Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

IMC-PINN-FE: A Physics-Informed Neural Network for Patient-Specific Left Ventricular Finite Element Modeling with Image Motion Consistency and Biomechanical Parameter Estimation

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A physics-informed neural network recovers myocardial stiffness and active tension from clinical images in about one minute, then runs whole-cycle heart simulations 75 times faster than conventional FE while matching imaged motion closely.

desk verdict A genuinely useful PINN-FE package with a real speedup, but the headline motion-fidelity gain is measured against the same tracker that built the model, so don't take the 0.85-to-0.93 Dice at face value. read the letter →

arxiv 2506.20696 v1 pith:R5A2MUNR submitted 2025-06-25 physics.med-ph cs.AIeess.IV

Siyu Mu , Wei Xuan Chan , Choon Hwai Yap This is my paper · ORCID
classification physics.med-phcs.AIeess.IV MSC 92C1068T07
keywords physics-informedneuralnetworkcardiacbiomechanicsleftventricularmodelingfiniteelementsimulationimagemotiontrackingmyocardialstiffnessestimationactivetensionproperorthogonaldecomposition
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper proposes IMC-PINN-FE, a physics-informed neural network framework intended to make patient-specific modeling of the beating left ventricle fast enough for clinical use. Its central claim is that myocardial stiffness and peak active tension can be back-computed from image-derived motion plus a few pressure measurements in roughly one minute, where iterative finite element fitting takes hours. The paper further claims that the same solver reproduces the full cardiac cycle about 75 times faster than conventional finite element analysis while matching imaged heart motion better (average Dice 0.927 versus 0.849) and preserving realistic pressure-volume loops. If correct, this would let clinicians obtain a patient's myocardial material properties and simulated heart function in near real time from routine MRI or echocardiography, and would make cardiac digital twins feasible at the point of care.

What carries the argument

The central objects are the subject-specific POD motion modes: the first ten Proper Orthogonal Decomposition basis functions of a single patient's tracked displacement fields across the cardiac cycle, which encode deformation compactly and stabilize PINN training. On top of these modes, the framework stacks an energy-balance loss, namely the difference between myocardial strain energy (exponential passive plus calcium-driven active terms) and the work done by cavity pressure on the endocardium, minimized by the two parameter-estimator networks and by the volume-constrained whole-cycle solver, which also enforces imaged volumes, end-diastolic and peak-systolic pressures, and monotone pressure-volume slopes.

What would settle it

The decisive test is to apply the pipeline to a heart whose true unloaded geometry is measured directly, for example by imaging an explanted heart at zero cavity pressure, alongside its in vivo imaging; then compare $C_{\text{stiffness}}$ and $T_{\max}$ recovered under the one-third-diastole reference assumption with values obtained by fitting the same constitutive model from the measured unloaded state. A systematic offset between the two would identify the reference-state assumption, which the paper itself flags as imperfect, as the dominant error source in the parameter estimates.

Watch

Extended reading notes

Core claim

IMC-PINN-FE establishes that one pipeline can solve both the inverse problem and the forward problem of left-ventricular biomechanics: motion extracted from 4D images, reduced-order motion encoding, and physics constraints are combined so that material parameters are first estimated, then the whole cardiac cycle is simulated. Motion is tracked by a proposed unsupervised Fourier-regularized registration network (EFDL) or a co-attention network, and is compressed into Proper Orthogonal Decomposition modes taken from a single subject's own cardiac cycle rather than from a population. Two small neural networks then estimate the exponential transversely-isotropic stiffness coefficient $C_{\text{stiffness}}$ and the maximum active tension $T_{\max}$ by minimizing the myocardial potential-energy balance at end-diastole and peak systole, recovering values within a few percent of finite element ground truths. The final volume-constrained PINN-FE solver computes pressure and displacement across the cycle from imaged volumes, and deliberately allows the displacements to deviate from the tracked motion so that the solution satisfies physics and image data simultaneously; this is why it claims to beat both conventional FE and a prior PINN-FE that lacks motion consistency.

Load-bearing premise

The load-bearing assumption is that the heart geometry at one third into the diastolic phase is the true zero-pressure, unloaded reference state; every displacement, strain, stiffness, and active-tension value in the pipeline is measured against this presumed state, which the paper acknowledges is imperfect.

Editorial extensions

If this is right

  • Myocardial stiffness and peak active tension can be recovered per patient in about one minute (63 ± 22 seconds per case), replacing iterative inverse finite element fitting that takes several hours per iteration.
  • Whole-cardiac-cycle simulation drops from roughly 7.5 hours to under 6 minutes on the same hardware, a near-75-fold speedup that makes near-real-time use plausible.
  • Simulated LV shapes match image-derived motion markedly better (average Dice 0.927 versus 0.849 for conventional FE) while predicted pressure-volume loops stay nearly identical to FE, supporting the claim that image consistency and physical consistency are compatible.
  • The motion basis comes from a single subject's own cardiac cycle, so no large multi-subject training dataset is needed for motion encoding and patient specificity is improved.
  • Because the solver uses the same governing equations as traditional FE, the framework can also serve as a fast forward model for building large simulation databases for further deep learning training.

