Pith. sign in

REVIEW 5 major objections 6 minor 12 references

Deep Learning Assisted Outer Volume Removal for Highly-Accelerated Real-Time Dynamic MRI

T0 review · 5 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read A post-processing pipeline that estimates and subtracts the signal from tissues surrounding the heart lets free-breathing real-time cine MRI be accelerated eightfold with image quality comparable to clinical breath-hold references.

desk verdict A genuinely new ghost-aware outer-volume removal method for RT cardiac MRI, with a prospective validation design that is better than most in the field, but the quantitative evidence is thinner than the abstract claims. read the letter →

arxiv 2505.00643 v1 pith:S33NGFP4 submitted 2025-05-01 eess.IV cs.AIcs.CVphysics.med-ph

Merve Gülle , Sebastian Weingärtner , Mehmet Akçakaya This is my paper · ORCID
classification eess.IVcs.AIcs.CVphysics.med-ph
keywords real-timeMRIdynamiccardiaccineoutervolumeremovalghostingartifactsdeeplearningreconstructionparallelimagingself-supervised
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 aims to show that a purely reconstruction-side step—estimating and subtracting the signal from tissues outside the heart—removes the main obstacle to high acceleration in real-time cine cardiac MRI. It analyzes low-temporal-resolution composite images formed from time-interleaved shifted undersampling as a sum of a motion-averaged heart, a pseudo-periodic ghost from cardiac motion, and a stationary background, then trains a neural network to detect the ghost and a second network to locate the heart. After subtracting the cleaned background estimate from k-space, a physics-driven unrolled network trained with a self-supervised, OVR-specific loss restores the image. The claimed payoff is that free-breathing real-time cine at R=8 acceleration matches the visual quality and left-ventricular function measurements of R=4 or breath-hold clinical references, with no change to the acquisition protocol.

What carries the argument

The governing identity is the composite-image decomposition $x_{com}(t_0) = x_{moving}(t_0) + x_{ghost}(t_0) + x_{background}(t_0)$, which turns the outer-volume-removal problem into three tractable subproblems. The load-bearing mechanism is a ResNet ghost detector that consumes four adjacent composite images and outputs the ghost maps for those frames; after ghost subtraction the stationary background is masked by a U-Net-predicted heart mask and subtracted from each frame's k-space. The final component is the OVR-specific SSDU loss for the 35-unroll PD-DL network, which mixes a standard self-supervised data-consistency loss with a consistency term between reconstructions using full and ROI-masked coil sensitivities, so the network learns to keep ROI signal while pushing residual outer-volume signal outside the mask.

What would settle it

Acquire a prospective R=8 free-breathing cine dataset from subjects with pronounced respiratory or diaphragmatic motion and compare the pipeline's systolic-frame reconstruction to a breath-hold reference at the same cardiac phase; visible ghost energy from other cardiac phases inside the myocardium, or an LV ejection fraction that drifts outside the agreement range reported in the prospective cohort, would falsify the central decomposition.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central claim is that outer-volume aliasing—not temporal resolution—is the limiting factor at high acceleration and that removing the outer volume in k-space unlocks R=8 real-time cine. The authors characterize composite cine images as $x_{com} = x_{moving} + x_{ghost} + x_{background}$, where time-interleaved shifted sampling gives each foldover a distinct modulation phase; the moving components add constructively at the true heart location, stationary components cancel at side foldovers, and the moving components form a pseudo-periodic ghost. A ResNet estimates the ghost, the background is obtained by subtracting it, an OVR mask isolates the heart, and k-space subtraction $y_{OVR} = y - F(m_{OVR} \cdot x_{background})$ removes the outer-volume signal before an unrolled physics-driven network performs frame-by-frame reconstruction. The PD-DL network is trained self-supervised with a loss that adds a consistency term between full-sensitivity and ROI-masked-sensitivity reconstructions, preventing both signal loss in the ROI and artifacts from residual outer-volume signal. Across retrospective bSSFP data and prospective GRE data at R=8, the method is reported to produce images visually comparable to clinical references and LV function values with no statistically significant difference from breath-hold cine in the prospective cohort.

Load-bearing premise

The argument stands or falls on the Section 2 assumption that, over the R consecutive frames used to form a composite image, every frame separates into a moving cardiac component and a stationary background, with respiration and diaphragm motion slow enough to be negligible; if that separation fails, the ghost estimate is wrong and subtracting it from k-space will corrupt the heart region rather than clean it.

