Pith. sign in

REVIEW 4 major objections 5 minor 33 references

Deep learning for temporal super-resolution 4D Flow MRI

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read A residual CNN temporally super-resolves 4D Flow MRI, recovering flow peaks that linear and sinc interpolation miss.

desk verdict A solid proof-of-principle for temporal super-resolution in 4D Flow MRI that overclaims in the abstract but is worth engaging with after adding in-vivo baselines. read the letter →

arxiv 2501.08780 v1 pith:YV6SFDP3 submitted 2025-01-15 cs.LG

Pia Callmer , Mia Bonini , Edward Ferdian , David Nordsletten , Daniel Giese , Alistair A. Young , Alexander Fyrdahl , David Marlevi This is my paper · ORCID
classification cs.LG
keywords temporalsuper-resolution4DFlowMRIresidualneuralnetworkdeeplearningcompressedsensingcomputationalfluiddynamicscardiacquantificationdenoising
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 asks whether a convolutional network trained on synthetic data can increase the temporal resolution of 4D Flow MRI, a technique whose clinical use is limited by the trade-off between acquisition time, noise, and frame rate. The authors adapt a spatial super-resolution network, 4DFlowNet, to upsample in time using 2D+t patches, and train it on low-resolution inputs generated from patient-specific computational fluid dynamics via a pipeline that mimics real MRI k-space sampling, noise, and compressed-sensing reconstruction. On a held-out in-silico model the network reaches a mean absolute error of 1.0 cm/s versus 2.3–2.6 cm/s for linear and sinc interpolation, and on two in-vivo cohorts it synthesizes missing frames with linear-regression slopes around 0.9 at peak systole. The point of the result is that high-frame-rate volumetric flow quantification could be obtained from acquisitions of roughly half the length, without patient-specific retraining.

What carries the argument

The load-bearing mechanism is a residual convolutional network derived from 4DFlowNet, in which the central upsampling layer was changed from trilinear to linear to increase the frame count only along the temporal axis; inputs are 16x16x16 patches of 2D spatial slices sampled over 16 time frames, and outputs are 16x16x32 patches at double the frame rate. The loss combines an MSE term computed separately over fluid and non-fluid voxels with the mutually projected loss, an angle- and magnitude-aware L1 measure from SRflow, plus an L2 weight regularizer. The second half of the machinery is the data-generation pipeline: CFD velocities are converted to phase-MRI signals, multiplied by simulated Biot-Savart coil sensitivities, corrupted with k-space noise, subsampled with a variable-density phyllotaxis pattern that accumulates consecutive frames to emulate temporal undersampling, and reconstructed by compressed sensing with FISTA in BART. That pipeline is what gives the network inputs whose temporal-smoothing and noise artifacts resemble clinically acquired 4D Flow MRI.

What would settle it

A concrete disconfirming experiment would be to acquire paired 20 ms and 40 ms 4D Flow MRI in patients with cardiac pathology outside the mitral regurgitation cohort and compute the network's RMSE on frames during late diastole or atrial contraction; if the error approaches or exceeds the linear-interpolation error on those frames, the claim of generalizable temporal super-resolution is refuted.

Watch

Extended reading notes

Core claim

The paper's central claim is that temporal super-resolution of 4D Flow MRI is tractable with a residual CNN: given three-dimensional spatial slices sampled over time at low frame rate, the network outputs denoised velocities at double the frame rate, synthesizing information that is absent in the low-resolution input. In the in-silico test the super-resolved velocities match the high-resolution reference with an average RMSE of 1.4 cm/s inside the fluid region (and 0.2 cm/s outside), against 3.0 cm/s for linear and 3.8 cm/s for sinc interpolation. In-vivo, the network recovers synthetically removed frames in five subjects with k=0.93 and $R^{2}$=0.89 at peak systole, and in a paired low/high-rate acquisition in five further subjects it super-resolves 40 ms input to 20 ms reference data with k=0.90 and $R^{2}$=0.95 in the aorta. The authors frame the work as a first-of-a-kind temporal super-resolution baseline for 4D Flow MRI that generalizes across anatomies without retraining.

