On the Foundation Model for Cardiac MRI Reconstruction

Cagan Alkan; Chi Zhang; Daniel B. Ennis; Mahmut Yurt; Michael Loecher; Shreyas S. Vasanawala

REVIEW 5 major objections 8 minor 18 references

On the Foundation Model for Cardiac MRI Reconstruction

T0 review · 5 major / 8 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A single adaptive, prompt-based network reconstructs cardiac MRI across mixed protocols without retraining per protocol.

desk verdict Plausible empirical result for multi-protocol CMR reconstruction, but the abstract oversells it and the adaptive-unrolling mechanism is untested. read the letter →

arxiv 2411.10403 v1 pith:6PDLULUW submitted 2024-11-15 eess.IV cs.CV

Chi Zhang , Michael Loecher , Cagan Alkan , Mahmut Yurt , Shreyas S. Vasanawala , Daniel B. Ennis This is my paper · ORCID

classification eess.IVcs.CV

keywords cardiacMRIreconstructionadaptiveunrollingfoundationmodelpatternandcontrastpromptchannel-shiftingundersampledk-spaceSSIMNMRSE

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper proposes a single machine-learning model for cardiac MRI reconstruction that works across a wide range of image contrasts, k-space sampling patterns, and acceleration rates, rather than requiring one trained network per acquisition protocol. The central claim is that combining adaptive unrolling (letting each image enter the reconstruction cascade at a point set by its acceleration rate), a Pattern and Contrast-Prompt-UNet (PCP-UNet), and channel-shifting gives the best reconstruction quality among the tested methods. Across in vivo experiments with mixed contrasts, sampling patterns, and acceleration rates, the authors report that Adaptive PCP-UNet achieved the highest structural similarity and the lowest NMRSE of the four compared designs. If the claim is correct, it points toward a general-purpose CMR reconstruction network that could be deployed without knowing the exact protocol in advance.

What carries the argument

The load-bearing mechanism is the acceleration-aware unrolled cascade: up to sixteen unrolled iterations, each with its own neural regularizer and conjugate-gradient data-consistency solver, and an entry point selected by the acceleration rate. PCP-UNet is the regularizer: it concatenates image-domain features with a contrast prompt and a sampling-pattern prompt built from statistical features of the k-space mask, letting one set of weights adapt to different contrasts and undersampling geometries. Channel-shifting widens the receptive field by concatenating circularly shifted copies of the input along the channel dimension, which the paper argues helps distinguish point-spread functions of different sampling patterns at a small computational cost.

What would settle it

Train a fixed-iteration PCP-UNet with the same total number of unrolled iterations, then compare its SSIM against Adaptive PCP-UNet separately at each acceleration rate; if the fixed model matches or beats the adaptive model at low rates, the adaptive schedule is not doing the claimed work. A second check is to compute the condition number of $E_{\Omega}$ for each rate and show that it varies enough to justify a different iteration count.

Watch

Extended reading notes

Core claim

The paper claims that one unrolled reconstruction network can handle mixed CMR protocols without retraining. It frames the task as the regularized problem $\arg\min_x \|E_{\Omega} x - y\|_2^2 + R(x)$, solved by variable splitting whose iterations are unrolled into a neural cascade. The adaptive component sends an undersampled image into the cascade at a starting unrolled iteration chosen by its acceleration rate, so the highest-rate images pass through all sixteen iterations and lower-rate images pass through only the latter ones, with each iteration's weights trained for a specific artifact level. The regularizer is PCP-UNet, which adds a k-space sampling-pattern prompt to the contrast prompt of Prompt-UNet, and channel-shifting concatenates circularly shifted copies of the input to widen the receptive field. In the reported in vivo experiments, Adaptive PCP-UNet had the highest SSIM and lowest NMRSE across uniform, Gaussian random, and radial sampling and across the tested acceleration rates and contrasts, with paired t-tests showing statistically significant differences between methods.