Reading between the lines

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

  • With parameter estimation this cheap, the practical ceiling on accuracy in this pipeline shifts from the mechanics solver to the quality of image-derived geometry and motion tracking; better tracking and explicit uncertainty quantification would likely improve every downstream quantity.
  • The single-subject POD basis is a natural scaffolding for cheap parameter sweeps: because the motion modes are already patient-specific, one could simulate many stiffness and tension variants of the same heart to generate training data for other deep learning models.
  • The imperfect load-free reference state, flagged in the paper, implies that a learned predictor of the true unloaded geometry would probably sharpen all pressure-volume and parameter estimates; this is a stated direction of the authors' future work.
  • Because the volume-constrained solver consumes imaged volumes directly, plugging in a Windkessel or other lumped circulation model (which the paper notes is absent) would extend the framework to settings with partial image coverage or to predictive hemodynamics.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper proposes IMC-PINN-FE, a three-stage pipeline for patient-specific left-ventricular biomechanics: (i) deep-learning motion tracking (a new unsupervised method, EFDL, or an existing co-attention STN) that produces 3D+t surface meshes; (ii) a PINN-based estimator that back-computes myocardial stiffness (C_stiffness) and maximal active tension (T_max) from end-diastolic and peak-systolic states; and (iii) a volume-constrained PINN-FE solver that runs a full cardiac-cycle FE simulation at about 75x speedup relative to a traditional volume-constrained FE model, using POD motion modes extracted from a single subject's tracked motion. The authors report parameter estimates within a few percent relative error on FE-generated validation cases, Dice similarity of 0.927 versus 0.849 for the PINN-FE solver compared with traditional FE, and P-V loops close to those of the FE reference.

Significance. If the central claims hold, this would be a meaningful step toward near-real-time, patient-specific cardiac biomechanical modeling: the pipeline is applicable to both MRI and echocardiography, does not require a large multi-subject training set, releases code, and demonstrates a large runtime reduction. The FE-based parameter-estimation validation in Table 2 is a correct self-consistency check and is a useful proof of concept. However, the headline motion-fidelity claim (Dice improvement and 'matches imaged displacements') is not independently validated because the same EFDL tracking that builds the POD basis and volume targets is also used as the ground truth for evaluation. In addition, the physics loss as written enforces zero total potential energy rather than equilibrium, which is a more fundamental concern. The load-free reference-state assumption is acknowledged but not tested for sensitivity, which affects all downstream parameter and P-V estimates. These issues are fixable in revision but are load-bearing for the paper's main claims.

major comments (4)
  1. [§3.3.2 / §2.3.2 / §2.2.2 / §2.3] The Dice improvement reported in Section 3.3.2 is evaluated against a ground truth that is itself produced by the same EFDL tracking that supplies the POD motion modes (Section 2.2.2) and the volume targets (Section 2.3). Section 2.3.2 states that the ground truth for imaged motion is '4D LV mesh reconstructions propagated across the cardiac cycle using EFDL', and Table 1 shows EFDL itself has DSC 0.829 on adult MRI and 0.802 on fetal echo against manual segmentations. Thus the abstract claim that IMC-PINN-FE 'matches imaged displacements more accurately' is not supported by an independent reference; the comparison primarily shows how well the solver reproduces the tracker that generated its reduced-order basis. I recommend adding an evaluation against manual segmentations (at least at ED and ES) or against an independent tracker (e.g., the co-attention STN), and reporting Dice on those references.
  2. [§2.2.3, Eq. (6) and §2.3.1, Eq. (18)] The 'governing equation' loss is defined as the squared total potential energy Π = ∫Ωmyo W dV - ∫Ωendo P·u dA. The equilibrium condition for the hyperelastic problem is the stationarity of Π with respect to the displacement field, δΠ = 0, not Π = 0. Enforcing Π^2 → 0 is a scalar energy-balance constraint, not the weak form of force equilibrium, and it can be satisfied by displacement-pressure pairs that violate local equilibrium. This calls into question the 'physics-informed' aspect of the method. The authors should replace this loss with a proper residual of the weak form (or train the network to minimize Π over admissible displacements), and then re-run the validation experiments.
  3. [§2.2, step 2] The load-free reference state is assumed to be the cardiac state at 1/3 into the diastolic duration. All displacements, Green-Lagrange strains, C_stiffness and T_max estimates, and P-V loops are computed relative to this geometry. The paper acknowledges the assumption is 'imperfect', but no sensitivity analysis is provided. If the true unloaded state differs, the estimated parameters and the entire P-V loop inherit a systematic bias. I request a sensitivity study that varies the reference frame (e.g., using frames at 1/4, 1/3, and 1/2 of diastole, or an estimated unloaded state) and reports the resulting changes in C_stiffness, T_max, and the P-V loop.
  4. [§2.2.5 / Table 2] The FE-based validation in Table 2 is a self-consistency check: the estimator recovers parameters prescribed in noise-free simulations of the same constitutive model. This demonstrates identifiability and internal consistency, but it does not validate the estimated parameters on real clinical data. For the two real-image cases, agreement is reported only as 'within' a literature range, which is a weak and vague criterion given the wide reported ranges. The abstract and discussion should be tempered accordingly; as it stands, the phrase 'rapidly estimates myocardial stiffness and active tension' overstates the level of clinical validation actually provided.
minor comments (4)
  1. [§2.2.2] The text says 'the first 10 dominant modes are retained' but Eq. (5) and the surrounding sentence say 'we use the first 20 bases'. Please state the exact number used in the experiments and ensure consistency throughout.
  2. [§2.2.3] The deformation gradient is written as F = ∂x/∂X = I + ∇u·u. This appears to be a typo; with the displacement u, the standard expression is F = I + ∇u (where ∇u is the displacement gradient). If the dotted form is actually implemented, the constitutive model would be incorrect.
  3. [§3.3.2] There are occasional typos, including 'EDFL' for 'EFDL' in Section 2.1.1 and 'enfocing' in Section 3.3.2. A careful proofread is needed.
  4. [Table 2] For the image-tracking cases, reporting only 'within' a literature interval is not informative. I suggest reporting the numerical value and the width of the interval, or a percentage deviation from the interval midpoint, so the reader can judge the closeness.

