Pith. sign in

REVIEW 3 major objections 5 minor 33 references

Brightness-Invariant Tracking Estimation in Tagged MRI

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read BRITE separates anatomy from fading tags, yielding accurate motion and strain estimates in tagged MRI.

desk verdict BRITE is a well-built tracking method, but 'brightness-invariant' is only shown for global brightness changes; local variations are not modeled or tested. read the letter →

arxiv 2505.18365 v1 pith:5EEWMD2G submitted 2025-05-23 eess.IV cs.CV

Zhangxing Bian , Shuwen Wei , Xiao Liang , Yuan-Chiao Lu , Samuel W. Remedios , Fangxu Xing , Jonghye Woo , Dzung L. Pham
show 5 more authors
Aaron Carass Philip V. Bayly Jiachen Zhuo Ahmed Alshareef Jerry L. Prince
This is my paper · ORCID
classification eess.IVcs.CV
keywords MRtaggingtaggedMRImotiontrackingtagfadingspectraloverlapstrainestimationdiffusionmodelphysics-informedneuralnetwork
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

Tagged MRI marks tissue with a striped saturation pattern that moves with the tissue, but the pattern fades and overall brightness drifts as the magnetization recovers, so both intensity-based optical flow and Fourier methods lose accuracy. The paper introduces BRITE, which treats each observed tagged frame as the product of an unknown anatomical image and a sinusoidal tag pattern, estimating both together with the motion of each material point from the first frame (the Lagrangian motion). Tag fading is absorbed by two per-frame scalars, the tag amplitude and offset, rather than being allowed to corrupt the motion estimate. On SPAMM-tagged gel phantom data with varied tag periods and flip angles, the paper reports that BRITE gives more accurate motion and strain estimates than HARP, SinMod, DRIMET, SyN variants, and DeepTag, and remains accurate as tags fade. If this holds, motion tracking from tagged MRI can tolerate brightness changes without the extra acquisitions used by methods such as CSPAMM and TruHARP.

What carries the argument

The load-bearing object is the factorized forward model $g^{h,v}_t = \tilde a_t \otimes \tilde p^{h,v}_t$, where $\tilde p^{h,v}_t = [A_t \sin(2\pi \tilde\mu s_{h,v} + \tilde\varphi^{h,v}) + B_t] \circ \phi_t^{-1}$, with $s_h=x$ and $s_v=y$. This factorization lets brightness change be absorbed by the two scalar parameters $A_t$ and $B_t$ instead of leaking into the motion field. The second essential piece is the deformation representation: a physics-informed neural network maps coordinates $(x,y)$ to a stationary velocity field $u_t$, and scaling-and-squaring exponential integration converts $u_t$ into the smooth, invertible deformation $\phi_t$ and its inverse. Because $\phi_t^{-1}$ warps both the anatomy and the tag pattern into the current frame, the same motion explains both image components, which is why no explicit smoothness penalty is needed.

What would settle it

Run BRITE on a tagged sequence with known deformation in which part of the field of view undergoes a local brightness change (for instance, a spatially varying T1 or flip-angle map, or an inflow of contrast agent) while the rest stays uniform; if the estimated displacement near that local region deviates from ground truth but is accurate elsewhere, the global-fading, constant-anatomy assumption is the reason.

Watch

Extended reading notes

Core claim

The central claim is that the ill-posed problem of jointly recovering the underlying anatomy, the fading tag pattern, and the Lagrangian motion (motion of each material point relative to the first frame) from a tagged MRI sequence has a tractable solution when the forward model is factored as $g^{h,v}_t = \tilde a_t \otimes \tilde p^{h,v}_t$. The anatomy is required to be constant in material coordinates, $\tilde a_t = \tilde a_0 \circ \phi_t^{-1}$, while the tag pattern is a sinusoid whose frequency and phase are fixed from the first frame and whose amplitude $A_t$ and offset $B_t$ are the only time-varying quantities. A diffusion model pretrained on anatomical images provides the prior that makes recovering $\tilde a_0$ from the product well-posed, and a physics-informed neural network outputs a stationary velocity field whose exponential map produces the deformation $\phi_t$, a smooth, invertible warping. The paper reports that on gel phantom data, this disentanglement yields more accurate displacement and strain estimates than the compared methods, with the advantage largest when tag fading and spectral overlap are severe.

