Pith. sign in

REVIEW 3 major objections 6 minor 74 references

Robust Physics-Informed Neural Network Approach for Estimating Heterogeneous Elastic Properties from Noisy Displacement Data

T0 review · 3 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read IE-PINN recovers absolute-scale heterogeneous elasticity maps from noisy displacement data, where direct finite-difference inversion collapses.

desk verdict Solid, well-ablated incremental advance in PINN-based inverse elasticity; the decoupled strain network is genuinely useful, but the absolute-scale calibration claim is shakier than the paper admits because the calibration factor inherits any boundary extrapolation error and the synthetic force is derived from the true model. read the letter →

arxiv 2506.14036 v2 pith:5IZRPEQK submitted 2025-06-16 cs.LG

Tatthapong Srikitrungruang , Matthew Lemon , Sina Aghaee Dabaghan Fard , Jaesung Lee , Yuxiao Zhou This is my paper · ORCID
classification cs.LG MSC 68T0765M3274B05
keywords inverseelasticityphysics-informedneuralnetworkselastographyYoung'smodulusestimationPoisson'srationoiserobustnessabsolutescalecalibrationheterogeneousmaterials
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

The paper is trying to establish that inverse elasticity—recovering the spatial distribution of material stiffness from measured deformation—can be made stable and noise-tolerant, despite being classically ill-posed. Its proposed IE-PINN separates the job across three neural networks: one fits the noisy displacement field, one predicts strain directly, and one predicts the elasticity parameters. That separation keeps noise from being amplified through repeated differentiation, and a two-phase procedure recovers the absolute Young's modulus scale by matching the predicted boundary stress to the known applied loading force. The claim is supported on synthetic plane-stress data with Gaussian noise at signal-to-noise ratios from 1000 down to 100, where the method reports low errors while a direct finite-difference baseline fails. If the claim holds, elastography-style measurements could produce absolute stiffness maps without priors on mean modulus or internal stress.

What carries the argument

The mechanism that carries the argument is the three-network decomposition. The displacement network is fitted to the noisy observations and acts as a smoother; the strain network predicts strain directly and is penalized for disagreeing with strain computed from the displacement network, so the PDE residual is built from strain-network outputs rather than from doubly differentiated noisy displacements; the elasticity network outputs $\hat{E}$ and $\hat{\nu}$. The elasticity prediction is constrained only to an arbitrary mean during training, which keeps the phase-one problem well-conditioned. The second load-bearing mechanism is the calibration step: the traction boundary condition gives $F = \int \hat{c}\,\hat{\sigma}_{xx}^{(b)}\,\mathrm{d}y$, so the scalar $\hat{c}$ is fixed by numerical integration of predicted boundary stress, and the final modulus is $\hat{c}\hat{E}(x,y)$. All coordinates enter through positional encoding, all hidden layers use sine activations, and the three networks are pretrained sequentially.

What would settle it

A decisive test would be to run the method on a phantom with a known stiffness inclusion while deliberately misspecifying the applied force by 10 percent; if the recovered absolute modulus does not shift by roughly 10 percent, the calibration equation fails, and if the method cannot handle a different heterogeneity pattern at SNR 100, the robustness claim is narrower than stated.

Watch

Extended reading notes

Core claim

The central discovery is that noise sensitivity in inverse elasticity is not an unavoidable feature of the problem, but a consequence of how derivatives are taken. IE-PINN therefore replaces second derivatives of noisy data with a dedicated strain network that is trained to agree with the displacement-derived strain, so that the equilibrium equations are enforced on a smooth strain representation. The absolute-scale problem is handled separately by the traction boundary condition: after phase one yields a relative modulus map $\hat{E}(x,y)$ and a relative boundary stress $\hat{\sigma}_{xx}^{(b)}$, phase two computes the multiplier $\hat{c} = F / \sum_i \hat{\sigma}_{xx}^{(b)}(x_b,y_i) h$ from the known applied force $F$, and reports $E_{\mathrm{absolute}}(x,y) = \hat{c}\hat{E}(x,y)$. With arbitrary mean-modulus constraints, the calibration still produces consistent absolute errors, and Poisson's ratio is recovered simultaneously without an incompressibility assumption. The author's claim, in short, is that decoupling displacement, strain, and elasticity networks plus boundary-force calibration turns a noisy ill-posed inversion into a tractable one.

