inference, FLEX-MRI perturbs a random subset of currently estimated k-space coefficients, allowing the learned predictor to stochastically explore alternative transition paths. Each predicted state is then reconciled with the acquired measurements through data consistency. We theoretically show that this stochastic exploration induces nonzero transition-level variation through the learned recovery map and data consistency without requiring an auxiliary generative sampler. Experiments on multi-coil MRI datasets show that FLEX-MRI outperforms competing methods while maintaining computational efficiency comparable to few-step baselines.

Abstract

Accelerated MRI Reconstruction

Explore similar work

Stochastic Finite-Transition Learning and Exploration for Accelerated MRI Reconstruction

Consistency Models for Fast MRI Reconstruction Using Regularization by Denoising

Information-Theoretic Optimization for Task-Adapted Compressed Sensing Magnetic Resonance Imaging