Load-bearing premise

The method's forward model assumes the anatomy is exactly carried along by the motion, with only a single global amplitude and offset changing in the tag pattern; if real tissue brightness changes locally rather than globally, the optimizer can translate that brightness change into a spurious apparent deformation.

Editorial extensions

If this is right

  • If BRITE's reported accuracy holds, tagged MRI motion tracking can be made brightness-invariant without extra acquisitions, removing a major practical argument for CSPAMM or TruHARP.
  • Because the tag frequency is estimated from the data rather than assumed from the pulse sequence, the method can tolerate gradient imperfections that shift or blur the spectral peaks.
  • Initializing each frame's velocity-field network from the previous frame's solution lets the tracker follow motion larger than half a tag period without tag-jumping, and avoids error accumulation from composing pairwise registrations.
  • Resistance to tag fading should extend the usable time window of a single tagged acquisition, exactly where HARP and intensity-based methods degrade most.
  • The reported processing time of about 8.4 seconds per frame with modest GPU memory suggests 2D clinical workflows are feasible today, and the paper argues 3D extension is straightforward.

Reading between the lines

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

  • A test the paper does not run is a spatially nonuniform brightness change, since its fading model has only global $A_t$ and $B_t$; one would expect local T1 variation or inflow to be absorbed as apparent deformation, which would set the boundary of the method's applicability.
  • The diffusion prior is trained only on synthetic ovals, and the paper's own limitations section restricts the tag model to sinusoids and validation to a 2D static gel phantom with simulated deformations; applying the method to textured human organs with independent motion measurements will show how much the prior and tag model must be generalized.
  • The same factorization could be carried over to other amplitude-modulated tagging schemes, such as grid tags or higher-order SPAMM, by swapping the sinusoidal tag model while keeping the PINN tracking stage.
  • A direct comparison against CSPAMM and TruHARP on the same phantom with the same ground truth would quantify how much accuracy is gained or lost by avoiding their extra acquisitions.
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

3 major / 5 minor

Summary. The paper introduces BRITE, a two-stage tracking method for SPAMM-tagged MRI. In the first stage, the method disentangles the anatomical image from the horizontal and vertical tag patterns at the reference frame by jointly optimizing a DDPM latent code and a CNN-predicted sinusoidal tag model (Eq. 1). In the second stage, for each subsequent frame a PINN produces a stationary velocity field whose exponential map defines a diffeomorphic Lagrangian deformation, while a Tag Fading Module estimates two global scalars A_t and B_t per frame; all unknowns are fit by minimizing the reconstruction MSE in Eq. (6). The method is evaluated on a silicone gel phantom with eight SPAMM acquisitions (tag periods 9, 12, 18, 26 mm; flip angles 5 and 10 degrees) under simulated non-rigid deformations (N=20), one rigid rotation, and a static case, comparing EPE and eMPS against HARP, SinMod, DRIMET, SyN variants, and DeepTag. The paper claims that BRITE is more accurate than these baselines and is resistant to tag fading.

Significance. If the claims survive closer quantitative scrutiny, BRITE would be a useful contribution: it attacks tag fading and spectral overlap simultaneously, requires no extra acquisitions beyond the standard horizontal/vertical SPAMM pair, and the validation design has real merit because the phantom images carry natural T1-driven tag fading while the ground-truth deformations come from external simulation. The runtime and memory figures are modest. The main evidence, however, is global-fading-only phantom data, and the quantitative support is incomplete (no error bars or significance tests, and no numbers for the rotation/static cases), so the strength of the central claim is currently larger than the evidence.