Circularity Check

2 steps flagged · score 6.0 of 10

The headline Dice gain is measured against the same EFDL tracker that supplies the POD motion modes and volume targets, so the motion-fidelity claim rests on a self-referential benchmark; the parameter-estimation validation similarly inverts the same equations that generated its synthetic ground truth.

  1. self definitional [Section 2.3.2 (Datasets and Evaluation); see also Section 2.2 steps 1-3 and Section 3.3.2]
    "The ground truth for imaged cardiac motion was defined as 4D LV mesh reconstructions propagated across the cardiac cycle using EDFL due to its highest accuracy (see Table 1) without requiring a large training dataset."

    The same EFDL tracker is used to propagate the reference mesh to all time points (Section 2.2, step 1), to build the POD motion modes in Section 2.2.2, and to supply the imaged LV volume targets used by the volume-constrained solver in Section 2.3. Section 3.3.2 then computes DSC between the predicted shapes and the 'image-derived ground truth (obtained via segmentation and EFDL motion tracking)'. Since the solver's displacement outputs are restricted to the EFDL-derived POD subspace and to EFDL-derived volumes, the reported Dice of 0.927 largely measures how well the physics-regularized solution reconstructs the tracker output that defined its reduced basis, not how well it matches the original images.

  2. other [Section 2.2.5 (Datasets and Evaluation); Table 2; Section 3.2]
    "To evaluate the accuracy of the back-computed parameters (C stiffness and Tmax), we performed FE simulations using predefined values of C stiffness and T max, which served as ground truths. From these simulations, a limited number of time frames are extracted and used as cardiac motion estimation inputs in the estimators."

    The synthetic ground-truth displacements are generated by the same potential-energy-based FE equations (Section 2.2.3, Eqs. 6-16) that the estimators use as their physics loss. The estimator inputs are the POD amplitudes extracted from those simulations, and the outputs are the scaling factors Cstiffness and Tmax that make the energy balance hold for the provided displacements and pressures. Recovering nearly identical parameters (relative errors below 6%) is therefore an inverse-crime self-consistency check: the forward and inverse models share the same constitutive equations, so the 'prediction' is the inverse of the generative model rather than an independent physiological measurement.

full rationale

The paper is not wholly circular: EFDL motion tracking is independently evaluated against manual segmentations on ACDC adult MRI and fetal echocardiography (Table 1), the P-V loop output is compared with traditional FE simulations, and the computational speedup is a genuine engineering result. However, the central motion-fidelity claim is benchmarked against EFDL-propagated meshes that are also the source of the POD motion basis and the volume constraints; therefore the improvement from Dice 0.849 to 0.927 is substantially constructed by the choice of ground truth, not by independent agreement with the original images. The stiffness and active-tension validation adds a second closed loop, since the FE-simulated 'ground truths' are produced with the same Fung-type constitutive law and energy balance that the estimators invert. Neither loop invalidates the method as a fast, physics-constrained solver, but both mean the paper's headline accuracy claims are currently self-referential evaluations rather than externally confirmed predictions. Score 6 reflects this partial circularity: the method has independent content, yet the central claim that it matches imaged displacements more accurately reduces, in its current evaluation, to matching the output of the tracker that generated its reduced-order motion representation.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central modeling chain assumes a chosen load-free reference, assumed fiber/sheet orientations, the Fung-type passive and calcium-activation active constitutive model with literature constants, empirical ED pressure formulas, and image tracking as a proxy for true motion. None of these are derived in the paper; they are standard domain assumptions. The only hand-chosen numbers that affect the outputs are the POD truncation count, the loss weights, and the EFDL Fourier mode count.

