Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

A single free-running 3D CMR acquisition can be reconstructed into continuous beat-to-beat ventricular volumes and ejection fractions, exposing functional heterogeneity in arrhythmia that standard gated imaging averages away.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-02 18:51 UTC pith:NOTOZZCE

load-bearing objection A genuinely new 3D real-time CMR method, but the abstract and full text disagree on cohort size and phantom EF — fix that before trusting the numbers. the 4 major comments →

arxiv 2603.04233 v2 pith:NOTOZZCE submitted 2026-03-04 physics.med-ph

Feasibility of Continuous Ventricular Volumetric Quantification in Arrhythmias using Real-Time 3D CMR-MOTUS

Thomas E. Olausson , Maarten L. Terpstra , Rizwan Ahmad , Edwin Versteeg , Casper Beijst , Yuchi Han , Marco Guglielmo , Birgitta K. Velthuis
show 2 more authors
Cornelis van den Berg Alessandro Sbrizzi
This is my paper
classification physics.med-ph
keywords free-running CMRreal-time 3D MRImotion fieldspremature ventricular contractionsbeat-to-beat ejection fractionvolumetric quantificationCMR-MOTUSarrhythmia imaging
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper claims that by jointly reconstructing a motion-corrected reference image and time-resolved 3D motion fields from a single free-running MRI scan—no ECG gating, no breath-holding—one can propagate a single segmentation through every cardiac cycle and obtain continuous beat-to-beat ventricular volumes and ejection fractions. In healthy volunteers this gives narrow EF distributions, while patients with premature ventricular contractions show broad, sometimes bimodal EF distributions where the low mode corresponds to PVC beats. If true, conventional gated or binned CMR, which averages many heartbeats, misses clinically relevant functional heterogeneity, and a single 2-minute scan could become a per-beat functional histogram.

Core claim

The paper extends CMR-MOTUS to 3D: from roughly two minutes of continuously acquired, ungated, free-breathing Cartesian data, it jointly reconstructs a motion-corrected reference image and a rank-16, time-smooth 3D motion field at about 16-21 Hz. By propagating one manual segmentation of the reference image through the motion fields, the authors obtain continuous volume curves and per-beat EF. Phantom EF agrees with ground truth (22.1% +/- 0.6% vs 21.9%), healthy volunteers show narrow EF distributions, and PVC patients show broader, sometimes bimodal distributions whose low-EF mode aligns with PVC episodes on simultaneously recorded ECG. The claim is that this makes beat-to-beat volumetric

What carries the argument

The load-bearing object is the 3D CMR-MOTUS reconstruction: an alternating minimization that alternates between estimating a time-independent motion-corrected reference image q and a low-rank deformation field D = Phi Psi^T, where Phi holds 16 spatial basis components and Psi the temporal basis, with isotropic total-variation spatial regularization, l1 data consistency to tolerate inflow effects, and a 4 Hz Butterworth low-pass filter on the temporal components. The OPRA variable-density Cartesian trajectory provides temporally incoherent sampling without non-Cartesian gridding. Segmentation propagation turns the motion fields into beat-resolved volume curves.

Load-bearing premise

The model assumes every acquired frame is a smooth, low-rank deformation of one time-invariant reference image; contrast changes or arrhythmic motion components outside the rank-16 and 4 Hz subspace will be misattributed to motion and bias the propagated volumes.