Editorial extensions

If this is right

  • At R=8, the OVR pipeline is claimed to yield images visually comparable to the clinical reference, whereas TGRAPPA at R=8 is severely artifact-degraded and PD-DL without OVR blurs the myocardium and papillary muscles.
  • Left-ventricular volumes, ejection fraction, stroke volume, and mass measured from OVR-based R=8 reconstructions agree with breath-hold segmented cine to the point of no statistically significant difference in the prospective study.
  • Because OVR acts in post-processing, it can be added to existing time-interleaved acquisitions without outer-volume-suppression pulses, avoiding SAR, steady-state disruption, and signal regrowth.
  • Ghost detection trained on R=4 data transfers to prospectively accelerated R=8 data, meaning no fully sampled reference at the target acceleration is needed to build the pipeline.
  • The comparisons are deliberately run without temporal regularization, so the reported improvements are attributed to outer-volume removal rather than temporal blurring.

Reading between the lines

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

  • If the decomposition holds, the same outer-volume-removal recipe should transfer to other dynamic MRI settings in which a large stationary FOV aliases into the ROI; for non-Cartesian trajectories the ghost structure would need its own derivation, a point the paper leaves open.
  • The two-stage pipeline (ghost detection then reconstruction) is likely not the end of the road: end-to-end training that backpropagates the reconstruction loss through the ghost detector could refine the background estimate, which the paper notes as future work.
  • A plausible next test is pushing to R=10 or higher; if outer-volume aliasing is genuinely eliminated, the remaining limit should be g-factor noise inside the ROI, and combining OVR with virtual coils or spatiotemporal regularization might extend the regime further.
  • The method's generalization boundaries (different scanner vendors, field strengths beyond 3T, arrhythmic patients, pediatric sizes) are not established in the paper, so those are natural stress tests before clinical adoption.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. The manuscript proposes a post-processing outer volume removal (OVR) framework for highly accelerated real-time cine cardiac MRI. The authors characterize the pseudo-periodic ghosting that appears in low-temporal-resolution composite images formed from time-interleaved shifted undersampling (Sec. 2, Eqs. 1-2), train a ResNet to predict these ghosts from four adjacent composite images (Sec. 3.1, Eq. 3), subtract the resulting clean outer-volume background from each timeframe's k-space (Eq. 5), and reconstruct the residual ROI data with a 35-unroll physics-driven network trained with multi-mask SSDU and a new dual-sensitivity consistency loss (Sec. 3.3, Eqs. 7-8). Evaluations include retrospectively R=8-undersampled bSSFP data (3 test subjects; Table 1) and prospectively R=8-acquired GRE data with breath-hold segmented cine as the clinical baseline (6 test subjects; Table 2), with TGRAPPA and a matched non-OVR PD-DL network as controls. The paper claims visual parity with clinical baselines and quantitative superiority over conventional reconstructions at R=8.

Significance. The core idea is attractive and timely: it removes aliasing from extra-cardiac tissue in post-processing rather than through acquisition-level suppression, so it is applicable to existing protocols, and it is evaluated with a genuine prospective R=8 acquisition plus a matched non-OVR deep-learning control. The analytical decomposition of composite images into an averaged moving component, a pseudo-periodic ghost, and a stationary background (Sec. 2) is a genuine contribution, and the masked/full-sensitivity consistency loss (Eq. 8) is a clever mechanism for avoiding both residual-background artifacts and ROI signal loss. The manuscript promises code release upon acceptance, which would be valuable for reproducibility. However, the significance of the claimed quantitative gains is currently limited by thin and partly self-referential evidence: the retrospective LV analysis uses only three subjects and shows sizable differences; the reported PSNR/SSIM values are never disclosed; and the prospective agreement claim rests on a low-power n=6 paired t-test. These gaps, not the plausibility of the method, are the main barrier to accepting the abstract's claims.