free parameters (3)
  • POD mode count = 10 or 20 (text inconsistent)
    The number of retained POD bases is a hand-chosen truncation; Eq. 5 says first 20, Section 2.2.2 says first 10. It controls the reduced-order subspace and therefore the expressible displacement solutions.
  • Loss weights w1-w6 = Not stated in main text (supplementary S3)
    The composite loss in Eq. 22 requires hand-tuned weights; they determine how strongly physics, volume, ED/PS pressure, and slope constraints are enforced.
  • Fourier terms N in EFDL = 4
    Chosen per previous optimization [49]; it fixes the temporal smoothness of the EFDL motion field.
assumptions (4)
  • domain assumption The zero-pressure unloaded LV geometry is the state at 1/3 into the diastolic duration.
    Section 2.2 step 2; all displacements, strains, stiffness and tension estimates are measured from this reference, and the paper explicitly calls the assumption imperfect.
  • domain assumption Fiber and sheet orientations follow a linear transmural helix angle with literature values from the supplementary text.
    The transversely isotropic constitutive model depends on these assumed fiber directions; wrong orientations propagate into stress and parameter estimates.
  • domain assumption The Fung-type passive and calcium-activation active stress models with literature constants describe the myocardium in adult, fetal, and pig cases.
    Equations 8-15; the estimator inverts the same constitutive law used to generate its validation data, so behavior outside this model class is untested.
  • domain assumption Image-derived motion from EFDL or Co-attention STN is a sufficiently accurate proxy for true cardiac motion.
    Motion modes and volume targets come from this tracking; the limitations section acknowledges sensitivity to tracking accuracy.

how reviews work

0 comments
Cite this review

Pith. "Pith review of IMC-PINN-FE: A Physics-Informed Neural Network for Patient-Specific Left Ventricular Finite Element Modeling with Image Motion Consistency and Biomechanical Parameter Estimation." pith.science (2026). https://pith.science/paper/R5A2MUNR

@misc{pith2026250620696,
  author       = {Pith},
  title        = {Pith review of: IMC-PINN-FE: A Physics-Informed Neural Network for Patient-Specific Left Ventricular Finite Element Modeling with Image Motion Consistency and Biomechanical Parameter Estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/R5A2MUNR}},
  note         = {Machine review of arXiv:2506.20696}
}
read the original abstract

Elucidating the biomechanical behavior of the myocardium is crucial for understanding cardiac physiology, but cannot be directly inferred from clinical imaging and typically requires finite element (FE) simulations. However, conventional FE methods are computationally expensive and often fail to reproduce observed cardiac motions. We propose IMC-PINN-FE, a physics-informed neural network (PINN) framework that integrates imaged motion consistency (IMC) with FE modeling for patient-specific left ventricular (LV) biomechanics. Cardiac motion is first estimated from MRI or echocardiography using either a pre-trained attention-based network or an unsupervised cyclic-regularized network, followed by extraction of motion modes. IMC-PINN-FE then rapidly estimates myocardial stiffness and active tension by fitting clinical pressure measurements, accelerating computation from hours to seconds compared to traditional inverse FE. Based on these parameters, it performs FE modeling across the cardiac cycle at 75x speedup. Through motion constraints, it matches imaged displacements more accurately, improving average Dice from 0.849 to 0.927, while preserving realistic pressure-volume behavior. IMC-PINN-FE advances previous PINN-FE models by introducing back-computation of material properties and better motion fidelity. Using motion from a single subject to reconstruct shape modes also avoids the need for large datasets and improves patient specificity. IMC-PINN-FE offers a robust and efficient approach for rapid, personalized, and image-consistent cardiac biomechanical modeling.

Figures

Figures reproduced from arXiv: 2506.20696 by the authors.