What would settle it

Drive a beating phantom with known ground-truth volumes through an abrupt, short contraction (e.g., less than 150 ms duration) and check whether the rank-16, 4 Hz reconstruction recovers the true volume dip; if the recovered EF change is attenuated, the method cannot be trusted for high-frequency arrhythmic beats.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • In arrhythmic patients, a single ~2-minute free-running scan could yield per-beat EF histograms instead of one averaged number, changing how PVC burden and treatment response are assessed.
  • Because the motion fields are explicit, the same acquisition can be reused to compute strain, desynchrony, and regional wall-motion metrics without re-scanning.
  • The approach removes the breath-hold and gating requirements of standard CMR cine, making volumetric function assessment feasible for patients who cannot hold their breath or have highly irregular rhythms.
  • The bimodal EF distribution offers a concrete, quantitative measure of the hemodynamic cost of PVC episodes, which standard 2D analysis cannot capture because arrhythmic beats appear in only some slices.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The in-vivo cohort mixes field strengths, sequences, and contrast agents, so the apparent difference between healthy and PVC distributions could be partly acquisition-related; a matched homogeneous cohort is needed before reading the histograms as purely physiological.
  • A likely failure mode beyond what the paper tests: abrupt, high-frequency contractions near or above the 4 Hz Butterworth cutoff may be smoothed, so the method could understate the true volume drop in very short-coupled PVCs.
  • The single-reference-image assumption means any slow contrast change (e.g., late gadolinium wash-in) would be folded into motion; a time-dependent contrast model may be necessary for non-steady-state contrast protocols.
  • If EF distribution metrics prove reproducible, an 'EF burden' index (e.g., percentage of beats below a threshold) might become a more sensitive outcome measure than mean EF for device and ablation trials.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper extends the CMR-MOTUS framework to 3D for free-running, ECG-free, continuous acquisition, jointly reconstructing a motion-corrected reference image and time-resolved motion fields expressed in a low-rank (rank-16) decomposition with temporal low-pass filtering. Ventricular volumes are obtained by propagating a single manual segmentation along the reconstructed motion fields, enabling beat-to-beat EF curves and histograms. The method is validated on a deformable cardiac phantom with static ground-truth acquisitions and applied to 4 healthy volunteers and 4 PVC patients (the abstract instead states 10+10). The authors report close phantom agreement (22.1±0.6% vs 21.9%), narrow beat-to-beat EF distributions in healthy subjects, bimodal EF distributions in PVC patients, and qualitative ECG correlation linking volume dips to PVC episodes. The central claim is that this workflow can quantify continuous volumetric function in arrhythmia where conventional binning and gating fail.

Significance. If the central claim is established, the method would be a meaningful step toward 3D real-time volumetric assessment in arrhythmia, with interpretable motion fields enabling downstream deformation metrics from a single free-running scan. The explicit low-rank motion model and joint reconstruction are technically sound and, unlike black-box deep-learning approaches, provide interpretable motion fields. The manuscript also candidly discloses limitations (small cohort, 2-hour reconstruction, scanner/protocol heterogeneity). The phantom validation and healthy-volunteer results support the feasibility of the motion-field-propagation workflow for regular, periodic cardiac motion. However, the evidence for the specific arrhythmic/PVC claim is incomplete: the 2D reference analysis excludes PVC beats, the phantom does not reproduce PVC dynamics, and the low-rank/4-Hz motion model is not independently validated for these events. The abstract/full-text numerical inconsistencies further reduce confidence in the reported results.