Load-bearing premise

Phase-two calibration assumes the total applied force on the loaded boundary is known and that the predicted relative boundary stress integrates to a value proportional to that force; in the synthetic benchmarks the force is computed from the true model, so any real measurement error in force magnitude, boundary geometry, or relative stress bias would enter the absolute modulus scale directly.

Editorial extensions

If this is right

  • Clinical elastography could estimate absolute tissue stiffness maps from noisy displacement data using only the applied load, without knowing the mean modulus beforehand.
  • Compressible materials with spatially varying Poisson's ratio can be handled directly, removing the common incompressibility assumption that earlier methods relied on.
  • Displacement data at SNR down to 100 could be used without aggressive pre-denosing that might blur stiffness boundaries.
  • The two-phase calibration removes the need for prior knowledge of internal or boundary stress distributions, which are usually unavailable in practice.

Reading between the lines

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

  • Beyond the paper's synthetic benchmarks, the calibration logic predicts that any measurement error in the applied force $F$ enters the absolute modulus scale linearly, so real-world accuracy reports should include force calibration uncertainty.
  • The strain-discrepancy trick is not specific to elasticity; other inverse PDE problems that differentiate noisy data, such as thermal conductivity imaging or hydraulic tomography, could borrow the decoupling idea.
  • The paper does not test three-dimensional or low-resolution clinical data, but the same two-phase scheme would need reworked finite-difference kernels and boundary integration before the claim could extend there.
  • A direct test on experimentally collected digital image correlation data with a known stiffness inclusion would show whether the synthetic-noise robustness transfers to real noise structure.
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

3 major / 6 minor

Summary. The manuscript proposes IE-PINN, a physics-informed neural network framework for estimating spatially heterogeneous Young's modulus and Poisson's ratio from noisy displacement data. The method uses three separate neural networks for displacement, strain, and elasticity, and trains them by minimizing a weighted sum of displacement fitting, strain discrepancy, equilibrium residual, and a mean-modulus constraint. A second phase calibrates the absolute scale of Young's modulus by comparing the predicted boundary stress to the applied force. Experiments on synthetic 2D plane-stress datasets at SNR 1000, 500, and 100 show lower error than Elastnet and than an ablation without the strain network; additional ablations support the choices of sinusoidal activation, positional encoding, and pretraining.

Significance. If the results are reproducible, the proposed architecture addresses a real limitation of existing inverse-elasticity PINNs: sensitivity to noise and reliance on a known mean Young's modulus. The two-phase calibration idea is simple and potentially practical when the applied load is known. The paper contains useful ablations (strain network, activation function, positional encoding, pretraining) and reports results on a public-style synthetic benchmark. However, the validation is narrow: one 2D synthetic problem family, a single baseline method, and the critical calibration step is tested with a force value derived from the ground-truth model, so the absolute-scale claim is not yet convincingly demonstrated in realistic conditions. The significance is moderate; the paper is likely to interest the PINN and elastography communities if the calibration is properly stress-tested.