Load-bearing premise

The load-bearing premise is that the synthetic pipeline produces low-resolution 4D Flow MRI inputs whose noise and undersampling artifacts match real clinical acquisitions, so that in-vivo performance reflects true recovery rather than the network learning to denoise generic flow patterns.

Editorial extensions

If this is right

  • Post-processing could raise the effective temporal resolution of existing 4D Flow MRI datasets from roughly 40 ms toward 20 ms without rescanning, since the network generates intermediate frames.
  • Peak-flow estimates (early diastole and peak systole) become more accurate: the in-silico RMSE at those frames is 1.5–1.7 cm/s, several-fold lower than for interpolation.
  • The network simultaneously denoises the input, reducing error in non-fluid regions to 0.2 cm/s, which could improve downstream segmentation and visualization.
  • The method requires no patient-specific retraining, no predefined fluid-region segmentation, and runs patch-based inference with stitching, so it can be applied to retrospective clinical data.
  • Generalization across two in-vivo protocols with different spatial resolutions and temporal resolutions indicates the network is not overfit to a single acquisition sequence.

Reading between the lines

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

  • We infer that the temporal smoothing in current 4D Flow MRI is partly an artifact of undersampling and reconstruction rather than truly absent information, since a network trained on realistic synthetic data reconstructs peaks not present in the low-rate input.
  • A testable extension is to ablate the synthetic pipeline stages (coil sensitivities, phyllotaxis sampling, compressed sensing, temporal accumulation) to determine which component is responsible for the in-vivo generalization; the paper only notes that removing the realistic pipeline degraded performance, not which stage matters most.
  • The focus on left-ventricular geometries with mitral regurgitation suggests the approach may need additional training data from other cardiovascular compartments before it becomes a general clinical tool, and the method could be extended to those domains using the same pipeline.
  • If the result transfers, the same 2D+t residual design could be applied to other time-resolved flow or cine imaging modalities, with the main challenge being construction of a similarly realistic synthetic training pipeline.
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.

Referee Report

4 major / 5 minor

Summary. The paper presents a residual convolutional network for 2x temporal super-resolution of 4D Flow MRI, adapting the spatial network 4DFlowNet to operate on 2D+t patches. Training uses paired synthetic data generated from six patient-specific CFD models of the left heart, processed through a pipeline that simulates coil sensitivities, k-space noise, phyllotaxis undersampling, temporal subsampling, and compressed-sensing reconstruction. The loss combines a fluid/non-fluid MSE term with a mutually projected directional L1 loss. Evaluation is performed on one held-out in-silico model and on two in-vivo cohorts (each n=5): one in which low-resolution input is created by removing every second frame from native-resolution data, and one with paired acquisitions at 20 ms and 40 ms temporal resolution. The authors report in-silico RMSE of 1.4 cm/s versus 3.0 cm/s (linear) and 3.8 cm/s (sinc), and in-vivo peak-flow correlations of k=0.93 with R^2=0.89 for recovery-to-native and k=0.90 with R^2=0.95 for paired aortic data.

Significance. If the claims are fully supported, this would be a useful first demonstration of a CNN for temporal super-resolution of 4D Flow MRI that generalizes without patient-specific retraining. The strengths include public release of code and training details, a comparatively realistic synthetic-data pipeline (Section 2.2.2), and an in-silico benchmark against deterministic linear and sinc interpolation that shows a substantial improvement in RMSE and correlation. The main unresolved issue is the strength of the in-vivo evidence: because the in-vivo evaluations omit the same deterministic baselines used in-silico, and because the synthetic-to-real degradation gap is not directly quantified, the paper's central clinical claim is not yet established at the level the abstract implies.