major comments (5)
  1. [5.1.2] The text states that 'Quantitative evaluations of PSNR and SSIM values ... further support the improvements,' but no PSNR or SSIM numbers appear anywhere in the manuscript. Since the abstract's claim of 'outperforming ... quantitatively' for the retrospective study rests on this sentence, the authors should report numeric PSNR and SSIM (mean and SD over the 3 test subjects) for TGRAPPA R=8, PD-DL without OVR, and the proposed method, and state the reference used (TGRAPPA R=4 or BH) and whether the metric is evaluated over the full FOV or restricted to the ROI.
  2. [Table 1 / Sec. 5.1.3] The retrospective LV analysis on n=3 subjects reports EF 55.0 (6) versus 65.4 (7), ESV 56.0 (13) versus 42.2 (15), and ESM 137.7 (30) versus 105.7 (24) against the TGRAPPA R=4 baseline. These are not 'close agreement' in any conventional sense, and with n=3 and no statistical test, the claim in Sec. 5.1.3 is unsupported. The authors should report subject-level values with limits of agreement, add a proper statistical treatment, or substantially temper the retrospective agreement claim.
  3. [4.2.1 vs 5.1.2 and Fig. 3e] The ghost-detection network is trained to reproduce composite-minus-TGRAPPA(R=4) reference labels, and the retrospective quantitative comparison is computed against the same TGRAPPA(R=4) reconstructions. Moreover, the final images recombine the masked background estimate (Fig. 3e), which is trained to resemble TGRAPPA(R=4) in the outer volume. The claimed quantitative advantage over the baseline is therefore entangled with the training target, and whole-FOV PSNR/SSIM against TGRAPPA(R=4) does not provide an independent evaluation. Metrics restricted to the ROI, and a comparison against the BH reference for the prospective data, would decouple the evaluation from the training labels.
  4. [3.3 / 4.2.3] The encoding operator E in Eqs. (6)-(8) requires coil sensitivity maps, but the manuscript does not state how these are estimated for the OVR pipeline (e.g., from raw R=8 k-space, from composite data, from OVR-subtracted data, or from a separate calibration scan). If sensitivities are computed from data containing the ghosting and outer-volume signal that the method is designed to remove, the forward model becomes inconsistent with the OVR target, and the improvements attributed to the Eq. (8) consistency loss could reflect that bias. Specify the estimation procedure and include a sensitivity analysis comparing reconstructions obtained with sensitivities from unmodified versus OVR-processed data.
  5. [5.2.3 / Sec. 4.3] For the prospective study, 'strong agreement ... with no statistical differences (P > .05 in all cases)' is inferred from a paired t-test on n=6 subjects. Failure to reject the null at n=6 has low power and cannot establish agreement or equivalence. Report Bland-Altman limits of agreement or an equivalence test, together with subject-level EDV/ESV/EF values, and discuss the fact that the BH comparator was acquired in a separate scan.
minor comments (6)
  1. [4.2.1 / 4.2.3] Learning rates are printed as '10-3' and '10 -4' with inconsistent spacing; they should be formatted as 10^-3 and 10^-4.
  2. [4.3] The Shapiro-Wilk test is stated as performed, but no test statistic or outcome is reported; with n=3 the test has little power, so the authors should either report the result or remove the sentence.
  3. [Eq. (3)] The notation in Eq. (3) and the surrounding text (e.g., xconcatcom(t0) and the indexing over tau = t0-2 ... t0+1) is hard to parse; it should be written out more explicitly.
  4. [Tables 1 and 2] The table captions should state the number of subjects and clarify that values are per-subject means with SDs, and whether EDV/ESV are body-surface-area indexed.
  5. [Abstract / 5.2.2] The 'clinical baseline' for the prospective study is a breath-hold ECG-gated segmented cine from a separate acquisition with no in-plane acceleration, and the RT sequence is GRE rather than bSSFP; a sentence stating that this cross-sequence comparison is inherent to the study design would improve clarity.
  6. [Sec. 2] The stationarity assumption over R consecutive frames is asserted as approximate, but at R=8 with 61 ms temporal resolution the window is about 0.5 s, over which respiratory chest-wall motion is non-negligible; a brief quantitative discussion of this timescale would strengthen the analytical derivation.

Circularity Check

1 steps flagged · score 4.0 of 10

Retrospective 'comparable to baseline' claim is partly built from the TGRAPPA R=4 labels used to train the ghosting detector; the prospective BH comparison remains independent.