major comments (4)
  1. [Abstract vs Methods/Results] The abstract reports 10 healthy volunteers and 10 PVC patients and a phantom EF of 17.86% versus 17.27%, while the Methods and Results sections report 4 healthy volunteers, 4 PVC patients, and phantom EF of 22.1±0.6% versus 21.9%. These are not minor typographical differences; they change the evidence base and the numerical result that is used as primary validation. The authors must reconcile the numbers and state which cohort and phantom values are correct.
  2. [Methods, 'In Vivo Study' / Results, Figure 8] The in-vivo validation for PVC patients is not a validation of PVC-beat volumes. The 2D real-time cine reference was deliberately limited to dominant sinus contractions ('For PVC patients, the arrhythmic beats were not captured consistently in every slice; therefore, the analysis was limited to dominant sinus contractions'). Thus the lower mean EF measured by the proposed method in subjects 5, 7, and 8 is expected by construction and cannot distinguish an accurate inclusion of PVC beats from an artifact. The manuscript should either provide an independent beat-resolved gold standard for PVC beats or explicitly restrict the claim to feasibility, with the PVC results treated as illustrative rather than quantitatively validated.
  3. [Equation (2) and Implementation (4-Hz Butterworth, rank-16)] The motion model constrains D = ΦΨᵀ with R=16 and the temporal basis is passed through a 4-Hz Butterworth low-pass filter. PVC contractions are short, non-periodic events that may contain components above 4 Hz or outside a rank-16 subspace; if so, the reconstruction can fold those components into errors that manifest as spurious low-EF beats. The phantom validation uses a 60-bpm sinusoidal, periodic, small-EF motion and does not exercise this regime. A concrete test would be to simulate or acquire a phantom motion waveform with a known PVC-like pattern (e.g., a premature, low-stroke-volume contraction) and verify that the reconstructed per-beat EF matches the injected waveform. Without such a test, the bimodal EF distributions in Figure 7 could be in part a model artifact.
  4. [Results, Figure 6] The claimed temporal correspondence between volume irregularities and PVC episodes is supported only by qualitative visual alignment of the ECG trace and volume curve. This does not quantify volumetric accuracy or even the specificity of the association. The authors should report a quantitative comparison, for example the detection rate of PVCs from the volume curve against the ECG, or the timing offset between ECG R-waves and reconstructed volumetric troughs/peaks. This would strengthen the argument that the observed low-EF peaks are physiological rather than reconstruction artifacts.
minor comments (5)
  1. [Equations (1)-(2)] In Eq. (1), the sum is written as ∑_t but the domain is not fully specified; in Eq. (2), the norm subscript is shown as ℓ without specifying which ℓ (the text says ℓ1 for in vivo, ℓ2 for phantom). Please make the notation explicit.
  2. [Methods, Phantom Study] The text states flip angle = 6° for the phantom GRE sequence, but Table 1 lists flip angle = 10°. Please reconcile.
  3. [Data Analysis / Implementation] The hyperparameter selection is described both as a parameter sweep on phantom data and as tuning on an additional volunteer's acquisition. Clarify which settings were used for in-vivo reconstructions and whether the 'additional volunteer' is part of the reported cohort.
  4. [Discussion] The sentence 'The authors have identified several computational solutions...' is vague; either name the specific algorithmic optimizations or remove the sentence.
  5. [Abstract] The abstract states '10 healthy volunteers, and 10 patients with PVCs' while the limitations section and full text state 4+4. This consistency issue is critical and must be corrected as part of the major revision.

Circularity Check

0 steps flagged

No significant circularity: the central feasibility claim is anchored by external phantom ground truth and 2D cine comparisons, not by the cited CMR-MOTUS prior work alone.

full rationale

The paper extends CMR-MOTUS [17] to 3D, and this is the authors' own prior method, so there is a self-citation. However, the load-bearing evidence does not reduce to that citation: the reconstruction is validated against independently acquired static-phantom ground truth (EF 22.1 ± 0.6% vs 21.9%) and against clinical 2D real-time cine EF software for in-vivo subjects. Equations (1)-(2) minimize data consistency against acquired k-space data; no EF or volume parameter is fitted and then renamed as a prediction. The bimodal PVC EF distributions arise from propagating a single segmentation through reconstructed motion fields, and the authors cross-check timing with simultaneous ECG. The rank-16/4-Hz motion model and the lambda sweep are limitations or risk factors for arrhythmic dynamics, but they are model assumptions, not circular reductions; the phantom and 2D comparisons are external benchmarks that do not enter the reconstruction. The manuscript's own limitations note the small cohort and lack of a dedicated per-beat ground truth, which is a validation gap rather than a circularity. The minor self-citation is not load-bearing because the present paper reimplements and empirically tests the extension.