major comments (4)
  1. [§3.2, Tables 3–4] The in-vivo evaluations contain no linear or sinc interpolation baselines applied to the same low-resolution inputs. In the absence of such baselines, the reported peak-flow correlations (k=0.90–0.93, R^2=0.89–0.95) could be produced by a network that mainly denoises and linearly interpolates, without recovering temporal information beyond interpolation. Please add deterministic interpolation results for the recovery-to-native and paired in-vivo datasets, and report the same metrics (RMSE, k, R^2) for those baselines.
  2. [§2.2.4 and §3.2.1] The recovery-to-native evaluation constructs low-resolution input by removing every second frame from already reconstructed native-resolution data. This does not reproduce the temporal footprint that the synthetic pipeline in §2.2.2 is designed to emulate, where temporal subsampling occurs through k-space readout accumulation followed by compressed-sensing reconstruction. The test therefore measures frame interpolation of reconstructed frames rather than recovery from an acquisition-like temporal degradation. Please either create low-resolution in-vivo inputs with a comparable temporal footprint (e.g., retrospective temporal binning or re-sampling of k-t data) or explicitly quantify and discuss the effect of this mismatch on the reported generalization.
  3. [§2.2.3 and §3.1] The in-silico test set consists of a single held-out CFD model, even though it is divided into many patches. The claim of generalization to an 'unseen in-silico setting' is therefore based on n=1 at the model level. Please report results for multiple held-out models or provide model-level confidence intervals, and discuss how patch-level statistics may overstate the diversity of the test set.
  4. [§3.2.2 and §4.5] The paired in-vivo evaluation compares super-resolved data against a separately acquired high-resolution dataset that is itself noisy and may differ in heart rate, breathing, and subject position. The manuscript acknowledges this in §4.5, but the quantitative metrics are presented as if the high-resolution reference were a gold standard. Please report the estimated noise level of the high-resolution reference (e.g., SNR or test–retest variability) and discuss how reference noise bounds the achievable k and R^2 values.
minor comments (5)
  1. [Figure 1 caption] The caption lists the input velocities as '˜vx, ˜vx, ˜vx' and the outputs as 'ˆvx, ˆvx, ˆvx'; these should be vx, vy, vz and vx, vy, vz respectively.
  2. [§2.3.3, Eq. (1), Table 2] The relationship between the relative error RE defined in Eq. (1) and the MRE values reported in Table 2 is not specified. Please state explicitly whether MRE is the average of RE over voxels and time, and how the percentage is obtained.
  3. [Table 2] RMSE values are reported as triples for the three velocity components while MAE and MRE are single numbers. Please specify whether MAE is computed on velocity magnitude or averaged over components, so the table is self-consistent.
  4. [§4.2] The preliminary comparison between the proposed pipeline and a simpler 'remove every second frame plus k-space noise' pipeline is described qualitatively as improving in-vivo performance. Please provide quantitative results for this comparison or clearly mark it as anecdotal.
  5. [§2.2.2] The text states that target SNR is 'randomly assigned between 14-17 dB or 40-45 db (two separate datasets)', but the validation and test sets are described as high-noise only. Please clarify which SNR ranges are used for training versus validation/test and whether the test model was exclusively high-noise.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: training and evaluation are held-out and externally benchmarked, so the temporal super-resolution claim is not equivalent to its inputs by construction.

full rationale

The paper's derivation chain is a standard supervised-learning evaluation rather than a circular reduction. Paired low- and high-resolution data are synthesized from patient-specific CFD simulations through a dedicated MRI degradation pipeline (Section 2.2.2), and the network is trained with a fixed loss (Section 2.1.2) on a training split separated from validation and test at the model level (Section 2.2.3). The central performance claims are measured on an unseen in-silico test model and on two in-vivo cohorts not used in training, with metrics computed against high-resolution references that are not fitted parameters of the model. The use of 4DFlowNet [7,18] as a base architecture and the adoption of a previously proposed mutually projected loss [19] are technical inheritances that do not smuggle in the target result; no uniqueness theorem or author-specific ansatz is invoked to force the outcome. The in-vivo 'recovery to native resolution' test does construct its low-resolution input by removing every second frame, but the network has no exposure to those in-vivo targets during training, so comparing its output to the removed frames is a legitimate generalization test rather than a fitted prediction. The absence of deterministic baselines in the in-vivo comparisons and the noise in the high-resolution in-vivo reference are validity limitations, not circular reductions.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No new physical entities are introduced. The central claim rests on the realism of the synthetic training pipeline and the validity of the in-vivo reference data.