Load-bearing premise

The method assumes that the number of unrolled iterations is the right control for the conditioning of the forward operator, but the paper never states the mapping from acceleration rate to starting iteration or demonstrates that this mapping is optimal.

Editorial extensions

If this is right

A single Adaptive PCP-UNet model can replace a suite of protocol-specific networks, since it maintains high SSIM across mixed contrasts, sampling patterns, and acceleration rates in the same training run.
Low-acceleration scans can be reconstructed with fewer unrolled iterations, which shortens the inference path for the most common clinically requested rates.
The pattern prompt's statistical mask features are sufficient to separate uniform, Gaussian random, and radial sampling patterns, as shown by the improvement of PCP-UNet over the contrast-only Prompt-UNet.
Because paired t-tests at P=0.05 found each method statistically different, the ranking in SSIM is reproducible on this dataset, though the paper does not translate these differences into a clinical acceptance threshold.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The acceleration-to-starting-iteration mapping is never spelled out, so the adaptive gain could be tested directly by training a fixed-iteration PCP-UNet with the same per-iteration weights and comparing per-rate SSIM; the paper does not include that ablation.
The pattern-prompt idea should transfer beyond cardiac MRI to other anatomies or to non-Cartesian trajectories, because the statistical mask features are expressed in k-space coordinates rather than anatomy-specific image content.
A full computational-cost comparison, including total inference time per acceleration rate and memory use of the channel-shifting input, would quantify the practical benefit of adaptive unrolling more directly than SSIM alone.
The method's 'foundation model' framing is currently a single-task reconstruction model; its generality is demonstrated only across contrasts and sampling patterns within the same cardiac anatomy, so calling it a foundation model requires further evidence of transfer across anatomies and tasks.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this is a useful empirical paper on a real deployment problem, but it overclaims in the abstract and the marquee component—adaptive unrolling—is not actually tested as a mechanism.

What's new and what works: the idea of one network handling mixed contrasts, three sampling patterns, and acceleration rates from 8x to 24x is genuinely practical. The PCP-UNet prompt extension is a sensible incremental step beyond Prompt-UNet, and the channel-shifting trick for enlarging receptive field is simple and cheap—the kind of thing that might be worth stealing. Experiments use the public CMRxRecon dataset, and the internal comparisons are consistent: adaptive PCP-UNet beats the fixed and prompt-free variants across rates, patterns, and contrasts. That is real evidence, and the paired t-tests are not doing much work on thousands of samples, but the ordering itself is credible.

The soft spots, in proportion: the abstract says the method "outperforms the conventional ML-based method," but no such baseline appears anywhere in the experiments. There is also no protocol-specific network trained per contrast/rate/pattern, which is the natural bar for a foundation model claim. That is the biggest gap. Second, the adaptive-unrolling schedule is never specified: no mapping from acceleration rate to starting UI, no ablation. The stress-test note is right that each UI has independent network and CG parameters, so this is not iterating to convergence of one optimization; the compressed-sensing conditioning story is a loose analogy. And the adaptive-vs-fixed comparison changes both the number of active UIs and how training samples are allocated, so the SSIM gain cannot be attributed to conditioning-matched adaptivity. That doesn't kill the empirical result, but it means the paper's central novelty is under-supported. Third, no code, no hyperparameters, and evaluation appears to be on the validation split only—reproducibility will be limited.

Who this is for: CMR reconstruction researchers, and anyone working on multi-protocol MRI reconstruction. It is not a methods breakthrough, but it is a credible engineering contribution on an important problem.

If I were the editor, I would send it to peer review, with major revision expected. The authors need to add the missing baselines (especially a protocol-specific model and a conventional compressed-sensing or classical ML method), report the adaptive schedule, and run a proper ablation that isolates the schedule from capacity allocation. The core idea is worth engaging with; the current write-up just doesn't yet back the full claim.