Figure 1
Figure 1. The overall framework of IMC-PINN-FE: inputs to the framework include 4D (3D+time) imaging data (e.g., echocardiography or MRI) combined with LV reconstruction obtained via deep learning (DL) and physiological inputs, including peak systolic (PS) and end-diastolic (ED) LV pressures. Outputs include estimates of patient-specific LV myocardial stiffness and maximum active tension, and LV finite element (FE) simulation… view at source ↗
Figure 2
Figure 2. Schematic of validation case generation: A limited number of states (blue points) are selected from the FE simulation, which is performed using a given unloaded surface mesh (red points), to serve as tracking results. S.Mu et al.: Preprint submitted to Elsevier Page 11 of 19 [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. Workflow of the IMC-PINN-FE framework. I. Deep Learning Motion Tracking: 4D full-cycle cardiac images (Echo/MRI) are used to generate myocardial motion meshes via either EFDL (our proposed method, no pre-training required) or Co-attention STN (previous SOTA method, pre-training required). II. Myocardial Stiffness and Maximal Tension Estimator: Starting from the 3D+t surface mesh, the pipeline includes: (1) tetrahedr… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: P–V loop evaluation for (1) adult-2 and (2) fetal-2 cases, comparing IMC-PINN-FE, FE simulation, fixed image tracking, and Buoso et al.’s PINN-FE [7]. Blue represents the P–V loop predicted by IMC-PINN-FE, orange represents the P–V loop from FE simulation, gray represe…
Figure 5
Figure 5. Figure 5: Box plots of DSC evaluation (20 uniformly sampled frames across the cardiac cycle for each of the 6 cases), comparing LV geometries from IMC-PINN-FE, traditional FE, and Buoso et al.’s PINN-FE [7] against image-derived results. Orange indicates the DSC between FE outpu…
Figure 6
Figure 6. Figure 6: Pointwise deformation error maps across different time points throughout the cardiac cycle (Adult-3). Each row corresponds to a different method: IMC-PINN-FE, traditional FE, and Buoso et al.’s PINN-FE [7]. Color encodes the magnitude of pointwise error with respect to…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. HeartUnloadNet: A Weakly-Supervised Cycle-Consistent Graph Network for Predicting Unloaded Cardiac Geometry from Diastolic States

    cs.CV 2025-07 conditional novelty 6.0 of 10

    HeartUnloadNet maps end-diastolic left-ventricular meshes to unloaded geometries in milliseconds on synthetic finite element data, but the reported DSC and sample-efficiency numbers are internally inconsistent.

Reference graph

Works this paper leans on

53 extracted references · 48 canonical work pages · cited by 1 Pith paper

  1. [1]

    Co-attention spatial transformernetworkforunsupervisedmotiontrackingandcardiacstrainanalysisin3dechocardiography

    Ahn, S.S., Ta, K., Thorn, S.L., Onofrey, J.A., Melvinsdottir, I.H., Lee, S., Langdon, J., Sinusas, A.J., Duncan, J.S., 2023. Co-attention spatial transformernetworkforunsupervisedmotiontrackingandcardiacstrainanalysisin3dechocardiography. Medicalimageanalysis84,102711

    work page 2023

  2. [2]

    A model of the mechanics of the left ventricle

    Arts, T., Reneman, R.S., Veenstra, P.C., 1979. A model of the mechanics of the left ventricle. Annals of biomedical engineering 7, 299–318

    work page 1979

  3. [3]

    Avril, S., 2023. Inverse problems in the characterization of soft connective tissue: perspective for reproduction system, in: Biomechanics of the Female Reproductive System: Breast and Pelvic Organs. Elsevier, pp. 115–138

    work page 2023

  4. [4]

    Ontheusageofaveragehausdorff distance for segmentation performance assessment: hidden error when used for ranking

    Aydin,O.U.,Taha,A.A.,Hilbert,A.,Khalil,A.A.,Galinovic,I.,Fiebach,J.B.,Frey,D.,Madai,V.I.,2021. Ontheusageofaveragehausdorff distance for segmentation performance assessment: hidden error when used for ranking. European radiology experimental 5, 1–7

    work page 2021

  5. [5]

    Voxelmorph: a learning framework for deformable medical image registration

    Balakrishnan, G., Zhao, A., Sabuncu, M.R., Guttag, J., Dalca, A.V., 2019. Voxelmorph: a learning framework for deformable medical image registration. IEEE transactions on medical imaging 38, 1788–1800

    work page 2019

  6. [6]

    Deep learning techniques for automatic mri cardiac multi-structures segmentation and diagnosis: Is the problem solved? IEEE Transactions on Medical Imaging 37, 2514–2525

    Bernard, O., Lalande, A., Zotti, C., Cervenansky, F., Yang, X., Heng, P.A., Cetin, I., Lekadir, K., Camara, O., Gonzalez Ballester, M.A., Sanroma, G., Napel, S., Petersen, S., Tziritas, G., Grinias, E., Khened, M., Kollerathu, V.A., Krishnamurthi, G., Rohé, M.M., Pennec, X., Sermesant, M., Isensee, F., Jäger, P., Maier-Hein, K.H., Full, P.M., Wolf, I., En...

    arXiv 2018

  7. [7]

    Personalising left-ventricular biophysical models of the heart using parametric physics-informed neural networks

    Buoso, S., Joyce, T., Kozerke, S., 2021. Personalising left-ventricular biophysical models of the heart using parametric physics-informed neural networks. Medical Image Analysis 71, 102066

    work page 2021

  8. [8]

    Reduced-order modeling of blood flow for noninvasive functional evaluation of coronary artery disease

    Buoso, S., Manzoni, A., Alkadhi, H., Plass, A., Quarteroni, A., Kurtcuoglu, V., 2019. Reduced-order modeling of blood flow for noninvasive functional evaluation of coronary artery disease. Biomechanics and modeling in mechanobiology 18, 1867–1881

    work page 2019

Show all 53 references
  1. [9]

    Assessment of systolic and diastolic ventricular properties via pressure-volume analysis: a guide for clinical, translational, and basic researchers