free parameters (5)
  • alpha (loss balance) = 0.8
    Relative weight of MSE vs directional loss in L_total; selected by hand without sensitivity analysis.
  • beta (directional loss balance) = 0.5
    Balance between two projected L1 losses in the mutually projected loss; selected by hand.
  • lambda_nn (weight decay) = 5e-7
    L2 regularization on network weights to prevent overfitting; selected by hand.
  • Target SNR ranges = 14-17 dB and 40-45 dB
    Random signal-to-noise ratios for synthetic training data to mimic realistic acquisitions; a modeling choice affecting training distribution.
  • Minimum fluid content per patch = 20%
    Patches with at least 20% fluid voxels over time are selected, with one low-fluid patch per iteration; this shapes the training distribution.
assumptions (4)
  • domain assumption CFD-simulated flow from six patient-specific geometries is representative of human left heart hemodynamics.
    Section 2.2.1. If the simulated flows do not match real hemodynamics, the learned temporal mapping may not transfer to patients.
  • domain assumption The synthetic MRI pipeline (coil sensitivities, k-space sampling, compressed sensing reconstruction) produces low-resolution inputs with realistic noise and artifacts.
    Section 2.2.2. The generalization claim depends on the synthetic inputs being similar to real 4D Flow MRI data.
  • domain assumption A 2D+t patch network with averaging across Cartesian directions captures temporally super-resolved flow without loss of critical spatial coherence.
    Section 2.1.1 and Section 4.5. The authors explicitly acknowledge the discard of some spatial information and mitigate by averaging.
  • domain assumption The high-resolution in-vivo acquisition (20 ms) is a valid reference for evaluating super-resolved frames from 40 ms input.
    Section 2.2.4 and Section 4.5. Paired acquisitions are sequential and may differ in heart rate, motion, and noise.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep learning for temporal super-resolution 4D Flow MRI." pith.science (2026). https://pith.science/paper/YV6SFDP3

@misc{pith2026250108780,
  author       = {Pith},
  title        = {Pith review of: Deep learning for temporal super-resolution 4D Flow MRI},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YV6SFDP3}},
  note         = {Machine review of arXiv:2501.08780}
}
read the original abstract

4D Flow Magnetic Resonance Imaging (4D Flow MRI) is a non-invasive technique for volumetric, time-resolved blood flow quantification. However, apparent trade-offs between acquisition time, image noise, and resolution limit clinical applicability. In particular, in regions of highly transient flow, coarse temporal resolution can hinder accurate capture of physiologically relevant flow variations. To overcome these issues, post-processing techniques using deep learning have shown promising results to enhance resolution post-scan using so-called super-resolution networks. However, while super-resolution has been focusing on spatial upsampling, temporal super-resolution remains largely unexplored. The aim of this study was therefore to implement and evaluate a residual network for temporal super-resolution 4D Flow MRI. To achieve this, an existing spatial network (4DFlowNet) was re-designed for temporal upsampling, adapting input dimensions, and optimizing internal layer structures. Training and testing were performed using synthetic 4D Flow MRI data originating from patient-specific in-silico models, as well as using in-vivo datasets. Overall, excellent performance was achieved with input velocities effectively denoised and temporally upsampled, with a mean absolute error (MAE) of 1.0 cm/s in an unseen in-silico setting, outperforming deterministic alternatives (linear interpolation MAE = 2.3 cm/s, sinc interpolation MAE = 2.6 cm/s). Further, the network synthesized high-resolution temporal information from unseen low-resolution in-vivo data, with strong correlation observed at peak flow frames. As such, our results highlight the potential of utilizing data-driven neural networks for temporal super-resolution 4D Flow MRI, enabling high-frame-rate flow quantification without extending acquisition times beyond clinically acceptable limits.