major comments (3)
  1. [Section 4.3, Eq. (26), Supplementary Note S2] The absolute-scale calibration is not validated independently: the applied force F in the synthetic benchmark is computed from the true Young's modulus and boundary strain (Supplementary Note S2), so the calibration is exactly consistent with the ground truth by construction. The paper presents F as 'experimentally measured' but no experiment with an independently measured force or perturbed F is reported. Since Eq. (27) makes the entire recovered modulus field proportional to F, the authors should report a sensitivity analysis with perturbed F and, if possible, test on data where F is measured separately from the displacement data.
  2. [Section 4.2, Eqs. (14),(15),(17)] The boundary stress used in the calibration equation (26) is evaluated at x = x_b, but the strain discrepancy loss (Eq. 14) is evaluated on the (Nx-1)x(Ny-1) interior, the equilibrium residual (Eq. 15) on the (Nx-3)x(Ny-3) interior, and the mean-modulus loss (Eq. 17) on the interior. Therefore the boundary stress is an extrapolation outside the support of all physics losses, and any systematic error in that extrapolation enters the scaling factor c_hat in Eq. (26) linearly, scaling the entire field in Eq. (27). The paper should quantify the boundary stress error (e.g., report relative error of sigma_xx at the loaded edge) or incorporate boundary-adjacent constraints so the calibration is not solely reliant on an unconstrained extrapolation.
  3. [Sections 2.3 and Figures 6-7] The central robustness claim is supported only by a single synthetic 2D plane-stress setup with Gaussian noise; no error bars or repeated-seed statistics are reported in Figures 6 and 7, and the comparison is limited to one baseline method (Elastnet). To substantiate the claims of robustness and state-of-the-art performance, the authors should report mean and variance over multiple independent runs (different noise realizations and network initializations) and compare with at least one additional inverse-elasticity method, particularly one that incorporates a learned denoising step.
minor comments (6)
  1. [Section 2.6] The text states that training was done 'with pretraining described in Section 10,' but the manuscript has no Section 10; the reference should be to the appropriate part of the Experimental Section.
  2. [References] References [70] and [73] are identical (Sitzmann et al. 2020); one should be removed or replaced with a distinct relevant reference.
  3. [Section 4.1, Eq. (9)] The finite-difference kernels in Eqs. (10)-(11) appear to implement central differences with a factor of 2 that is not accounted for by the division by h_t; the authors should clarify the exact convolution convention, including the role of the factor 1/2 if the kernels are meant to be averaged over adjacent rows or columns.
  4. [Figure 4] The figure reports 'MRE across 50 independent datasets' but does not specify whether the plotted quantity is a mean or median, and no measure of dispersion is shown; please provide this information and add error bars or box plots.
  5. [Table S1] The table reports MAE values without standard deviations, and the text does not state how many random initializations and noise realizations were used; please provide this information for reproducibility.
  6. [Section 4.2, Eq. (17)] The mean-modulus loss L_E is defined as the sum of absolute deviations of each predicted E(i,j) from E_c, which is not exactly a constraint on the spatial mean; the text says it constrains the mean, so please clarify the relationship (e.g., by noting that it is a softened mean constraint or changing the loss to penalize the difference of means).

Circularity Check

0 steps flagged · score 1.0 of 10

No by-construction circularity: the absolute-scale calibration in Eq. (26) uses an independently supplied traction force; the synthetic benchmark's F being ground-truth-derived is a validation caveat, not a circular reduction.

full rationale

The derivation chain is self-contained. Phase 1 trains the displacement, strain, and elasticity networks against noisy displacement data, strain compatibility, equilibrium residuals, and an arbitrary mean-modulus constraint, so it produces only a relative Young's modulus field. Phase 2 introduces one scalar c = F / (sum of predicted boundary stress x h) (Eq. 26) using the externally imposed loading force F, and then sets E_absolute = c * E_hat (Eq. 27). This is a calibration identity, not a hidden fit: the denominator is the network's predicted boundary stress, which is not made equal to the true stress by any loss term or construction; the network must still learn the relative field from noisy displacement data. The absolute-scale result therefore is not equivalent to the inputs by definition. In the synthetic benchmark, Supplementary Note S2 computes F from the true boundary strain and elasticity, so the validation uses the exact force that generated the data; this is a favorable benchmark condition and any error in the unregularized boundary stress would propagate linearly through c, but it is a benchmark caveat rather than circular reasoning. There are no load-bearing self-citations: the dataset and Elastnet comparisons are external, and no prior result by the present authors is invoked to force the outcome. The boundary-stress extrapolation concern is a correctness/robustness risk, not a circular step.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