  1. fitted input called prediction [Sec. 4.2.1 (ghosting detection training) + Sec. 5.1.2 (retrospective quantitative evaluation)]
    "Reference ghosting artifact images were generated for each coil by subtracting TGRAPPA (Breuer et al., 2005) reconstructions of the R = 4 data from the composite images. ... Quantitative evaluations of PSNR and SSIM values, relative to the baseline R = 4 reconstruction, further support the improvements made by the proposed method."

    The ghosting-detection network is trained to output the difference between composite R=8 images and TGRAPPA R=4 reconstructions (Sec. 4.2.1). Equations (4)-(5) then use this output to form the background estimate and subtract it from k-space, so the final R=8 OVR reconstruction is explicitly constructed to agree with the TGRAPPA R=4-derived outer-volume estimate. The retrospective quality metrics (Sec. 5.1.2) are then computed relative to the same TGRAPPA R=4 reconstruction. Thus, the reported 'comparable to baseline R=4' performance in the retrospective study is partially ensured by construction: the reference defines the training target.

full rationale

The most defensible circular step is retrospective and partial. The ghosting detector is trained on labels formed by subtracting TGRAPPA R=4 reconstructions from composite images, and the same TGRAPPA R=4 reconstruction is used as the baseline for PSNR/SSIM. Since the OVR background estimate is derived from this trained output, the retrospective 'matches baseline R=4' claim is partly self-referential. However, the paper's central prospective claim is evaluated against independently acquired breath-hold cine data, and the method's primary outperformance comparison is against TGRAPPA R=8 and non-OVR PD-DL, neither of which is the label source. The Section 2 decomposition (Eqs. 1-2) is a definitional modelling assumption rather than a fitted-hidden-input, so I do not treat it as a separate circular step. The self-citations to SSDU and earlier PD-DL work are standard technique references, not uniqueness theorems or unverified load-bearing constraints. The skeptic's sensitivity-map concern is a potential validation gap, but it is not a demonstrated circular reduction. Overall, the derivation is self-contained for the prospective claim, with partial circularity only in the retrospective baseline comparison, hence a score of 4.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The central pipeline rests on a standard parallel-imaging encoding model plus three domain assumptions: the moving/stationary split in Eq. (1), the ghost/background decomposition in Eq. (2), and the use of TGRAPPA(R=4) difference maps as ghosting labels. The only hand-set numeric parameters listed are lambda=0.02 and K=3 SSDU masks; they are not fitted to test data but are chosen by the authors. No new physical entities (particles, forces, dimensions) are introduced.

free parameters (2)
  • lambda (OVR consistency loss weight) = 0.02
    Set by hand in Sec. 4.2.3 for the loss in Eq. (8). Controls the balance between the SSDU loss and the masked-sensitivity consistency term.
  • K (number of SSDU masks) = 3
    Chosen by hand in Sec. 4.2.3. Determines how the acquired k-space locations are partitioned for self-supervised training.
assumptions (5)
  • domain assumption x(t) = x_moving(t) + x_stationary(t), with the heart moving and surrounding tissues stationary over R consecutive frames.
    Invoked in Section 2, Eq. (1). If diaphragm, chest, or other non-cardiac motion is non-negligible at the composite timescale, the ghost estimate is biased.
  • domain assumption The composite image decomposes as x_com = x_moving_avg + x_ghost + x_background, with a well-defined pseudo-periodic ghost term.
    Eq. (2) is definitional. The pseudo-periodic structure is illustrated with Fig. 1 and example data, not derived from the sampling equations in a formal proof.
  • ad hoc to paper Reference ghosting maps computed as TGRAPPA(R=4) minus composite are a valid supervised target for ghosts at R=8.
    Section 4.2.1. TGRAPPA(R=4) contains spatially varying g-factor noise, and prospective R=8 generalization relies on this proxy because no R=8 ghosting reference exists.
  • domain assumption Subtracting a masked outer-volume estimate from k-space and reconstructing with full sensitivity maps maps residual outer-volume signal outside the ROI.
    Section 3.3 and Fig. 4. The proposed loss in Eq. (8) is designed to enforce this behavior, but it is an assumed modeling property of the coil-sensitivity and encoding model.
  • standard math Standard parallel-MRI encoding model with Fourier transform, coil sensitivities, and Cartesian time-interleaved undersampling.
    Used throughout Sec. 3 for the data-fidelity operators and the OVR subtraction in Eq. (5). Standard in the MRI reconstruction literature.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Learning Assisted Outer Volume Removal for Highly-Accelerated Real-Time Dynamic MRI." pith.science (2026). https://pith.science/paper/S33NGFP4