Referee Report

5 major / 8 minor

Summary. The manuscript proposes a single reconstruction network intended to work across cardiac MRI contrasts, views, sampling patterns, and acceleration rates. It combines adaptive unrolling, in which the number of unrolled iterations is chosen based on the acceleration rate, with a Pattern and Contrast-Prompt UNet (PCP-UNet) and a channel-shifting input augmentation. Experiments on the CMRxRecon dataset compare four variants—Fixed UNet, Adaptive UNet, Fixed PCP-UNet, and Adaptive PCP-UNet—reporting SSIM and NRMSE boxplots for acceleration rates 8x, 16x, and 24x, three sampling patterns, and multiple contrasts. The authors conclude that Adaptive PCP-UNet significantly improves image quality and outperforms the 'conventional ML-based method.'

Significance. If substantiated, the ability to reconstruct a wide range of CMR protocols with a single trained network would address a real clinical bottleneck, since protocol-specific training is impractical in many clinical settings. The paper uses a large public in vivo dataset, evaluates multiple contrasts and sampling patterns, and includes component-wise ablation among its four variants, which is a useful experimental design. However, the current evidence does not yet establish the central claim: the key adaptive-unrolling schedule is not specified, and the comparison set does not include the 'conventional ML-based method' named in the abstract. The claimed improvements are therefore not yet fully supported.

major comments (5)

[Abstract and Section 3] The abstract claims the proposed model 'outperforms the conventional ML-based method,' but no such baseline appears in Section 3. The four evaluated methods are all variants of the proposed unrolled network (Fixed/Adaptive UNet and Fixed/Adaptive PCP-UNet). A protocol-specific network, a standard published unrolled network such as MoDL or VarNet, or a compressed-sensing baseline is needed to support the claim of outperforming existing ML-based reconstruction. Without such a comparison, the central 'foundation model' claim is not established.
[Section 2.2] The adaptive-unrolling schedule is the paper's linchpin but is never specified. The text states that 'images with the highest acceleration rate go through all sixteen UI' and that lower-acceleration images use 'the latter UI only,' yet the exact mapping from the six evaluated acceleration rates (4, 8, 12, 16, 20, 24) to the entry UI is never reported. Moreover, because each UI has its own network and conjugate-gradient parameters, the cascade is not iterating a single optimization to convergence; hence the compressed-sensing conditioning intuition in Section 2.2 does not directly transfer. Please report the full entry-UI schedule and provide an ablation that varies the mapping while holding total network capacity fixed.
[Figure 6] The Adaptive-versus-Fixed comparison is confounded. In the Adaptive condition, images of different acceleration rates use different numbers of active unrolled iterations, so the comparison changes both the per-rate architecture depth and the allocation of training samples across subnetworks. The reported SSIM gain of Adaptive over Fixed cannot therefore be attributed specifically to matching iteration count to conditioning. A matched-capacity comparison—for example, Fixed networks with the same per-rate UI counts, or Adaptive networks with shuffled entry schedules—is needed to isolate the mechanism.
[Section 3.1] The statement that 'paired t-tests show that the SSIM for each method is statistically different (P = 0.05)' is not accompanied by test statistics, multiplicity-corrected p-values, confidence intervals, or effect sizes. With 4,158 validation images and many subgroup comparisons (3 patterns x 3 rates x multiple contrasts), uncorrected paired t-tests are insufficient. Please provide a quantitative table with means, standard deviations, and a corrected significance analysis.
[Section 2.5] Several training and reconstruction hyperparameters are omitted: the data-consistency weight mu in Eqs. (3)-(4), the number of conjugate-gradient iterations and its tolerance, the optimizer, loss function, learning rate, batch size, number of epochs, and the channel-shift replica configuration (how many shifts and along which axes for each sampling pattern). These details are necessary to reproduce the results and to assess the claim that channel-shifting has only 'minor computational overhead.'

minor comments (8)