No new physical entities are introduced; the three neural networks are computational components, not physical postulates. The main hand-chosen numbers are the mean modulus constraint, loss weights, and positional encoding settings, all of which the paper tests for robustness.

free parameters (3)
  • Mean modulus constraint E_c = varied: 0.20, 0.2327, 0.25, 0.30, 0.40
    Phase 1 imposes an arbitrary mean Young's modulus to regularize the ill-posed problem; the paper shows robustness to this choice after calibration (Section 2.4, Figure 8).
  • Loss weights lambda_u, lambda_eps, lambda_r, lambda_E = 2, 1, 3, 0.02
    Hyperparameters chosen by hand to balance data fitting, strain discrepancy, PDE residual, and mean constraint (Section 4.2, Eq. 12).
  • Positional encoding parameters f, omega = f=0.0001, omega=64
    Input encoding hyperparameters chosen for the 2D coordinate inputs (Section 4.4).
assumptions (6)
  • domain assumption Linear elasticity PDE with isotropic material under plane stress governs the deformation (Eq. 5-8).
    The entire loss is built on this constitutive model; not valid for large deformations, anisotropy, or 3D.
  • domain assumption The applied loading force (traction) F is known and measurable, and the boundary is a straight edge with normal in x (Eq. 23-26).
    Phase 2 absolute calibration uses F as an input; if F is unknown, only relative modulus is recovered.
  • domain assumption Displacement measurements are available on a regular pixel grid so that the fixed 2x2 and 3x3 convolution kernels apply (Eq. 2-4, 9-11).
    The finite-difference scheme assumes uniform spacing h and t; irregular or sparse measurement grids are not handled.
  • domain assumption Gaussian zero-mean noise with known SNR is an adequate model for measurement error (Supplementary Note S1).
    All robustness experiments are generated under this noise model; real DIC/MRE noise may be correlated or non-Gaussian.
  • standard math The convolution kernels in Eq. (10)-(11) correctly approximate the divergence of the stress tensor under static equilibrium.
    This is a standard finite-difference approximation on a uniform grid.
  • ad hoc to paper A sufficiently expressive neural network trained by Adam can minimize the weighted loss to reach the correct solution (Section 4.2).
    Convergence and global optimality are not proven; the method relies on empirical training success.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Robust Physics-Informed Neural Network Approach for Estimating Heterogeneous Elastic Properties from Noisy Displacement Data." pith.science (2026). https://pith.science/paper/5IZRPEQK

@misc{pith2026250614036,
  author       = {Pith},
  title        = {Pith review of: Robust Physics-Informed Neural Network Approach for Estimating Heterogeneous Elastic Properties from Noisy Displacement Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5IZRPEQK}},
  note         = {Machine review of arXiv:2506.14036}
}
read the original abstract

Accurately estimating spatially heterogeneous elasticity parameters, particularly Young's modulus and Poisson's ratio, from noisy displacement measurements remains significantly challenging in inverse elasticity problems. Existing inverse estimation techniques are often limited by instability, pronounced sensitivity to measurement noise, and difficulty in recovering absolute-scale Young's modulus. This work presents a novel Inverse Elasticity Physics-Informed Neural Network (IE-PINN) specifically designed to robustly reconstruct heterogeneous distributions of elasticity parameters from noisy displacement data based on linear elasticity physics. IE-PINN integrates three distinct neural network architectures dedicated to separately modeling displacement fields, strain fields, and elasticity distributions, thereby significantly enhancing stability and accuracy against measurement noise. Additionally, a two-phase estimation strategy is introduced: the first phase recovers relative spatial distributions of Young's modulus and Poisson's ratio, and the second phase calibrates the absolute scale of Young's modulus using imposed loading boundary conditions. Additional methodological innovations, including positional encoding, sine activation functions, and a sequential pretraining protocol, further enhance the model's performance and robustness. Extensive numerical experiments demonstrate that IE-PINN effectively overcomes critical limitations encountered by existing methods, delivering accurate absolute-scale elasticity estimations even under severe noise conditions. This advancement holds substantial potential for clinical imaging diagnostics and mechanical characterization, where measurements typically encounter substantial noise.