Figures

Figures reproduced from arXiv: 2501.08780 by the authors.

Figure 1
Figure 1. Illustration of the proposed temporal super-resolution network for 4D Flow MRI, based on the previously described 4DFlowNet [18]. The residual convolutional neural network takes a sequence of 2D low resolution, noisy input velocities v˜x, v˜x, v˜x, magnitude, speed and a phase￾contrast magnetic resonance angiogram (PC-MRA) mask, and outputs denoised and super-resolved velocities vˆx, vˆx, vˆx. where α ∈ [0, 1] deter… view at source ↗
Figure 2
Figure 2. Overview of in-silico data and patch generation for one of six data models. Left to right: voxelized geometry (2 mm resolution), velocity field (0–68 cm/s), and patch selection. Patches generate 2D slice sequences (2D+t) in all three Cartesian directions across multiple time frames. into realistic synthetic 4D Flow MRI datasets to use as low reso￾lution input data. This was performed to closely mimic the char￾acteri… view at source ↗
Figure 3
Figure 3. Overview of the synthetic MRI data preparation pipeline. (A) Combination of phase and magnitude images into a complex image. (B) Coil sensitivity maps are generated from Biot-Savart simulations, each of which is multiplied by the complex image. (C) Application of FFT and addition of complex Gaussian noise in the k-space. (D) Sampling of frequency domain data with a Cartesian variable-density phyllotaxis sampling pat… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: In-silico evaluation on the test set. From left to right: (A) Comparison of Vy between low resolution (LR), high-resolution (HR), super￾resolution (SR) and sinc interpolation during a few selected time frames at early diastole. (B) Linear regression plot for one of the…
Figure 5
Figure 5. Figure 5: Flow evaluation over the defined aortic and mitral valve planes of the in-silico test set. Qualitative evaluation comparing LR, HR and SR results for aortic and mitral valve planes (A, C). Mean velocity [m/s] across outlets plotted as a function of time (B, D). contain…
Figure 6
Figure 6. Figure 6: Qualitative comparison of Vx for subject A1 between LR, HR and SR in peak systolic frames and the correlation plots between SR and HR for each synthetic frame prediction. The correlation plots include all voxels within the left ventricle and aorta. in Fig. 4B-C, where …
Figure 7
Figure 7. Figure 7: Qualitative comparison of velocity Vx for subject B1 between LR, HR and SR in peak systolic frames and correlation plots between SR and HR, including points from the entire fluid region of the left ventricle and aorta [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Mean velocity through planes at both ascending (A) and descending aorta (B) comparing LR, HR and SR of volunteer B1, B3 and B5. Each plane has a thickness of around 8-10 mm to account for pixel shifts and mimic the slice thickness in 2D flow measurements of aortic flow…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 33 canonical work pages

  1. [1]

    Yoram Richter and Elazer R. Edelman. Cardiology is flow. Circulation, 113(23):2679–2682, 6 2006

    work page 2006

  2. [2]

    Bolger, Einar Heiberg, Matts Karlsson, Lars Wigstr¨om, Jan Engvall, Andreas Sigfridsson, Tino Ebbers, John Peder Escobar Kvitting, Carl Johan Carlh ¨all, and Bengt Wranne

    Ann F. Bolger, Einar Heiberg, Matts Karlsson, Lars Wigstr¨om, Jan Engvall, Andreas Sigfridsson, Tino Ebbers, John Peder Escobar Kvitting, Carl Johan Carlh ¨all, and Bengt Wranne. Transit of blood flow through the hu- man left ventricle mapped by cardiovascular magnetic res- onance. Journal of Cardiovascular Magnetic Resonance , 9(5):741–747, 9 2007

    work page 2007

  3. [3]