major comments (3)
  1. [§3, Eqs. (4)–(6); §5, Limitations] The forward model represents brightness evolution by only two global scalars per frame, A_t and B_t, in the Tag Fading Module. The phantom is a homogeneous gel, so its T1-driven fading is spatially uniform and exactly matches this model. Real tissue has spatially varying T1, B1 inhomogeneity, inflow, and off-resonance; none of these are exercised by the experiments, and the manuscript's Limitations section does not discuss them. Under the pixel-wise MSE objective of Eq. (6), a spatially varying brightness change could be absorbed as spurious deformation, which is precisely the failure mode the title promises to avoid. Please either narrow the claim to global tag fading or add a validation with spatially varying brightness modulation (e.g., synthetic local intensity scaling on the acquired phantom data) and report EPE/eMPS for that case.
  2. [§5, Fig. 5] The main quantitative evidence is reported only as curves without error bars, confidence intervals, or significance tests across the 20 non-rigid deformations. The claim that BRITE 'generally outperforms' all other methods cannot be verified statistically from the figures. Please provide per-sequence results, summary statistics with error bars, and, if appropriate, paired tests (e.g., Wilcoxon) on EPE and eMPS.
  3. [§5, Fig. 7] The rigid-rotation and static no-motion experiments are central to the tag-fading-resistance claim, yet the text states that 'quantitative results are omitted due to limited space.' In particular, the static case directly measures spurious motion caused by fading (cf. Fig. 2), so omitting its numbers weakens the main claim. Please include EPE and eMPS values for the rotation and static scenarios, alongside the qualitative images.
minor comments (5)
  1. [§4, Evaluation Metrics] Please specify how MPS is computed from the estimated displacement field (e.g., finite-difference strain and Lagrangian strain tensor), since eMPS is a central metric.
  2. [§4, Datasets] Please report the B-spline control-point spacing and displacement magnitudes used to generate the 20 non-rigid deformations, as these parameters determine the difficulty of the test.
  3. [§3, Implementation Details] The text says 'a multilayer perception' (should be 'perceptron'), and the notation for the tag-pattern CNN and the PINN weights should be made consistent (f_{\theta_{tag}} versus f_{\theta_{pinn,t}}).
  4. [§5, Figs. 5–7] The figure captions contain garbled strings such as '10-12mm-15-18mm-2' and the legends in Fig. 5 are difficult to read; please clean up all figures and make the per-method plotting conventions explicit.
  5. [Overall] No code or data availability statement is present; please add one or state that the phantom data can be shared on request.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: BRITE's motion and strain estimates are validated against externally simulated ground-truth deformations, and its fitted brightness parameters are nuisance variables rather than disguised predictions.

full rationale

The central derivation chain is not circular. BRITE fits an anatomical prior (DDPM over synthetic ovals), tag-pattern parameters, per-frame fading scalars (At, Bt), and PINN velocity fields by minimizing reconstruction error (Eqs. 1 and 6), but its motion outputs are evaluated against simulated rigid and non-rigid ground-truth deformations that are not used in the fitting objective. The per-frame At and Bt in Eqs. (5)-(6) are nuisance parameters that absorb global tag fading, and the validation shows that this fitted brightness model does not prevent accurate motion recovery on the phantom data, including a static no-motion case. Self-citations (e.g., ref. [8], DRIMET baseline) are present but not load-bearing: they motivate prior work and provide baselines rather than justify the core inverse problem. The paper's own limitation passage correctly notes the sinusoidal tag-pattern assumption and the need for human validation. The residual concern—that spatially varying brightness changes not captured by two global scalars could be absorbed as spurious deformation (Eqs. 4-6)—is a generalization risk, not an internal circularity, and is not exercised by the homogeneous gel phantom experiments.

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

The central claim rests on the forward model g=a⊗p, material constancy of anatomy, sinusoidal tag model, and generalization of the synthetic DDPM prior. These are assumptions, not derived; failure of any of them would weaken the method. The free parameters are numerous but are fit to data and evaluated against external ground truth.