    Burkhoff, D., Mirsky, I., Suga, H., 2005. Assessment of systolic and diastolic ventricular properties via pressure-volume analysis: a guide for clinical, translational, and basic researchers. American Journal of Physiology-Heart and Circulatory Physiology

    2005

  2. [10]

    Full cardiac cycle asynchronous temporal compounding of 3d echocardiography images

    Chan, W.X., Zheng, Y., Wiputra, H., Leo, H.L., Yap, C.H., 2021. Full cardiac cycle asynchronous temporal compounding of 3d echocardiography images. Medical Image Analysis 74, 102229

    2021

  3. [11]

    Gaussian process regressions for inverse problems and parameter searches in models of ventricular mechanics

    Di Achille, P., Harouni, A., Khamzin, S., Solovyova, O., Rice, J.J., Gurev, V., 2018. Gaussian process regressions for inverse problems and parameter searches in models of ventricular mechanics. Frontiers in physiology 9, 1002

    2018

  4. [12]

    Cardiac Diagnosis and Treatment

    Fowler, N.O., 1980. Cardiac Diagnosis and Treatment. 3 ed., JB Lippincott, Philadelphia

    1980

  5. [13]

    Dynamicfinite-strainmodellingofthehumanleftventricleinhealthanddisease using an immersed boundary-finite element method

    Gao,H.,Carrick,D.,Berry,C.,Griffith,B.E.,Luo,X.,2014. Dynamicfinite-strainmodellingofthehumanleftventricleinhealthanddisease using an immersed boundary-finite element method. IMA journal of applied mathematics 79, 978–1010

    2014

  6. [14]

    Electro-mechanical modelling of in-vivo human left ventricle

    Gao, H., Griffith, B., Carrick, D., McComb, C., Berry, C., Luo, X., 2013. Electro-mechanical modelling of in-vivo human left ventricle. Journal of Electrocardiology 46, e26

    2013

  7. [15]

    Equilibrated warping: Finite element image registration with finite strain equilibrium gap regularization

    Genet, M., Stoeck, C.T., von Deuster, C., Lee, L.C., Kozerke, S., 2018. Equilibrated warping: Finite element image registration with finite strain equilibrium gap regularization. Medical image analysis 50, 1–22

    2018

  8. [16]

    Gmsh: A 3-d finite element mesh generator with built-in pre-and post-processing facilities

    Geuzaine, C., Remacle, J.F., 2009. Gmsh: A 3-d finite element mesh generator with built-in pre-and post-processing facilities. International journal for numerical methods in engineering 79, 1309–1331

    2009

  9. [17]

    Pre-intervention myocardial stress is a good predictor of aortic valvuloplasty outcome for fetal critical aortic stenosis and evolving hlhs

    Green, L., Chan, W.X., Prakash, I., Tulzer, A., Tulzer, G., Yap, C.H., 2024a. Pre-intervention myocardial stress is a good predictor of aortic valvuloplasty outcome for fetal critical aortic stenosis and evolving hlhs. The Journal of Physiology 602, 663–681

  10. [18]

    The dependency of fetal left ventricular biomechanics function on myocardium helix angle configuration

    Green, L., Chan, W.X., Ren, M., Mattar, C.N.Z., Lee, L.C., Yap, C.H., 2023. The dependency of fetal left ventricular biomechanics function on myocardium helix angle configuration. Biomechanics and Modeling in Mechanobiology 22, 629–643

    2023

  11. [19]

    Myocardialbiomechanicaleffectsoffetalaorticvalvuloplasty

    Green,L.,Chan,W.X.,Tulzer,A.,Tulzer,G.,Yap,C.H.,2024b. Myocardialbiomechanicaleffectsoffetalaorticvalvuloplasty. Biomechanics and Modeling in Mechanobiology 23, 1433–1448

  12. [20]

    Passive material properties of intact ventricular myocardium determined from a cylindrical model

    Guccione, J.M., McCulloch, A.D., Waldman, L., 1991. Passive material properties of intact ventricular myocardium determined from a cylindrical model

    1991

  13. [21]

    Feedback attention to enhance unsupervised deep learning image registration in 3d echocardiography

    Hasan, M.K., Luo, Y., Yang, G., Yap, C.H., 2025. Feedback attention to enhance unsupervised deep learning image registration in 3d echocardiography. IEEE Transactions on Medical Imaging doi:10.1109/TMI.2025.3530501

    2025

  14. [22]

    Heyde, B., Claus, P., Jasaityte, R., Barbosa, D., Bouchez, S., Vandenheuvel, M., Wouters, P., Maes, F., D’hooge, J., 2012. Motion and deformation estimation of cardiac ultrasound sequences using an anatomical B-spline transformation model, in: 2012 9th IEEE International Sympo...