[Section 2.2] There is a typographical error: 'up to sixteeen UI' should read 'up to sixteen UI.'
[Section 2.5] The phrase 'Fixed amount of UP' should be 'Fixed amount of UI.'
[Section 3.1] The text says 'Fixed UNet exhibits the lowest SSIM and PSNR,' but Figure 6 reports NRMSE, not PSNR; please make the metric names consistent.
[Figure 6] The abbreviation 'NMRSE' appears in the figure and text; the standard abbreviation is NRMSE, and the paper should use one form throughout.
[Section 2.3] The phrase 'which we refereed as Pattern and Contrast-Prompt-UNet' should read 'which we refer to as Pattern and Contrast-Prompt-UNet.'
[Introduction] There is a typo: 'Any mistmatch' should read 'Any mismatch.'
[Conclusion] The phrase 'a wide range if CMR image contrast' should read 'a wide range of CMR image contrasts.'
[Figure 6] The boxplots are difficult to read without numerical axis annotations; please add a supplementary table or annotate the medians and interquartile ranges.

Circularity Check

0 steps flagged · score 1.0 of 10

No material circularity: the method is trained end-to-end and evaluated on held-out CMRxRecon data, with independent external citations carrying the load-bearing derivations.

full rationale

The paper's central claims are empirical: a PCP-UNet with adaptive unrolling and channel-shifting is trained on the public CMRxRecon dataset and evaluated on validation images that were not used for training. The reported SSIM/NMRSE numbers are validation metrics, not fitted constants, so no prediction reduces to a fitted input by construction. The adaptive-unrolling idea is justified by citing compressed sensing [15] and specifying that more unrolled iterations are used for higher acceleration rates; the exact UI schedule is not reported, but this is a missing-detail or correctness limitation, not circularity. The PCP-UNet architecture extends Prompt-UNet [16], an external prior work, and the pattern embeddings are computed directly from the sampling mask statistics, not from reconstruction outputs. Self-citations [8,9] are contextual references to the authors' own prior review and phase-contrast reconstruction work, and neither is load-bearing for the present method's derivation. The data-consistency step uses ESPIRiT [18] and conjugate gradient, both external and standard. Accordingly, the derivation chain is self-contained with respect to its empirical evaluation, and the only minor concern is the presence of self-citations, which do not carry the argument.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The central claim depends on hand-chosen design elements: adaptive unrolling schedule, 16-iteration depth, channel-shift settings, and the sampling-mask feature set. Standard supervised training of network weights is not listed as a free parameter because it is the normal estimation procedure. No new physical entities are introduced.

free parameters (5)

Adaptive unrolling entry-point schedule = not reported
Section 2.2: inputs enter at a UI determined by acceleration rate, but the schedule mapping rates 4, 8, 12, 16, 20, 24 to starting UI is not stated.
Total number of unrolled iterations = 16
Section 2.2: the cascade is implemented with up to sixteen UI, chosen without sensitivity analysis.
Channel-shift replica configuration = not reported
Section 2.4: the number and spacing of circular shifts are unspecified, although they control the receptive field and computational cost.
Sampling pattern statistical feature set = mean and variance of sample count and spacing
Section 2.3: the prompt encoding is hand-crafted from these statistics; no ablation justifies the choice.
Data consistency weight mu = not reported
Eq. 3 includes mu in the variable-splitting update, but its value is not given.

assumptions (6)

standard math Variable splitting converges to a solution of Eq. 1
Eqs. 2 to 4 assume alternating minimization approaches the regularized reconstruction.
domain assumption A neural network can represent the proximal operator of R(u)
Standard unrolled-learning assumption inherited from Hammernik et al. [1].
ad hoc to paper Acceleration rate is a sufficient proxy for E_Omega conditioning
Section 2.2 asserts this to motivate adaptive unrolling; no derivation or ablation is provided.
domain assumption Retrospectively undersampled CMRxRecon data are a valid proxy for prospective acceleration
Section 2.5 uses retrospective undersampling only; no prospective acquisitions are tested.
domain assumption ESPIRiT coil sensitivity maps are accurate
Section 2.5 relies on ESPIRiT [18]; estimation errors would propagate.
ad hoc to paper Hand-crafted mask statistics distinguish sampling patterns
Section 2.3 chooses mean and variance features without evidence that they capture pattern identity.