free parameters (5)
  • DDPM latent code z = optimized, no closed form
    Eq (1) jointly optimizes z to reconstruct the initial tagged images; it is a high-dimensional latent parameter, not a physical constant.
  • Tag pattern parameters (A0, B0, mu, phi_h, phi_v) = A0,B0 initialized 0.45/0.55; phi at 2pi; mu from pulse sequence, all refined by CNN
    Estimated from data by f_theta_tag in Eq (1); they define the disentangled tag pattern.
  • Per-frame fading parameters (A_t, B_t) = initialized 0.5 for each t
    Two scalar parameters per time frame fit by Eq (6) to absorb tag fading; these are empirical and not derived from T1/TR/FA.
  • PINN velocity field weights theta_pinn,t = fully connected net, 3x128 hidden; trained per frame
    Optimized in Eq (6) to produce the motion field; the motion estimate is the output of this fitted network.
  • CNN encoder weights theta_tag = 4-layer ResNet, trained with Eq (1)
    Predicts tag parameters from images; part of the disentanglement fit.
assumptions (5)
  • domain assumption Tagged MR image equals element-wise product of anatomy and sinusoidal tag pattern (g = a ⊗ p)
    This forward model defines the disentanglement objective (Eq 1) and is a simplification of the MR signal equation.
  • domain assumption Anatomy is materially constant: a_t = a0 ∘ phi_t^{-1}
    Used in Eq (4); the method models only tag amplitude/offset changes (Eq 5-6), not changes in tissue brightness.
  • domain assumption Tag pattern remains a single sinusoid with fixed frequency and phase after t=0
    Eq (5) carries mu and phi from t=0; the authors list this as a limitation (Section 5).
  • domain assumption DDPM trained on synthetic random ovals generalizes to real phantom anatomy
    Pretrained on synthetic ovals (Section 3, Implementation Details) and applied to gel phantom; no evidence it works for arbitrary anatomy.
  • standard math Exponential map (scaling and squaring) produces diffeomorphic deformations
    Eq (3), standard result from Lie group methods.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Brightness-Invariant Tracking Estimation in Tagged MRI." pith.science (2026). https://pith.science/paper/5EEWMD2G

@misc{pith2026250518365,
  author       = {Pith},
  title        = {Pith review of: Brightness-Invariant Tracking Estimation in Tagged MRI},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5EEWMD2G}},
  note         = {Machine review of arXiv:2505.18365}
}
read the original abstract

Magnetic resonance (MR) tagging is an imaging technique for noninvasively tracking tissue motion in vivo by creating a visible pattern of magnetization saturation (tags) that deforms with the tissue. Due to longitudinal relaxation and progression to steady-state, the tags and tissue brightnesses change over time, which makes tracking with optical flow methods error-prone. Although Fourier methods can alleviate these problems, they are also sensitive to brightness changes as well as spectral spreading due to motion. To address these problems, we introduce the brightness-invariant tracking estimation (BRITE) technique for tagged MRI. BRITE disentangles the anatomy from the tag pattern in the observed tagged image sequence and simultaneously estimates the Lagrangian motion. The inherent ill-posedness of this problem is addressed by leveraging the expressive power of denoising diffusion probabilistic models to represent the probabilistic distribution of the underlying anatomy and the flexibility of physics-informed neural networks to estimate biologically-plausible motion. A set of tagged MR images of a gel phantom was acquired with various tag periods and imaging flip angles to demonstrate the impact of brightness variations and to validate our method. The results show that BRITE achieves more accurate motion and strain estimates as compared to other state of the art methods, while also being resistant to tag fading.

Figures

Figures reproduced from arXiv: 2505.18365 by the authors.

Figure 1
Figure 1. Top row: Tagged MRI. Middle Row: Their corresponding Fourier spectrum. Bottom Row: The midline profiles extracted from the Fourier spectra. The column names “Xmm@Ys” denote a tag period of X mm acquired at Y s following tagging. cardiac motion [15], investigating musculoskeletal dynamics [19], studying tongue movements during speech [20, 23, 32], quantifying brain deformation during mild head impact [7, 16], and eva… view at source ↗
Figure 2
Figure 2. Tagged MRI of a static phantom. Harmonic phases are distorted over time due to tag fading and spectral overlap, resulting in erroneous non-zero motion estimation. The acquisition of additional tagged MR images has been used to avoid this problem. The complementary SPAMM (CSPAMM) [12] method suppresses the central peak by combining two SPAMM acquisitions; to yield both horizontal and vertical tags, four acquisitions … view at source ↗
Figure 3
Figure 3. Overview of the Proposed BRITE Framework. The pipeline consists of two stages: disentangling and tracking. At t = 0, the “Tag Pattern Module” and a pretrained DDPM model are used to separate the anatomical image a˜0 from the tag patterns p˜ h,v 0 . Subsequently, for each time t > 0, the PINN estimates the Lagrangian motion while the “Tag Fading Module” captures tag fading. from the observed data an ill-posed problem… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: (a) Input tagged MR sequence with horizontal and vertical tags. (b) Synthetic oval-shaped images used for training the DDPM. (c) Disentanglement process: shown are the optimization trajectories of µ, φh , φv , A0, B0, the disentanglement loss, estimated anatomy a˜0, an…
Figure 5
Figure 5. Figure 5: Non-rigid Deformation Evaluation. Results are shown for EPE (top two rows) and eMPS (bottom two rows) at different time points. Each column represents a different tag period, and each row corresponds to a flip angle. t gt − |ϕ| [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: Qualitative Results under Non-Rigid Deformations. Two examples are shown with FA = 5◦ , TP = 18mm on the top and FA = 10◦ , TP = 12mm on the bottom. The left column displays the input images at t = 0s and 1.1s. The Motion Estimation panel shows the displacement fields …
Figure 7
Figure 7. Figure 7: Qualitative Results under Rigid-Body Rotations and No Motion. The top row has FA = 10◦ , TP = 9mm for the 15◦ rigid-body rotation, and the bottom row shows FA = 10◦ , TP = 26mm for a static, no-motion scenario. Evaluation: Rigid-body Rotation and No Motion All methods …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 32 canonical work pages