    2012

  15. [23]

    Determining optical flow

    Horn, B.K., Schunck, B.G., 1981. Determining optical flow. Artificial intelligence 17, 185–203

    1981

  16. [24]

    Improvingtheaccuracyofregistration-basedbiomechanicalanalysis: a finite element approach to lung regional strain quantification

    Hurtado,D.E.,Villarroel,N.,Retamal,J.,Bugedo,G.,Bruhn,A.,2015. Improvingtheaccuracyofregistration-basedbiomechanicalanalysis: a finite element approach to lung regional strain quantification. IEEE Transactions on Medical Imaging 35, 580–588

    2015

  17. [25]

    Reviewing 3d convolutional neural network approaches for medical image segmentation

    Ilesanmi, A.E., Ilesanmi, T.O., Ajayi, B.O., 2024. Reviewing 3d convolutional neural network approaches for medical image segmentation. Heliyon . S.Mu et al.:Preprint submitted to ElsevierPage 18 of 19 IMC-PINN-FE

    2024

  18. [26]

    Intracardiac pressures in the human fetus

    Johnson, P., Maxwell, D., Tynan, M., Allan, L., 2000. Intracardiac pressures in the human fetus. Heart 84, 59–63

    2000

  19. [27]

    Elastix:atoolboxforintensity-basedmedicalimageregistration

    Klein,S.,Staring,M.,Murphy,K.,Viergever,M.A.,Pluim,J.P.,2009. Elastix:atoolboxforintensity-basedmedicalimageregistration. IEEE transactions on medical imaging 29, 196–205

    2009

  20. [28]

    Scattereddatainterpolationwithmultilevelb-splines

    Lee,S.,Wolberg,G.,Shin,S.,1997. Scattereddatainterpolationwithmultilevelb-splines. IEEETransactionsonVisualizationandComputer Graphics 3, 228–244. doi:10.1109/2945.620490

    1997

  21. [29]

    URL:https://arxiv.org/abs/2409.02070,arXiv:2409.02070

    Luo,Y.,Sesia,D.,Wang,F.,Wu,Y.,Ding,W.,Huang,J.,Shi,F.,Shah,A.,Kaural,A.,Mayet,J.,Yang,G.,Yap,C.,2024.Explicitdifferentiable slicing and global deformation for cardiac mesh reconstruction. URL:https://arxiv.org/abs/2409.02070,arXiv:2409.02070

    2024 arXiv

  22. [30]

    A comparison of methods for gradient field estimation on simplicial meshes

    Mancinelli, C., Livesu, M., Puppo, E., 2019. A comparison of methods for gradient field estimation on simplicial meshes. Computers & Graphics 80, 37–50

    2019

  23. [31]

    Model reduction techniques for fast blood flow simulation in parametrized geometries

    Manzoni, A., Quarteroni, A., Rozza, G., 2012. Model reduction techniques for fast blood flow simulation in parametrized geometries. International journal for numerical methods in biomedical engineering 28, 604–625

    2012

  24. [32]

    Marstal, K., Berendsen, F., Staring, M., Klein, S., 2016. Simpleelastix: A user-friendly, multi-lingual library for medical image registration, in: Proceedings of the IEEE conference on computer vision and pattern recognition workshops, pp. 134–142

    2016

  25. [33]

    Normal pressures in the heart and great vessels

    Merck Manuals Professional Version, 2025. Normal pressures in the heart and great vessels. URL:https://www.merckmanuals.com/ professional/multimedia/table/normal-pressures-in-the-heart-and-great-vessels. accessed: 2025-05-11

    2025

  26. [34]

    Strain energy density

    MOOSE Framework, 2025. Strain energy density. URL:https://mooseframework.inl.gov/source/materials/ StrainEnergyDensity.html. accessed: 2025-02-24

    2025

  27. [35]

    A neural network finite element approach for high speed cardiac mechanics simulations

    Motiwale, S., Zhang, W., Feldmeier, R., Sacks, M.S., 2024. A neural network finite element approach for high speed cardiac mechanics simulations. Computer Methods in Applied Mechanics and Engineering 427, 117060

    2024

  28. [36]

    Effects of local activation times on the tension development of human cardiomyocytes in a computational model

    Müller, A., Kovacheva, E., Schuler, S., Dössel, O., Baron, L., 2018. Effects of local activation times on the tension development of human cardiomyocytes in a computational model. Current Directions in Biomedical Engineering 4, 247–250

    2018

  29. [37]

    Nvidia modulus

    NVIDIA Modulus Team, 2023. Nvidia modulus. Computer Software. URL:https://github.com/NVIDIA/modulus/tree/main

    2023

  30. [38]

    Biomechanics of human fetal hearts with critical aortic stenosis

    Ong, C.W., Ren, M., Wiputra, H., Mojumder, J., Chan, W.X., Tulzer, A., Tulzer, G., Buist, M.L., Mattar, C.N.Z., Lee, L.C., et al., 2021. Biomechanics of human fetal hearts with critical aortic stenosis. Annals of biomedical engineering 49, 1364–1379

    2021

  31. [39]

    Numerical solution of parametrized navier–stokes equations by reduced basis methods