@misc{pith2026250500643,
  author       = {Pith},
  title        = {Pith review of: Deep Learning Assisted Outer Volume Removal for Highly-Accelerated Real-Time Dynamic MRI},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/S33NGFP4}},
  note         = {Machine review of arXiv:2505.00643}
}
read the original abstract

Real-time (RT) dynamic MRI plays a vital role in capturing rapid physiological processes, offering unique insights into organ motion and function. Among these applications, RT cine MRI is particularly important for functional assessment of the heart with high temporal resolution. RT imaging enables free-breathing, ungated imaging of cardiac motion, making it a crucial alternative for patients who cannot tolerate conventional breath-hold, ECG-gated acquisitions. However, achieving high acceleration rates in RT cine MRI is challenging due to aliasing artifacts from extra-cardiac tissues, particularly at high undersampling factors. In this study, we propose a novel outer volume removal (OVR) method to address this challenge by eliminating aliasing contributions from non-cardiac regions in a post-processing framework. Our approach estimates the outer volume signal for each timeframe using composite temporal images from time-interleaved undersampling patterns, which inherently contain pseudo-periodic ghosting artifacts. A deep learning (DL) model is trained to identify and remove these artifacts, producing a clean outer volume estimate that is subsequently subtracted from the corresponding k-space data. The final reconstruction is performed with a physics-driven DL (PD-DL) method trained using an OVR-specific loss function to restore high spatio-temporal resolution images. Experimental results show that the proposed method at high accelerations achieves image quality that is visually comparable to clinical baseline images, while outperforming conventional reconstruction techniques, both qualitatively and quantitatively. The proposed approach provides a practical and effective solution for artifact reduction in RT cine MRI without requiring acquisition modifications, offering a pathway to higher acceleration rates while preserving diagnostic quality.

Figures

Figures reproduced from arXiv: 2505.00643 by the authors.

Figure 1
Figure 1. Illustration of the decomposition of a composite real-time cine cardiac MR image into various [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Importance of accounting for ghosting artifacts during OVR. (a) TGRAPPA reconstruction at [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Proposed reconstruction pipeline: (a) DL-based ghosting detection: A ResNet with 15 residual [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: PD-DL reconstruction of the OVR k-space using (a) masked sensitivity maps, (b) full sensitivity [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Detection of the ghosting artifact of a composite RT cine image (retrospective acceleration [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Comparison of reconstructed images from a retrospectively [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Detection and removal of ghosting artifacts for the prospectively accelerated [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: Representative reconstruction results for the prospective [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

12 extracted references · 12 canonical work pages

  1. [1]

    K., Mani, M

    Aggarwal, H. K., Mani, M. P., & Jacob, M. (2018). MoDL: Model-based deep learning architecture for inverse problems. IEEE Transactions on Medical Imaging , 38(2), 394–405. Ak¸ cakaya, M., Yaman, B., Chung, H., & Ye, J. C. (2022). Unsupervised deep learning methods for biological image reconstruction and enhancement. IEEE Signal Processing Magazine , 39(2), 28–

    work page 2018

  2. [17]

    Y., Rashid, S., Cheng, Y

    Li, Y. Y., Rashid, S., Cheng, Y. J., Schapiro, W., Gliganic, K., Yamashita, A.-M., . . . others (2018). Real-time cardiac MRI with radial acquisition and k-space variant reduced-FOV reconstruction. Magnetic Resonance Imaging , 53, 98–104. Lin, L., Li, Y., Wang, J., Cao, L., Liu, Y., Pang, J., . . . Wang, Y. (2023). Free-breathing cardiac cine MRI with com...

    work page 2018

  3. [29]

    L., Maspero, M., Verhoeff, J

    Terpstra, M. L., Maspero, M., Verhoeff, J. J. C., & Van Den Berg, C. A. T. (2023). Accelerated respiratory-resolved 4D-MRI with separable spatio-temporal neural networks. Medical Physics , 50(9), 5331–5342. Tian, Y., Lim, Y., Zhao, Z., Byrd, D., Narayanan, S., & Nayak, K. S. (2021). Aliasing artifact reduction in spiral real-time MRI. Magnetic Resonance i...