how reviews work

0 comments

Cite this review

Pith. "Pith review of On the Foundation Model for Cardiac MRI Reconstruction." pith.science (2026). https://pith.science/paper/6PDLULUW

@misc{pith2026241110403,
  author       = {Pith},
  title        = {Pith review of: On the Foundation Model for Cardiac MRI Reconstruction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6PDLULUW}},
  note         = {Machine review of arXiv:2411.10403}
}

read the original abstract

In recent years, machine learning (ML) based reconstruction has been widely investigated and employed in cardiac magnetic resonance (CMR) imaging. ML-based reconstructions can deliver clinically acceptable image quality under substantially accelerated scans. ML-based reconstruction, however, also requires substantial data and computational time to train the neural network, which is often optimized for a fixed acceleration rate or image contrast. In practice, imaging parameters are often tuned to best suit the diagnosis, which may differ from the training data. This can result in degraded image quality, and multiple trained networks are needed to fulfill the clinical demands. In this study, we propose a foundation model that uses adaptive unrolling, channel-shifting, and Pattern and Contrast-Prompt-UNet (PCP-UNet) to tackle the problem. In particular, the undersampled data goes through a different number of unrolled iterations according to its acceleration rate. Channel-shifting improves reconstructed data quality. The PCP-UNet is equipped with an image contrast and sampling pattern prompt. In vivo CMR experiments were performed using mixed combinations of image contrasts, acceleration rates, and (under)sampling patterns. The proposed foundation model has significantly improved image quality for a wide range of CMR protocols and outperforms the conventional ML-based method.

Figures

Figures reproduced from arXiv: 2411.10403 by the authors.

**Figure 1.** Adaptive unrolling. Undersampled measurements are fed to a specific unrolled iteration according to the acceleration rate. Images of higher acceleration rate proceed through more unrolled iterations than those of lower acceleration rates. Each unrolled iteration has its own network regularizer and conjugate gradient with parameters independently determined relative to the other unrolled iterations [PITH_FULL_IMAGE… view at source ↗

**Figure 2.** Pattern and Contrast-Prompt (PCP) UNet. PCP-UNet inherits the PromptUNet concept with additional k-space undersampling pattern prompt modules that are similar to the contrast prompt modules in Prompt-UNet [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** k-space sampling pattern embeddings were estimated from statistical information of the sampling mask, with an example of statistics along kx direction. The sampling mask is first split in two halves along the kx direction, in order to distinguish between uniform and random sampling. Summation along kx is applied to obtain the total number of samples as well as its distribution. The mean and variance of the sample … view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Enlarging receptive field via channel-shifting. Circular shifting along the subsampled direction(s) is performed to produce shifted replicas of the input image. The replicas are subsequently concatenated with the input along channel dimension to produce an augmented …

**Figure 5.** Figure 5: Representative reconstructed image results for selected combinations of sampling patterns, acceleration rates, and contrasts, using all tested methods: FU: Fixed amount of UI with a UNet regularizer; AU: Adaptive amount of UI with a UNet regularizer; FP: Fixed amount …

**Figure 6.** Figure 6: SSIM and NMRSE statistics for each reconstruction method (Fixed UNet, Adaptive UNet, Fixed PCP-UNet and Adaptive PCP-UNet) with respect to: (A) the k-space undersampling rate (8x, 16x, or 24x) and sampling pattern (uniform, random and radial), and (B) image contrast. A…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

18 extracted references · 17 canonical work pages

[1]

and Knoll,F.,2018.LearningavariationalnetworkforreconstructionofacceleratedMRI data