    Quarteroni, A., Rozza, G., 2007. Numerical solution of parametrized navier–stokes equations by reduced basis methods. Numerical methods for partial differential equations: an international journal 23, 923–948

    2007

  32. [40]

    Contractile properties of developing human fetal cardiac muscle

    Racca, A.W., Klaiman, J.M., Pioner, J.M., Cheng, Y., Beck, A.E., Moussavi-Harami, F., Bamshad, M.J., Regnier, M., 2016. Contractile properties of developing human fetal cardiac muscle. The Journal of physiology 594, 437–452

    2016

  33. [41]

    Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations

    Raissi, M., Perdikaris, P., Karniadakis, G.E., 2019. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational physics 378, 686–707

    2019

  34. [42]

    Model reduction for fluids, using balanced proper orthogonal decomposition

    Rowley, C.W., 2005. Model reduction for fluids, using balanced proper orthogonal decomposition. International Journal of Bifurcation and Chaos 15, 997–1013

    2005

  35. [43]

    Advancements in finite element modeling for cardiac device leads and 3d heart models

    Salih, A., Hamandi, F., Goswami, T., 2024. Advancements in finite element modeling for cardiac device leads and 3d heart models. Bioengineering 11, 564

    2024

  36. [44]

    Highspatialresolutionmulti-organfiniteelementmodelingofventricular-arterialcoupling

    Shavik,S.M.,Jiang,Z.,Baek,S.,Lee,L.C.,2018. Highspatialresolutionmulti-organfiniteelementmodelingofventricular-arterialcoupling. Frontiers in physiology 9, 119

    2018

  37. [45]

    3ddigitalgeometrydesignsforpoland’ssyndromeusingmagicsandgeomagic®freeform®

    VanHeerden,I.,Fossey,A.,Booysen,G.J.,2018. 3ddigitalgeometrydesignsforpoland’ssyndromeusingmagicsandgeomagic®freeform®. Rapid Prototyping Journal 24, 229–236

    2018

  38. [46]

    Image-based predictive modeling of heart mechanics

    Wang, V., Nielsen, P., Nash, M., 2015. Image-based predictive modeling of heart mechanics. Annual review of biomedical engineering 17, 351–383

    2015

  39. [47]

    Wang, V.Y., Lam, H.I., Ennis, D.B., Cowan, B.R., Young, A.A., Nash, M.P., 2010. Cardiac active contraction parameters estimated from magnetic resonance imaging, in: Statistical Atlases and Computational Models of the Heart: First International Workshop, STACOM 2010, andCardiac...

    2010

  40. [48]

    First evidence of depressed contractility in the border zone of a human myocardial infarction

    Wenk,J.F.,Klepach,D.,Lee,L.C.,Zhang,Z.,Ge,L.,Tseng,E.E.,Martin,A.,Kozerke,S.,GormanIII,J.H.,Gorman,R.C.,etal.,2012. First evidence of depressed contractility in the border zone of a human myocardial infarction. The Annals of thoracic surgery 93, 1188–1193

    2012

  41. [49]

    Cardiac motion estimation from medical images: a regularisation framework applied on pairwise image registration displacement fields

    Wiputra, H., Chan, W.X., Foo, Y.Y., Ho, S., Yap, C.H., 2020. Cardiac motion estimation from medical images: a regularisation framework applied on pairwise image registration displacement fields. Scientific reports 10, 18510

    2020

  42. [50]

    Fluid mechanics of human fetal right ventricles from image-based computational fluid dynamics using 4d clinical ultrasound scans

    Wiputra, H., Lai, C.Q., Lim, G.L., Heng, J.J.W., Guo, L., Soomar, S.M., Leo, H.L., Biwas, A., Mattar, C.N.Z., Yap, C.H., 2016. Fluid mechanics of human fetal right ventricles from image-based computational fluid dynamics using 4d clinical ultrasound scans. American Journal of ...

    2016

  43. [51]

    Anovelmri-basedfiniteelement modeling method for calculation of myocardial ischemia effect in patients with functional mitral regurgitation

    Zhang,Y.,Wang,V.Y.,Morgan,A.E.,Kim,J.,Ge,L.,Guccione,J.M.,Weinsaft,J.W.,Ratcliffe,M.B.,2020. Anovelmri-basedfiniteelement modeling method for calculation of myocardial ischemia effect in patients with functional mitral regurgitation. Frontiers in physiology 11, 158

    2020

  44. [52]

    Morphological, functional, and biomechanical progression of lv remodelling in a porcine model of hfpef

    Zheng, Y., Chan, W.X., Charles, C.J., Richards, A.M., Lee, L.C., Leo, H.L., Yap, C.H., 2022. Morphological, functional, and biomechanical progression of lv remodelling in a porcine model of hfpef. Journal of biomechanics 144, 111348. S.Mu et al.:Preprint submitted to ElsevierP...

    2022

  45. [2010]

    Proceedings 1, Springer. pp. 194–203

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.