Axiom & Free-Parameter Ledger

4 free parameters · 5 axioms · 0 invented entities

The method relies on a fixed model: a single motion-corrected reference image plus low-rank, temporally smoothed deformation fields. The main free parameters are algorithmic (λ, rank, cutoff, batch size) selected by empirical tuning rather than derived from first principles. There are no invented physical entities. The core assumption that all dynamics are representable as motion of a static image is the central modeling risk.

free parameters (4)
  • TV regularization weight λ = not reported (selected via phantom sweep)
    The specific value of λ is not stated in the paper; it was selected empirically through a parameter sweep on phantom data and kept for in vivo data (Methods, Framework).
  • Motion model rank R = 16
    Rank-16 low-rank factorization D=ΦΨ^T used for all datasets; chosen as memory/compute tradeoff, not derived from data (Implementation).
  • Temporal low-pass cutoff = 4 Hz
    Butterworth filter cutoff applied to temporal components Ψ to enforce temporal smoothness; may suppress high-frequency motion (Implementation).
  • SGD batch size = 40 frames
    Selected empirically to minimize noise in reconstructed motion fields while maintaining convergence (Implementation).
axioms (5)
  • domain assumption All time frames are described by a single motion-corrected reference image q warped by D_t (Eq. 1-2)
    The reconstruction alternates between q and D_t, assuming time-stable contrast and no dynamics beyond motion; stated in Framework and contrasted with the time-dependent reference in original 2D CMR-MOTUS.
  • domain assumption Cardiac/respiratory motion lies in a rank-16 subspace with temporally smooth (≤4 Hz) components
    Low-rank factorization and Butterworth filtering; if rank/cutoff too low for PVC dynamics, motion fields are biased (Implementation).
  • domain assumption Time-stable contrast in all in vivo acquisitions
    Authors state 'all data used in this study have a time stable contrast as there are no contrast agent inflow effects'; however in vivo scans were after contrast administration (Framework).
  • domain assumption Isotropic TV spatial smoothness produces physically plausible motion fields
    Isotropic TV on D is intended to promote smooth motion while preserving sliding motion; chosen empirically (Framework).
  • domain assumption Phantom tissue relaxation properties approximate human myocardium and blood
    Phantom validation assumes the silicone/liquid T1/T2 values are representative enough that motion-field accuracy transfers to human myocardium (Phantom Study).

pith-pipeline@v1.3.0-alltime-deepseek · 12511 in / 14151 out tokens · 133335 ms · 2026-08-02T18:51:24.666922+00:00 · methodology

0 comments
read the original abstract

Conventional cardiovascular magnetic resonance (CMR) cine imaging combines data across multiple heartbeats, an assumption that can fail in arrhythmia because beat-to-beat variation causes motion artifacts and obscures functional heterogeneity. Although 2D real-time cine resolves individual beats, stacked slices are suboptimal for capturing complex 3D cardiac dynamics. We investigated continuous beat-to-beat volumetric quantification in patients with premature ventricular contractions (PVCs) using free-running 3D real-time CMR. CMR-MOTUS was extended to jointly reconstruct time-resolved 3D motion fields and a motion-corrected reference image from continuously acquired data without breath-holding or ECG gating. Data were acquired with a variable-density Cartesian trajectory and either 3D spoiled gradient-echo or balanced steady-state free-precession imaging. Ventricular volumes were obtained by propagating one manual reference-image segmentation through all reconstructed frames. The method was evaluated in a cardiac motion phantom with static ground-truth acquisitions, 10 healthy volunteers, and 10 patients with PVCs; all in vivo scans were performed after contrast administration. Phantom ejection fraction (EF) agreed closely with ground truth (17.86% versus 17.27%). Healthy volunteers showed narrow beat-to-beat EF distributions, whereas patients with PVCs showed broader and sometimes bimodal distributions. Simultaneous ECG recordings supported the temporal correspondence between volume irregularities and PVC episodes. Free-running joint 3D motion-field and image reconstruction enables continuous beat-to-beat volumetric assessment and can reveal functional heterogeneity obscured by gated or heartbeat-averaged methods. Larger studies are required to establish clinical validity and determine its role alongside standard 2D cine analysis.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Fast ungated five-dimensional cardiac MRI on a 1.5 T MR-linac for MRI-guided radiotherapy

    physics.med-ph 2026-07 conditional novelty 6.0

    One minute of ungated Cartesian MRI on an MR-linac can be reconstructed into separated cardiac and respiratory 5D motion states in six minutes using low-rank deformation fields.