    4D Flow MRI

    Michael Markl, Alex Frydrychowicz, Sebastian Kozerke, Mike Hope, and Oliver Wieben. 4D Flow MRI. Journal of Magnetic Resonance Imaging, 2012

    work page 2012

  4. [4]

    The role of 4D flow MRI for clinical applications in cardiovascular disease: current status and future per- spectives

    Baiyan Zhuang, Arlene Sirajuddin, Shihua Zhao, and Min- jie Lu. The role of 4D flow MRI for clinical applications in cardiovascular disease: current status and future per- spectives. Quantitative Imaging in Medicine and Surgery , 11(9):4193, 9 2021

    work page 2021

  5. [5]

    4D Flow MRI quantification of blood flow patterns, turbulence and pressure drop in normal and stenotic prosthetic heart valves.Magnetic Resonance Imag- ing, 55:118–127, 1 2019

    Hojin Ha, John Peder Escobar Kvitting, Petter Dyverfeldt, and Tino Ebbers. 4D Flow MRI quantification of blood flow patterns, turbulence and pressure drop in normal and stenotic prosthetic heart valves.Magnetic Resonance Imag- ing, 55:118–127, 1 2019

    work page 2019

  6. [6]

    Vasanawala, Kate Hanneman, Marcus T

    Shreyas S. Vasanawala, Kate Hanneman, Marcus T. Alley, and Albert Hsiao. Congenital heart disease assessment with 4D flow MRI. Journal of Magnetic Resonance Imaging , 42(4):870–886, 10 2015

    work page 2015

  7. [7]

    Cerebrovascu- lar super-resolution 4D Flow MRI-using deep learning to non-invasively quantify velocity, flow, and relative pressure Cerebrovascular super-resolution 4D Flow MRI

    Edward Ferdian, David Marlevi, J Schollenberger, M Aris- tova, E R Edelman, S Schnell, C A Figueroa, D A Nord- sletten, A A Young, and D Al Marlevi. Cerebrovascu- lar super-resolution 4D Flow MRI-using deep learning to non-invasively quantify velocity, flow, and relative pressure Cerebrovascular super-resolution 4D Flow MRI. Medical Image Analysis, 88:102...

    work page 2023

  8. [8]

    Russe, James Carr, Wulf Euringer, Lena Stehlin, Zoltan Csatari, Peter C

    Zoran Stankovic, Bernd Jung, Jeremy Collins, Maximil- ian F. Russe, James Carr, Wulf Euringer, Lena Stehlin, Zoltan Csatari, Peter C. Strohm, Mathias Langer, and Michael Markl. Reproducibility study of four-dimensional flow MRI of arterial and portal venous liver hemodynam- ics: Influence of spatio-temporal resolution. Magnetic Res- onance in Medicine, 72...

    work page 2014

Show all 33 references
  1. [9]

    Marcus Carlsson, Johannes T ¨oger, Mikael Kanski, Karin Markenroth Bloch, Freddy St˚ahlberg, Einar Heiberg, and H ˚akan Arheden. Quantification and visualization of cardiovascular 4D velocity mapping accelerated with par- allel imaging or k-t BLAST: Head to head comparison and...

    2011

  2. [10]

    Nguyen, Rachel Wald, Andreas Greiser, Andrew M

    Kate Hanneman, Milani Sivagnanam, Elsie T. Nguyen, Rachel Wald, Andreas Greiser, Andrew M. Crean, Sebas- tian Ley, and Bernd J. Wintersperger. Magnetic resonance assessment of pulmonary (QP) to systemic (QS) flows us- ing 4D phase-contrast imaging: pilot study comparison with ...

    2014

  3. [11]

    Ramos, Maria J

    Alexander Fyrdahl, Joao G. Ramos, Maria J. Eriksson, Kenneth Caidahl, Martin Ugander, and Andreas Sigfrids- son. Sector-wise golden-angle phase contrast with high temporal resolution for evaluation of left ventricular di- astolic dysfunction. Magnetic Resonance in Medicine , 8...