Figures

Figures reproduced from arXiv: 2506.14036 by the authors.

Figure 1
Figure 1. Framework for heterogeneous elasticity estimation from noisy displacement data. The framework consists of [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. The prediction field of mechanical quantities. The model is applied to a measured displacement that contains [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Young’s modulus scale calibration procedure. Upon [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (4 more)
Figure 5
Figure 5. Figure 5: Young’s modulus (E) and Poisson’s ratio (ν) predictions and corresponding error maps were obtained from different models. All models were trained using the same noisy displacement data with a signal-to-noise (SNR) ratio of 1000. (i) IE-PINN (Proposed) incorporates both…
Figure 6
Figure 6. Figure 6: Predicted (a) Young’s modulus (E) and (b) Poisson’s ratio (ν), along with their corresponding error maps, evaluated across varying noise levels (signal-to-noise ratio, SNR). IE-PINN was trained using the same displacement data across three different SNRs: (i) SNR = 100…
Figure 7
Figure 7. Figure 7: Robustness of prediction errors across different [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 9
Figure 9. Figure 9: Performance comparison across different activation functions. The prediction errors in Young’s modulus (E) and Poisson’s ratio (ν) are presented for models using various activation functions in neu￾ral networks (Fitting denotes both displacement and strain networks, El…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

74 extracted references · 47 canonical work pages

  1. [1]

    Boutouyrie, P

    P. Boutouyrie, P. Chowienczyk, J. D. Humphrey, G. F. Mitchell, Arterial stiffness and cardiovascular risk in hypertension, Circulation research 128 (7) (2021) 864–886

    work page 2021

  2. [2]

    J. A. Chirinos, P. Segers, T. Hughes, R. Townsend, Large-artery stiffness in health and disease: Jacc state-of-the-art review, Journal of the American College of Cardiology 74 (9) (2019) 1237–1263

    work page 2019

  3. [3]

    Y . Wang, C. Negron, A. Khoshnaw, S. Edwards, H. Vu, J. Quatela, N. Park, F. Maldonado, C. Demarest, V . Simon, et al., Sensory artificial cilia for in situ monitoring of airway physiological properties, Proceedings of the National Academy of Sciences 121 (46) (2024) e2412086121

    work page 2024

  4. [4]

    K. Kim, S. Edwards, K. Fuxa, H. Lin, S. Shrestha, H. Fan, N. Diaz, J. Berinstein, R. Naik, Y . Zhou, et al., Mucosa- interfacing capsule for in situ sensing the elasticity of biological tissues, Advanced Materials Technologies 2401487

    work page

  5. [5]

    J. Han, X. Dong, Z. Yin, S. Zhang, M. Li, Z. Zheng, M. C. Ugurlu, W. Jiang, H. Liu, M. Sitti, Actuation-enhanced multifunctional sensing and information recognition by magnetic artificial cilia arrays, Proceedings of the National Academy of Sciences 120 (42) (2023) e2308301120

    work page 2023

  6. [6]

    Goenezen, J.-F

    S. Goenezen, J.-F. Dord, Z. Sink, P. E. Barbone, J. Jiang, T. J. Hall, A. A. Oberai, Linear and nonlinear elastic modulus imaging: An application to breast cancer diagnosis, IEEE Transactions on Medical Imaging 31 (8) (2012) 1628–1637.doi:10.1109/tmi.2012.2201497. URLhttp://dx.doi.org/10.1109/TMI.2012.2201497

    work page arXiv 2012

  7. [7]