Hammernik, K., Klatzer, T., Kobler, E., Recht, M.P., Sodickson, D.K., Pock, T. and Knoll,F.,2018.LearningavariationalnetworkforreconstructionofacceleratedMRI data. Magnetic resonance in medicine, 79(6), pp.3055-3071

work page 2018
[2]

and Jacob, M., 2018

Aggarwal, H.K., Mani, M.P. and Jacob, M., 2018. MoDL: Model-based deep learning architecture for inverse problems. IEEE transactions on medical imaging, 38(2), pp.394-405. 10 F. Author et al

work page 2018
[3]

and Akcakaya, M., 2020

Knoll, F., Hammernik, K., Zhang, C., Moeller, S., Pock, T., Sodickson, D.K. and Akcakaya, M., 2020. Deep-learning methods for parallel magnetic resonance imaging reconstruction: A survey of the current approaches, trends, and issues. IEEE signal processing magazine, 37(1), pp.128-140

work page 2020
[4]

and Rueckert, D., 2017

Schlemper, J., Caballero, J., Hajnal, J.V., Price, A.N. and Rueckert, D., 2017. A deep cascade of convolutional neural networks for dynamic MR image reconstruc- tion. IEEE transactions on Medical Imaging, 37(2), pp.491-503

work page 2017
[5]

and Uğurbil, K., 2019

Akçakaya, M., Moeller, S., Weingärtner, S. and Uğurbil, K., 2019. Scan-specific robust artificial-neural-networks for k-space interpolation (RAKI) reconstruction: database-free deep learning for fast imaging. Magnetic resonance in medicine, 81(1), pp.439-453

work page 2019
[6]

and Akçakaya, M., 2020

Hosseini, S.A.H., Zhang, C., Weingärtner, S., Moeller, S., Stuber, M., Ugurbil, K. and Akçakaya, M., 2020. Accelerated coronary MRI with sRAKI: A database-free self-consistent neural network k-space reconstruction for arbitrary undersampling. Plos one, 15(2), p.e0229418

work page 2020
[7]

and Akçakaya, M., 2022

Zhang, C., Moeller, S., Demirel, O.B., Uğurbil, K. and Akçakaya, M., 2022. Resid- ual RAKI: A hybrid linear and non-linear approach for scan-specific k-space deep learning. NeuroImage, 256, p.119248

work page 2022
[8]

and Ennis, D.B., 2023

Oscanoa, J.A., Middione, M.J., Alkan, C., Yurt, M., Loecher, M., Vasanawala, S.S. and Ennis, D.B., 2023. Deep learning-based reconstruction for cardiac MRI: a re- view. Bioengineering, 10(3), p.334

work page 2023

Show all 18 references

[9]

and Ennis, D.B., 2023

Oscanoa, J.A., Middione, M.J., Syed, A.B., Sandino, C.M., Vasanawala, S.S. and Ennis, D.B., 2023. Accelerated two-dimensional phase-contrast for cardiovascular MRI using deep learning-based reconstruction with complex difference estimation. Magnetic resonance in medicine, 89(1...

2023
[10]

and Yang, G., 2022

Huang, J., Fang, Y., Wu, Y., Wu, H., Gao, Z., Li, Y., Del Ser, J., Xia, J. and Yang, G., 2022. Swin transformer for fast MRI. Neurocomputing, 493, pp.281-304

2022
[11]

and Ye, J.C., 2022

Chung, H. and Ye, J.C., 2022. Score-based diffusion models for accelerated MRI. Medical image analysis, 80, p.102479

2022
[12]

and Leiner, T., 2023

Rajiah, P.S., François, C.J. and Leiner, T., 2023. Cardiac MRI: state of the art. Radiology, 307(3), p.e223008

2023
[13]

and Cheng, H.L.M., 2022

Curtis, A.D. and Cheng, H.L.M., 2022. Primer and historical review on rapid car- diac CINE MRI. Journal of Magnetic Resonance Imaging, 55(2), pp.373-388

2022
[14]

and Galanski, M., 2002

Lotz, J., Meier, C., Leppert, A. and Galanski, M., 2002. Cardiovascular flow mea- surement with phase-contrast MR imaging: basic facts and implementation. Radio- graphics, 22(3), pp.651-671

2002
[15]

and Pauly, J.M., 2007

Lustig, M., Donoho, D. and Pauly, J.M., 2007. Sparse MRI: The application of compressed sensing for rapid MR imaging. Magnetic Resonance in Medicine: An Official Journal of the International Society for Magnetic Resonance in Medicine, 58(6), pp.1182-1195

2007
[16]

and Metaxas, D.N., 2023, October

Xin, B., Ye, M., Axel, L. and Metaxas, D.N., 2023, October. Fill the k-space and refine the image: Prompting for dynamic and multi-contrast MRI reconstruction. In International Workshop on Statistical Atlases and Computational Models of the Heart (pp. 261-273). Cham: Springer ...