    work page 2023

  4. [34]

    Retrieved from http:// dx.doi.org/10.1186/1532-429X-16-S1-P34 Van Amerom, J. F. P., Lloyd, D. F. A., Price, A. N., Kuklisova Murgasova, M., Aljabar, P., Malik, S. J., . . . others (2018). Fetal cardiac cine imaging using highly accelerated dynamic MRI with retrospective motion correction and outlier rejection. Magnetic Resonance in Medicine , 79(1), 327–3...

    work page Pith review arXiv 2018

  5. [39]

    M., Lai, P., Vasanawala, S

    Sandino, C. M., Lai, P., Vasanawala, S. S., & Cheng, J. Y. (2021). Accelerating cardiac cine MRI using a deep learning-based ESPIRiT reconstruction. Magnetic Resonance in Medicine , 85(1), 152–167. Schlemper, J., Caballero, J., Hajnal, J. V., Price, A. N., & Rueckert, D. (2017). A deep cascade of convolutional neural networks for dynamic MR image reconstr...

    work page 2021

  6. [41]

    D., Kouwenhoven, M., Clemence, M.,

    G´ omez-Talavera, S., Fernandez-Jimenez, R., Fuster, V., Nothnagel, N. D., Kouwenhoven, M., Clemence, M., . . . others (2021). Clinical validation of a 3-dimensional ultrafast cardiac magnetic resonance protocol including single breath-hold 3-dimensional sequences. JACC Cardiovasc Imaging , 14(9), 1742–1754. Griswold, M. A., Kannengiesser, S., Heidemann, ...

    work page 2021

  7. [44]

    P., Breunig, F.,

    Beer, M., Stamm, H., Machann, W., Weng, A., Goltz, J. P., Breunig, F., . . . K¨ ostler, H. (2010). Free breathing cardiac real-time cine MR without ECG triggering. International Journal of Cardiology , 145(2), 380–382. Blaimer, M., Breuer, F. A., Jacob, P. M., Kellman, P., & Griswold, M. A. (2008). A sparse TSENSE approach for improved dynamic parallel MR...

    work page 2010

  8. [313]

    Zhang, Y., Stolt-Ans´ o, N., Pan, J., Huang, W., Hammernik, K., & Rueckert, D. (2024). Direct car- diac segmentation from undersampled k-space using transformers. In Proceedings of the ieee 21st international symposium on biomedical imaging (ISBI) (pp. 1–4). Zhu, Y., Cheng, J., Cui, Z.-X., Zhu, Q., Ying, L., & Liang, D. (2023). Physics-driven deep learnin...

    work page 2024

Show all 12 references
  1. [334]

    Otazo, R., Candes, E., & Sodickson, D. K. (2015). Low-rank plus sparse matrix decomposition for accelerated dynamic MRI with separation of background and dynamic components. Magnetic Resonance in Medicine , 73(3), 1125–1136. Otazo, R., Kim, D., Axel, L., & Sodickson, D. K. (20...

    2015

  2. [359]

    Yaman, B., Gu, H., Hosseini, S. A. H., Demirel, ¨O. B., Moeller, S., Ellermann, J., . . . Ak¸ cakaya, M. (2022, Dec). Multi-mask self-supervised learning for physics-guided neural networks in highly accelerated magnetic resonance imaging. Nuclear Magnetic Resonance in Biomedic...

    2020

  3. [1270]

    A., Kellman, P., Griswold, M

    Breuer, F. A., Kellman, P., Griswold, M. A., & Jakob, P. M. (2005). Dynamic autocalibrated parallel imaging using temporal GRAPPA (TGRAPPA).Magnetic Resonance in Medicine , 53(4), 981–985. Chubb, H., Harrison, J. L., Weiss, S., Krueger, S., Koken, P., Bloch, L. Ø., . . . Razav...

    2005

  4. [1371]

    P., Sodickson, D

    Hammernik, K., Klatzer, T., Kobler, E., Recht, M. P., Sodickson, D. K., Pock, T., & Knoll, F. (2018). Learning a variational network for reconstruction of accelerated MRI data. Magnetic Resonance in Medicine, 79(6), 3055–3071. Hammernik, K., K¨ ustner, T., Yaman, B., Huang, Z....

    2018

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.