    F.-S. Quan, K. S. Kim, Medical applications of the intrinsic mechanical properties of single cells, Acta Biochimica et Biophysica Sinica 48 (10) (2016) 865–871.doi:10.1093/abbs/gmw081. URLhttp://dx.doi.org/10.1093/abbs/gmw081

    work page doi:10.1093/abbs/gmw081 2016

  8. [8]

    H. Wu, W. Liang, Y . Jiao, H. Song, Q. Peng, H. Luo, Y . Zhang, J. Xu, F. Dong, A preliminary comparative study of young’s modulus versus shear modulus in the diagnosis of breast cancer, Ultrasound Quarterly 35 (1) (2019) 88–92.doi:10.1097/ruq.0000000000000434. URLhttp://dx.doi.org/10.1097/RUQ.0000000000000434

    work page doi:10.1097/ruq.0000000000000434 2019

Show all 74 references
  1. [9]

    C. Wang, Y . Wu, X. Dong, M. Armacki, M. Sitti, In situ sensing physiological properties of biological tissues using wireless miniature soft robots, Science advances 9 (23) (2023) eadg3988

    2023

  2. [10]

    J. Xue, S. Singh, Y . Zhou, A. Perdomo-Pantoja, Y . Tian, N. Gupta, T. F. Witham, W. L. Grayson, T. P. Weihs, A biodegradable 3d woven magnesium-based scaffold for orthopedic implants, Biofabrication 14 (3) (2022) 034107

    2022

  3. [11]

    Y . Zhou, W. Grayson, Three-dimensional printing of scaffolds for facial reconstruction, MRS Bulletin 47 (1) (2022) 91–97

    2022

  4. [12]

    J. Wang, B. Zhou, I. Parkinson, C. D. L. Thomas, J. G. Clement, N. Fazzalari, X. E. Guo, Trabecular plate loss and deteriorating elastic modulus of femoral trabecular bone in intertrochanteric hip fractures, Bone research 1 (1) (2013) 346–354

    2013

  5. [13]

    J. A. Choren, S. M. Heinrich, M. B. Silver-Thorn, Young’s modulus and volume porosity relationships for additive manufacturing applications, Journal of Materials Science 48 (15) (2013) 5103–5112. doi:10.1007/ s10853-013-7237-5. URLhttp://dx.doi.org/10.1007/s10853-013-7237-5

    2013 doi

  6. [14]

    A. Roy, T. Babuska, B. Krick, G. Balasubramanian, Machine learned feature identification for predicting phase and young’s modulus of low-, medium- and high-entropy alloys, Scripta Materialia 185 (2020) 152–158. doi: 10.1016/j.scriptamat.2020.04.016. URLhttp://dx.doi.org/10.101...

    2020 doi

  7. [15]

    H. Yan, Q. Li, B. Yang, Y . Yang, Y . Wang, H. Zhang, Modeling and prediction method for young’s moduli of ti al- loys based on residual muti-layer perceptron, JOM 77 (1) (2024) 76–90. doi:10.1007/s11837-024-06942-3 . URLhttp://dx.doi.org/10.1007/s11837-024-06942-3

    2024 doi

  8. [16]

    Q. Mao, K. Su, Y . Zhou, M. Hossaini-Zadeh, G. S. Lewis, J. Du, V oxel-based micro-finite element analysis of dental implants in a human cadaveric mandible: Tissue modulus assignment and sensitivity analyses, Journal of the mechanical behavior of biomedical materials 94 (2019) 229–237

    2019

  9. [17]

    K. Su, Y . Zhou, M. Hossaini-Zadeh, J. Du, Effects of implant buccal distance on peri-implant strain: A micro-ct based finite element analysis, journal of the mechanical behavior of biomedical materials 116 (2021) 104325

    2021

  10. [18]

    Oliver, G

    W. Oliver, G. Pharr, An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments, Journal of Materials Research 7 (6) (1992) 1564–1583. doi: 15 Inverse Elasticity Physics-informed Neural Network (IE-PINN)A PREP...