    2020

  4. [12]

    Blood flow in the human ascending aorta: A combined MRI and CFD study

    Armin Leuprecht, Sebastian Kozerke, Peter Boesiger, and Karl Perktold. Blood flow in the human ascending aorta: A combined MRI and CFD study. Journal of Engineering Mathematics, 47(3-4):387–404, 12 2003

    2003

  5. [13]

    Bonini, M

    M. Bonini, M. Hirschvogel, Y . Ahmed, H. Xu, A. Young, P.C. Tang, and D. Nordsletten. Hemodynamic Modeling for Mitral Regurgitation. The Journal of Heart and Lung Transplantation, 41(4):S218, 4 2022

    2022

  6. [14]

    Barker, Ann F

    Petter Dyverfeldt, Malenka Bissell, Alex J. Barker, Ann F. Bolger, Carl Johan Carlh ¨all, Tino Ebbers, Christopher J. Francios, Alex Frydrychowicz, Julia Geiger, Daniel Giese, Michael D. Hope, Philip J. Kilner, Sebastian Kozerke, Saul Myerson, Stefan Neubauer, Oliver Wieben, a...

    2015

  7. [15]

    Image Super-Resolution Using Deep Convolutional Networks

    Chao Dong, Chen Change Loy, Kaiming He, and Xiaoou Tang. Image Super-Resolution Using Deep Convolutional Networks. IEEE Transactions on Pattern Analysis and Ma- chine Intelligence, 38(2):295–307, 12 2014

    2014

  8. [16]

    Implicit neural representations for unsupervised super-resolution and denoising of 4D flow MRI

    Simone Saitta, Marcello Carioni, Subhadip Mukherjee, Carola-Bibiane Sch ¨onlieb, and Alberto Redaelli. Implicit neural representations for unsupervised super-resolution and denoising of 4D flow MRI. Computer Methods and Programs in Biomedicine, (246):108057, 4 2024

    2024

  9. [17]

    Fathi, Isaac Perez-Raya, Ahmadreza Baghaie, Philipp Berg, Gabor Janiga, Amirhossein Arzani, and Roshan M

    Mojtaba F. Fathi, Isaac Perez-Raya, Ahmadreza Baghaie, Philipp Berg, Gabor Janiga, Amirhossein Arzani, and Roshan M. D’Souza. Super-resolution and denoising of 4D-Flow MRI using physics-Informed deep neural nets. Computer Methods and Programs in Biomedicine, 197, 12 2020

    2020

  10. [18]

    Dubowitz, Debbie Zhao, Alan Wang, Brett Cowan, and Alistair A

    Edward Ferdian, Avan Suinesiaputra, David J. Dubowitz, Debbie Zhao, Alan Wang, Brett Cowan, and Alistair A. Young. 4DFlowNet: Super-Resolution 4D Flow MRI Us- ing Deep Learning and Computational Fluid Dynamics. Frontiers in Physics, 8:138, 5 2020

    2020

  11. [19]

    Paetzold, Augusto F

    Suprosanna Shit, Judith Zimmermann, Ivan Ezhov, Jo- hannes C. Paetzold, Augusto F. Sanches, Carolin Pirkl, and Bjoern H. Menze. SRflow: Deep learning based super- resolution of 4D-flow MRI data. Frontiers in Artificial In- telligence, 5:171, 8 2022

    2022

  12. [20]

    Mauger, David Marlevi, Martyn P

    Derek Long, Cameron McMurdo, Edward Ferdian, Charl`ene A. Mauger, David Marlevi, Martyn P. Nash, and Alistair A. Young. Super-resolution 4D flow MRI to quan- tify aortic regurgitation using computational fluid dynamics and deep learning. International Journal of Cardiovascular...

    2023

  13. [21]

    Alberto Figueroa, David Nordsletten, Alis- tair A

    Leon Ericsson, Adam Hjalmarsson, Muhammad Usman Akbar, Edward Ferdian, Mia Bonini, Brandon Hardy, Jonas Schollenberger, Maria Aristova, Patrick Winter, Nicholas Burris, Alexander Fyrdahl, Andreas Sigfridsson, Susanne Schnell, C. Alberto Figueroa, David Nordsletten, Alis- tair ...