  1. [1]

    Magnetic Resonance in Medicine64(2), 574–585 (2010)

    Agarwal, H.K., Prince, J.L., Abd-Elmoniem, K.Z.: Total removal of unwanted harmonic peaks (TruHARP) MRI for single breath-hold high-resolution myocardial motion and strain quantification. Magnetic Resonance in Medicine64(2), 574–585 (2010)

    work page 2010

  2. [2]

    Journal of Magnetic Resonance 137(1), 247 (1999)

    Aletras, A.H., Ding, S., Balaban, R.S., Wen, H.: DENSE: displacement encoding with stimulated echoes in cardiac functional MRI. Journal of Magnetic Resonance 137(1), 247 (1999)

    work page 1999

  3. [3]

    In: American Associ- ation for Artificial Intelligence (AAAI) Spring Symposium Series

    Amini, A., Curwen, R., Constable, R.T., Gore, J.C.: MR physics-based snake tracking and dense deformations from tagged cardiac images. In: American Associ- ation for Artificial Intelligence (AAAI) Spring Symposium Series. Applications of Computer Vision in Medical Image Processing. pp. 126–129 (1994)

    work page 1994

  4. [4]

    In: MICCAI

    Arsigny, V., Commowick, O., Pennec, X., Ayache, N.: A log-euclidean framework for statistics on diffeomorphisms. In: MICCAI. pp. 924–931. Springer (2006)

    work page 2006

  5. [5]

    IEEE Transactions on Medical Imaging29(5), 1114–1123 (2010)

    Arts, T., Prinzen, F.W., Delhaas, T., Milles, J.R., Rossi, A.C., Clarysse, P.: Mapping displacement and deformation of the heart with local sine-wave modeling. IEEE Transactions on Medical Imaging29(5), 1114–1123 (2010)

    work page 2010

  6. [6]

    Radiology172(2), 349–350 (1989)

    Axel, L., Dougherty, L.: Heart wall motion: improved method of spatial modulation of magnetization for MR imaging. Radiology172(2), 349–350 (1989)

    work page 1989

  7. [7]

    Annals of Biomedical Engineering49, 2677–2692 (2021)

    Bayly, P.V., Alshareef, A., Knutsen, A.K., Upadhyay, K., Okamoto, R.J., Carass, A., Butman, J.A., Pham, D.L., Prince, J.L., Ramesh, K.T., Johnson, C.L.: MR imaging of human brain mechanics in vivo: New measurements and applications to the development of computational models of brain injury. Annals of Biomedical Engineering49, 2677–2692 (2021)

    work page 2021

  8. [8]

    Bian, Z., Alshareef, A., Wei, S., Chen, J., Wang, Y., Woo, J., Pham, D.L., Zhuo, J., Carass, A., Prince, J.L.: Is registering raw tagged-MR enough for strain estimation in the era of deep learning? In: Medical Imaging 2024: Image Processing. vol. 12926, pp. 79–85. SPIE (2024)

    work page 2024

Show all 33 references
  1. [9]