Reference graph

Works this paper leans on

24 extracted references · 12 canonical work pages · cited by 1 Pith paper

  1. [1]

    Barriers to cardiovascular magnetic resonance imaging scan performance and reporting by cardiologists: a systematic literature review,

    T. Betemariam et al., “Barriers to cardiovascular magnetic resonance imaging scan performance and reporting by cardiologists: a systematic literature review, ” Eur. Heart J. - Imaging Methods Pract., vol. 3, no. 1, p. qyaf010, Jan. 2025, doi: 10.1093/ehjimp/qyaf010

    work page doi:10.1093/ehjimp/qyaf010 2025

  2. [2]

    Heart deformation analysis: measuring regional myocardial velocity with MR imaging,

    K. Lin, J. D. Collins, V . Chowdhary, M. Markl, and J. C. Carr, “Heart deformation analysis: measuring regional myocardial velocity with MR imaging, ” Int. J. Cardiovasc. Imaging, vol. 32, no. 7, pp. 1103–1111, Jul. 2016, doi: 10.1007/s10554- 016-0879-z

    work page doi:10.1007/s10554- 2016

  3. [3]

    Systolic Myocardial Dysfunction in Patients with Type 2 Diabetes Mellitus: Identification at MR Imaging with Cine Displacement Encoding with Stimulated Echoes,

    L. Ernande et al., “Systolic Myocardial Dysfunction in Patients with Type 2 Diabetes Mellitus: Identification at MR Imaging with Cine Displacement Encoding with Stimulated Echoes, ” Radiology, vol. 265, no. 2, pp. 402–409, Nov. 2012, doi: 10.1148/radiol.12112571

    work page doi:10.1148/radiol.12112571 2012

  4. [4]

    Myocardial tissue phase mapping reveals impaired myocardial tissue velocities in obesity,

    O. J. Rider et al., “Myocardial tissue phase mapping reveals impaired myocardial tissue velocities in obesity, ” Int. J. Cardiovasc. Imaging, vol. 31, no. 2, pp. 339–347, Feb. 2015, doi: 10.1007/s10554-014-0548-z

    work page doi:10.1007/s10554-014-0548-z 2015

  5. [5]

    Automated Assessment of Left Ventricular Function and Mass Using Heart Deformation Analysis:,

    K. Lin et al., “Automated Assessment of Left Ventricular Function and Mass Using Heart Deformation Analysis:, ” Acad. Radiol., vol. 23, no. 3, pp. 321–325, Mar. 2016, doi: 10.1016/j.acra.2015.10.020

    work page doi:10.1016/j.acra.2015.10.020 2016

  6. [6]

    Segmented radial cardiac MRI during arrhythmia using retrospective electrocardiogram and respiratory gating,

    R. Chava, F . Assis, D. Herzka, and A. Kolandaivelu, “Segmented radial cardiac MRI during arrhythmia using retrospective electrocardiogram and respiratory gating, ” Magn. Reson. Med., vol. 81, no. 3, pp. 1726–1738, Mar. 2019, doi: 10.1002/mrm.27533

    work page doi:10.1002/mrm.27533 2019

  7. [7]

    5D whole‐heart sparse MRI,

    L. Feng et al., “5D whole‐heart sparse MRI,” Magn. Reson. Med., vol. 79, no. 2, pp. 826–838, Feb. 2018, doi: 10.1002/mrm.26745

    doi:10.1002/mrm.26745 2018

  8. [8]