    2024

  14. [22]

    Adam: A Method for Stochastic Optimization

    Diederik P Kingma and Jimmy Lei Ba. Adam: A Method for Stochastic Optimization. In International Conference on Learning Representations, 2015

    2015

  15. [23]

    Christoph Forman, Davide Piccini, Robert Grimm, Jana Hutter, Joachim Hornegger, and Michael O. Zenge. High- resolution 3D whole-heart coronary MRA: a study on the combination of data acquisition in multiple breath- holds and 1D residual respiratory motion compensation. Magneti...

    2014

  16. [24]

    Michael Lustig, David Donoho, and John M. Pauly. Sparse MRI: The application of compressed sensing for rapid MR imaging. Magnetic Resonance in Medicine , 58(6):1182– 1195, 12 2007

    2007

  17. [25]

    A fast iterative shrinkage- thresholding algorithm for linear inverse problems

    Amir Beck and Marc Teboulle. A fast iterative shrinkage- thresholding algorithm for linear inverse problems. SIAM Journal on Imaging Sciences, 2(1), 2009

    2009

  18. [26]

    Martin Uecker, F Ong, JI Tamir, D Bahri, P Virtue, JY Cheng, T Zhang, and M. Lustig. Berkeley Advanced Reconstruction Toolbox. In ISMRM, 2015

    2015

  19. [27]

    Andia, Cristian Tejos, Pablo Irarrazaval, Daniel E

    Cristian Montalba, Jesus Urbina, Julio Sotelo, Marcelo E. Andia, Cristian Tejos, Pablo Irarrazaval, Daniel E. Hurtado, Israel Valverde, and Sergio Uribe. Variability of 4D flow parameters when subjected to changes in MRI acquisition parameters using a realistic thoracic aortic...

    2018

  20. [28]

    Karolina Dorniak, Einar Heiberg, Marcin Hellmann, Dorota Rawicz-Zegrzda, Maria Wesierska, Rafal Galaska, Agnieszka Sabisz, Edyta Szurowska, Maria Dudziak, and Erik Hedstr ¨om. Required temporal resolution for accurate thoracic aortic pulse wave velocity measurements by phase- ...

    2016

  21. [29]

    Park, Diogo C

    Retta El Sayed, Alireza Sharifi, Charlie C. Park, Diogo C. Haussen, Jason W. Allen, and John N. Oshinski. Optimiza- tion of 4D Flow MRI Spatial and Temporal Resolution for Examining Complex Hemodynamics in the Carotid Artery Bifurcation. Cardiovascular Engineering and Technolo...

    2023

  22. [30]

    Jagtap and George Em Karniadakis

    Ameya D. Jagtap and George Em Karniadakis. Extended physics-informed neural networks (XPINNs): A general- ized space-time domain decomposition based deep learn- ing framework for nonlinear partial differential equations. Communications in Computational Physics , 28(5):2002– 20...

    2002

  23. [31]

    Locality-Aware Generalizable Implicit Neural Rep- resentation

    Doyup Lee, Chiheon Kim, Minsu Cho, and Wook-Shin Han. Locality-Aware Generalizable Implicit Neural Rep- resentation. Advances in Neural Information Processing Systems, 36:48363–48381, 12 2023

    2023

  24. [32]

    Long Short- Term Memory

    Sepp Hochreiter and J ¨urgen Schmidhuber. Long Short- Term Memory. Neural Computation, 9(8):1735–1780, 11 1997

    1997

  25. [33]

    Video Frame Interpolation With Transformer

    Liying Lu, Ruizheng Wu, Huaijia Lin, Jiangbo Lu, and Ji- aya Jia. Video Frame Interpolation With Transformer. In 2022 IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 3522–3532, 2022

    2022

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.