    In: Medical Imaging with Deep Learning

    Bian, Z., Xing, F., Yu, J., Shao, M., Liu, Y., Carass, A., Woo, J., Prince, J.L.: DRIMET: Deep registration-based 3d incompressible motion estimation in tagged- MRI with application to the tongue. In: Medical Imaging with Deep Learning. pp. 134–150. PMLR (2024) 14 Bian et al

    2024

  2. [10]

    In: International Conference on Machine Learning

    Bora, A., Jalal, A., Price, E., Dimakis, A.G.: Compressed sensing using generative models. In: International Conference on Machine Learning. pp. 537–546. PMLR (2017)

    2017

  3. [11]

    IEEE Transactions on Medical Imaging18(4), 359–363 (1999)

    Dougherty, L., Asmuth, J.C., Blom, A.S., Axel, L., Kumar, R.: Validation of an optical flow method for tag displacement estimation. IEEE Transactions on Medical Imaging18(4), 359–363 (1999)

    1999

  4. [12]

    Magnetic Resonance in Medicine30(2), 191–200 (1993)

    Fischer, S.E., McKinnon, G., Maier, S., Boesiger, P.: Improved myocardial tagging contrast. Magnetic Resonance in Medicine30(2), 191–200 (1993)

    1993

  5. [13]

    The University of Iowa (1990)

    Fisher, D.J.: Automatic tracking of cardiac wall motion using magnetic resonance markers. The University of Iowa (1990)

    1990

  6. [14]

    In: Information Processing in Medical Imaging

    Gupta, S.N., Prince, J.L.: On variable brightness optical flow for tagged MRI. In: Information Processing in Medical Imaging. vol. 3, pp. 323–334. Kluwer Dordrecht (1995)

    1995

  7. [15]

    Journal of Cardiovascular Magnetic Resonance13(1), 36 (2011)

    Ibrahim, E.S.H.: Myocardial tagging by cardiovascular magnetic resonance: evolu- tion of techniques–pulse sequences, analysis algorithms, and applications. Journal of Cardiovascular Magnetic Resonance13(1), 36 (2011)

    2011

  8. [16]

    Journal of Biomechanics 47(14), 3475–3481 (2014)

    Knutsen, A.K., Magrath, E., McEntee, J.E., Xing, F., Prince, J.L., Bayly, P.V., Butman, J.A., Pham, D.L.: Improved measurement of brain deformation during mild head acceleration using a novel tagged MRI sequence. Journal of Biomechanics 47(14), 3475–3481 (2014)

    2014

  9. [17]

    Journal of Magnetic Resonance Imaging36(6), 1490–1495 (2012)

    Mannelli, L., Wilson, G.J., Dubinsky, T.J., Potter, C.A., Bhargava, P., Cuevas, C., Linnau, K.F., Kolokythas, O., Gunn, M.L., Maki, J.H.: Assessment of the liver strain among cirrhotic and normal livers using tagged MRI. Journal of Magnetic Resonance Imaging36(6), 1490–1495 (2012)

    2012

  10. [18]

    IEEE Transactions on Medical Imaging40(4), 1240–1252 (2021)

    Mella, H., Mura, J., Wang, H., Taylor, M.D., Chabiniok, R., Tintera, J., Sotelo, J., Uribe, S.: HARP-I: a harmonic phase interpolation method for the estimation of motion from tagged MR images. IEEE Transactions on Medical Imaging40(4), 1240–1252 (2021)

    2021

  11. [19]

    Medical Physics39(4), 1793–1810 (2012)

    Moerman, K.M., Sprengers, A.M., Simms, C.K., Lamerichs, R.M., Stoker, J., Nederveen, A.J.: Validation of continuously tagged MRI for the measurement of dynamic 3D skeletal muscle tissue deformation. Medical Physics39(4), 1793–1810 (2012)

    2012

  12. [20]

    Radiology191(2), 578–580 (1994)

    Niitsu, M., Kumada, M., Campeau, N.G., Niimi, S., Riederer, S.J., Itai, Y.: Tongue displacement: visualization with rapid tagged magnetization-prepared MR imaging. Radiology191(2), 578–580 (1994)

    1994

  13. [21]

    Magnetic Resonance in Medicine42(6), 1048–1060 (1999)