    An automated approach to fully self‐gated free‐running cardiac and respiratory motion‐resolved 5D whole‐ heart MRI,

    L. Di Sopra, D. Piccini, S. Coppo, M. Stuber, and J. Yerly, “An automated approach to fully self‐gated free‐running cardiac and respiratory motion‐resolved 5D whole‐ heart MRI,” Magn. Reson. Med., vol. 82, no. 6, pp. 2118–2132, Dec. 2019, doi: 10.1002/mrm.27898

    doi:10.1002/mrm.27898 2019

  9. [9]

    Free breathing whole- heart 3D CINE MRI with self-gated Cartesian trajectory,

    M. Usman, B. Ruijsink, M. S. Nazir, G. Cruz, and C. Prieto, “Free breathing whole- heart 3D CINE MRI with self-gated Cartesian trajectory, ” Magn. Reson. Imaging, vol. 38, pp. 129–137, May 2017, doi: 10.1016/j.mri.2016.12.021

    work page doi:10.1016/j.mri.2016.12.021 2017

  10. [10]

    Fast Real-Time Cardiac MRI: a Review of Current Techniques and Future Directions,

    X. Wang, M. Uecker, and L. Feng, “Fast Real-Time Cardiac MRI: a Review of Current Techniques and Future Directions, ” Investig. Magn. Reson. Imaging, vol. 25, no. 4, p. 252, 2021, doi: 10.13104/imri.2021.25.4.252

    work page doi:10.13104/imri.2021.25.4.252 2021

  11. [11]

    Motion‐robust free‐running volumetric cardiovascular MRI,

    S. M. Arshad et al., “Motion‐robust free‐running volumetric cardiovascular MRI,” Magn. Reson. Med., vol. 92, no. 3, pp. 1248–1262, Sep. 2024, doi: 10.1002/mrm.30123. 23

    work page doi:10.1002/mrm.30123 2024

  12. [12]

    Respiratory variations in pulmonary and systemic blood flow in healthy humans,

    M. Elstad, “Respiratory variations in pulmonary and systemic blood flow in healthy humans, ” Acta Physiol., vol. 205, no. 3, pp. 341–348, Jul. 2012, doi: 10.1111/j.1748- 1716.2012.02419.x

    arXiv 2012

  13. [13]

    Differentiation of Constriction and Restriction,

    J. B. Geske, N. S. Anavekar, R. A. Nishimura, J. K. Oh, and B. J. Gersh, “Differentiation of Constriction and Restriction, ” J. Am. Coll. Cardiol., vol. 68, no. 21, pp. 2329–2347, Nov. 2016, doi: 10.1016/j.jacc.2016.08.050

    work page doi:10.1016/j.jacc.2016.08.050 2016

  14. [14]

    4D Golden‐Angle Radial MRI at Subsecond Temporal Resolution,

    L. Feng, “4D Golden‐Angle Radial MRI at Subsecond Temporal Resolution,” NMR Biomed., vol. 36, no. 2, p. e4844, Feb. 2023, doi: 10.1002/nbm.4844

    doi:10.1002/nbm.4844 2023

  15. [15]

    Extreme MRI: Large‐scale volumetric dynamic imaging from continuous non‐gated acquisitions,

    F . Ong et al., “Extreme MRI: Large‐scale volumetric dynamic imaging from continuous non‐gated acquisitions,” Magn. Reson. Med., vol. 84, no. 4, pp. 1763– 1780, Oct. 2020, doi: 10.1002/mrm.28235

    work page doi:10.1002/mrm.28235 2020

  16. [16]

    A multi-dynamic low-rank deep image prior for three-dimensional real-time cardiovascular magnetic resonance imaging,

    C. Chen et al., “A multi-dynamic low-rank deep image prior for three-dimensional real-time cardiovascular magnetic resonance imaging, ” J. Cardiovasc. Magn. Reson., vol. 28, no. 1, p. 102015, 2026, doi: 10.1016/j.jocmr.2025.102015

    arXiv 2026

  17. [17]