2023
[17]

and Wang, S., 2023

Wang, C., Lyu, J. and Wang, S., 2023. CMRxRecon: an open cardiac MRI dataset for the competition of accelerated image reconstruction (2023). arXiv preprint arXiv:2309.10836

2023 arXiv
[18]

and Lustig,M., 2014

Uecker, M., Lai, P., Murphy, M.J., Virtue, P., Elad, M., Pauly, J.M., Vasanawala, S.S. and Lustig,M., 2014. ESPIRiT—aneigenvalueapproachto autocalibratingpar- allel MRI: where SENSE meets GRAPPA. Magnetic resonance in medicine, 71(3), pp.990-1001

2014

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

and Knoll,F.,2018.LearningavariationalnetworkforreconstructionofacceleratedMRI data

Hammernik, K., Klatzer, T., Kobler, E., Recht, M.P., Sodickson, D.K., Pock, T. and Knoll,F.,2018.LearningavariationalnetworkforreconstructionofacceleratedMRI data. Magnetic resonance in medicine, 79(6), pp.3055-3071

work page 2018

[2] [2]

and Jacob, M., 2018

Aggarwal, H.K., Mani, M.P. and Jacob, M., 2018. MoDL: Model-based deep learning architecture for inverse problems. IEEE transactions on medical imaging, 38(2), pp.394-405. 10 F. Author et al

work page 2018

[3] [3]

and Akcakaya, M., 2020

Knoll, F., Hammernik, K., Zhang, C., Moeller, S., Pock, T., Sodickson, D.K. and Akcakaya, M., 2020. Deep-learning methods for parallel magnetic resonance imaging reconstruction: A survey of the current approaches, trends, and issues. IEEE signal processing magazine, 37(1), pp.128-140

work page 2020

[4] [4]

and Rueckert, D., 2017

Schlemper, J., Caballero, J., Hajnal, J.V., Price, A.N. and Rueckert, D., 2017. A deep cascade of convolutional neural networks for dynamic MR image reconstruc- tion. IEEE transactions on Medical Imaging, 37(2), pp.491-503

work page 2017

[5] [5]

and Uğurbil, K., 2019

Akçakaya, M., Moeller, S., Weingärtner, S. and Uğurbil, K., 2019. Scan-specific robust artificial-neural-networks for k-space interpolation (RAKI) reconstruction: database-free deep learning for fast imaging. Magnetic resonance in medicine, 81(1), pp.439-453

work page 2019

[6] [6]

and Akçakaya, M., 2020

Hosseini, S.A.H., Zhang, C., Weingärtner, S., Moeller, S., Stuber, M., Ugurbil, K. and Akçakaya, M., 2020. Accelerated coronary MRI with sRAKI: A database-free self-consistent neural network k-space reconstruction for arbitrary undersampling. Plos one, 15(2), p.e0229418

work page 2020

[7] [7]

and Akçakaya, M., 2022

Zhang, C., Moeller, S., Demirel, O.B., Uğurbil, K. and Akçakaya, M., 2022. Resid- ual RAKI: A hybrid linear and non-linear approach for scan-specific k-space deep learning. NeuroImage, 256, p.119248

work page 2022

[8] [8]

and Ennis, D.B., 2023

Oscanoa, J.A., Middione, M.J., Alkan, C., Yurt, M., Loecher, M., Vasanawala, S.S. and Ennis, D.B., 2023. Deep learning-based reconstruction for cardiac MRI: a re- view. Bioengineering, 10(3), p.334

work page 2023