    Osman, N.F., Kerwin, W.S., McVeigh, E.R., Prince, J.L.: Cardiac motion track- ing using CINE harmonic phase (HARP) magnetic resonance imaging. Magnetic Resonance in Medicine42(6), 1048–1060 (1999)

    1999

  14. [22]

    Magnetic Resonance in Medicine46(2), 324–334 (2001)

    Osman, N.F., Sampath, S., Atalar, E., Prince, J.L.: Imaging longitudinal cardiac strain on short-axis images using strain-encoded MRI. Magnetic Resonance in Medicine46(2), 324–334 (2001)

    2001

  15. [23]

    The Journal of the Acoustical Society of America121(1), 491–504 (2007)

    Parthasarathy, V., Prince, J.L., Stone, M., Murano, E.Z., NessAiver, M.: Measuring tongue motion from tagged cine-MRI using harmonic phase (HARP) processing. The Journal of the Acoustical Society of America121(1), 491–504 (2007)

    2007

  16. [24]

    IEEE Transactions on Medical Imaging11(2), 238–249 (1992)

    Prince, J.L., McVeigh, E.R.: Motion estimation from tagged MR image sequences. IEEE Transactions on Medical Imaging11(2), 238–249 (1992)

    1992

  17. [25]

    In: 2006 International Conference of the IEEE Engineering in Medicine and Biology Society

    Qian, Z., Metaxas, D.N., Axel, L.: Extraction and tracking of MRI tagging sheets using a 3D Gabor filter bank. In: 2006 International Conference of the IEEE Engineering in Medicine and Biology Society. pp. 711–714. IEEE (2006) BRITE: Brightness-Invariant Tracking Estimation 15

    2006

  18. [26]

    In: Proceedings of the 25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (IEEE Cat

    Qian, Z., Montillo, A., Metaxas, D.N., Axel, L.: Segmenting cardiac MRI tagging lines using Gabor filter banks. In: Proceedings of the 25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (IEEE Cat. No. 03CH37439). vol. 1, pp. 630–633. I...

    2003

  19. [27]

    Journal of Computational Physics378, 686–707 (2019)

    Raissi, M., Perdikaris, P., Karniadakis, G.E.: Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics378, 686–707 (2019)

    2019

  20. [28]

    arXiv preprint arXiv:2010.02502 (2020)

    Song, J., Meng, C., Ermon, S.: Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502 (2020)

    2020 arXiv

  21. [29]

    In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

    Ulyanov, D., Vedaldi, A., Lempitsky, V.: Deep image prior. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR). pp. 9446–9454 (2018)

    2018

  22. [30]

    IEEE Transactions on Medical Imaging36(10), 2116–2128 (2017)

    Xing, F., Woo, J., Gomez, A.D., Pham, D.L., Bayly, P.V., Stone, M., Prince, J.L.: Phase vector incompressible registration algorithm for motion estimation from tagged magnetic resonance images. IEEE Transactions on Medical Imaging36(10), 2116–2128 (2017)

    2017

  23. [31]

    In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

    Ye, M., Kanski, M., Yang, D., Chang, Q., Yan, Z., Huang, Q., Axel, L., Metaxas, D.: Deeptag: An unsupervised deep learning method for motion tracking on cardiac tagging magnetic resonance images. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR). pp. 7261–7...

    2021

  24. [32]

    In: Proceedings of SPIE Medical Imaging (SPIE-MI 2023), San Diego, CA, February 19 – 23, 2023

    Yu, J., Shao, M., Bian, Z., Liang, X., Zhuo, J., Stone, M., Prince, J.L.: New starting point registration method for tagged MRI tongue motion estimation. In: Proceedings of SPIE Medical Imaging (SPIE-MI 2023), San Diego, CA, February 19 – 23, 2023. p. 1246429 (2023)

    2023

  25. [33]

    Medical Physics23(8), 1359–1369 (1996)

    Zhang, S., Douglas, M.A., Yaroslavsky, L., Summers, R.M., Dilsizian, V., Fanana- pazir, L., Bacharach, S.L.: A Fourier based algorithm for tracking SPAMM tags in gated magnetic resonance cardiac images. Medical Physics23(8), 1359–1369 (1996)

    1996

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.