    Free-running time-resolved first-pass myocardial perfusion using a multi-scale dynamics decomposition: CMR-MOTUS,

    T. E. Olausson et al., “Free-running time-resolved first-pass myocardial perfusion using a multi-scale dynamics decomposition: CMR-MOTUS, ” Magn. Reson. Mater. Phys. Biol. Med., Sep. 2025, doi: 10.1007/s10334-025-01291-x

    doi:10.1007/s10334-025-01291-x 2025

  18. [18]

    Characterization of a deformable beating cardiac phantom with real‐time dosimetric capabilities for validation of MRI‐guided heart radiotherapy,

    M. M. N. Aubert et al., “Characterization of a deformable beating cardiac phantom with real‐time dosimetric capabilities for validation of MRI‐guided heart radiotherapy,” Med. Phys., vol. 53, no. 2, p. e70313, Feb. 2026, doi: 10.1002/mp.70313

    doi:10.1002/mp.70313 2026

  19. [19]

    WarpPINN: Cine-MR image registration with physics-informed neural networks,

    P . Arratia López, H. Mella, S. Uribe, D. E. Hurtado, and F . Sahli Costabal, “WarpPINN: Cine-MR image registration with physics-informed neural networks, ” Med. Image Anal., vol. 89, p. 102925, Oct. 2023, doi: 10.1016/j.media.2023.102925

    arXiv 2023

  20. [20]

    Isotropic Total Variation Regularization of Displacements in Parametric Image Registration,

    V . Vishnevskiy, T. Gass, G. Szekely, C. Tanner, and O. Goksel, “Isotropic Total Variation Regularization of Displacements in Parametric Image Registration, ” IEEE Trans. Med. Imaging, vol. 36, no. 2, pp. 385–395, Feb. 2017, doi: 10.1109/TMI.2016.2610583

    arXiv 2017

  21. [21]

    Technical Report (v1.0)--Pseudo- random Cartesian Sampling for Dynamic MRI,

    M. Joshi, A. Pruitt, C. Chen, Y . Liu, and R. Ahmad, “Technical Report (v1.0)--Pseudo- random Cartesian Sampling for Dynamic MRI, ” Jun. 07, 2022, arXiv: arXiv:2206.03630. Accessed: Sep. 19, 2023. [Online]. Available: http://arxiv.org/abs/2206.03630

    Pith/arXiv arXiv 2022

  22. [22]

    MRI-Based Quantification of the Deformation of an Anthropomorphic Heart Phantom,

    M. Aubert, “MRI-Based Quantification of the Deformation of an Anthropomorphic Heart Phantom, ” presented at the ISMRM Workshop on Motion Correction in MR, Quebec, QB, Canada, 2024

    2024

  23. [23]

    J. Schulz-Menger et al., “Standardized image interpretation and post processing in cardiovascular magnetic resonance: Society for Cardiovascular Magnetic Resonance (SCMR) Board of Trustees Task Force on Standardized Post Processing, ” J. Cardiovasc. Magn. Reson., vol. 15, no. 1, p. 35, Jan. 2013, doi: 10.1186/1532-429X-15-35

    work page doi:10.1186/1532-429x-15-35 2013

  24. [24]

    ESPIRiT—an eigenvalue approach to autocalibrating parallel MRI: Where SENSE meets GRAPPA,

    M. Uecker et al., “ESPIRiT—an eigenvalue approach to autocalibrating parallel MRI: Where SENSE meets GRAPPA, ” Magn. Reson. Med., vol. 71, no. 3, pp. 990–1001, Mar. 2014, doi: 10.1002/mrm.24751. 24 Supplementary Information subject_1.mp4 Supplementary Figure 1: Data in figure is from subject 1. The figure shows a video of the CMR-MOTUS reconstruction, whi...

    doi:10.1002/mrm.24751 2014