    1992

  11. [19]

    Oliver, G

    W. Oliver, G. Pharr, Measurement of hardness and elastic modulus by instrumented indentation: Advances in understanding and refinements to methodology, Journal of Materials Research 19 (1) (2004) 3–20. doi: 10.1557/jmr.2004.19.1.3. URLhttp://dx.doi.org/10.1557/jmr.2004.19.1.3

    2004 doi

  12. [20]

    D. Xu, T. Harvey, E. Begiristain, C. Domínguez, L. Sánchez-Abella, M. Browne, R. B. Cook, Measuring the elastic modulus of soft biomaterials using nanoindentation, Journal of the Mechanical Behavior of Biomedical Materials 133 (2022) 105329.doi:10.1016/j.jmbbm.2022.105329. URL...

    2022

  13. [21]

    Y . Zhou, J. Du, Atomic force microscopy (afm) and its applications to bone-related research, Progress in Biophysics and Molecular Biology 176 (2022) 52–66

    2022

  14. [22]

    Muthupillai, D

    R. Muthupillai, D. Lomas, P. Rossman, J. F. Greenleaf, A. Manduca, R. L. Ehman, Magnetic resonance elastogra- phy by direct visualization of propagating acoustic strain waves, science 269 (5232) (1995) 1854–1857

    1995

  15. [23]

    Manduca, P

    A. Manduca, P. V . Bayly, R. L. Ehman, A. Kolipaka, T. J. Royston, I. Sack, R. Sinkus, B. E. Van Beers, Mr elastography: Principles, guidelines, and terminology, Magnetic resonance in medicine 85 (5) (2021) 2377–2390

    2021

  16. [24]

    S. B. Lang, Ultrasonic method for measuring elastic coefficients of bone and results on fresh and dried bovine bones, IEEE Transactions on Biomedical Engineering BME-17 (2) (1970) 101–105. doi:10.1109/tbme.1970. 4502706. URLhttp://dx.doi.org/10.1109/tbme.1970.4502706

    1970

  17. [25]

    C. M. Buntin, F. H. Silver, Noninvasive assessment of mechanical properties of peripheral arteries, Annals of Biomedical Engineering 18 (5) (1990) 549–566.doi:10.1007/bf02364617. URLhttp://dx.doi.org/10.1007/BF02364617

    1990 doi

  18. [26]

    M. A. Michael A., J.-J. Orteu, H. W. Schreier, Digital Image Correlation (DIC), Springer US, 2009, p. 1–37. doi:10.1007/978-0-387-78747-3\_5. URLhttp://dx.doi.org/10.1007/978-0-387-78747-3_5

    2009 doi

  19. [27]

    B. K. Bay, T. S. Smith, D. P. Fyhrie, M. Saad, Digital volume correlation: three-dimensional strain mapping using x-ray tomography, Experimental mechanics 39 (1999) 217–226

    1999

  20. [28]

    T.-A. Chen, B. B. Zhao, R. A. Balbin, S. Sharma, D. Ha, T. J. Kamp, Y . Zhou, F. Zhao, Engineering a robust and anisotropic cardiac-specific extracellular matrix scaffold for cardiac patch tissue engineering, Matrix Biology Plus 23 (2024) 100151

    2024

  21. [29]

    Tanoto, Z

    H. Tanoto, Z. Zhou, K. Chen, R. Qiu, H. Fan, J. Z. Chen, E. Milton, Y . Zhou, M. Liu, Predicting biaxial failure strengths of aortic tissues using a dispersed fiber failure model, Extreme Mechanics Letters 75 (2025) 102287

    2025

  22. [30]

    Tanoto, H

    H. Tanoto, H. Fan, D. Ha, C. G. Ambrose, E. O. Klineberg, Y . Zhou, Quantifying the relation between aging-related trabecular bone microstructure and mechanical properties with digital volume correlation approach, Extreme Mechanics Letters 73 (2024) 102265

    2024

  23. [31]

    H. Gao, B. Ji, I. L. Jäger, E. Arzt, P. Fratzl, Materials become insensitive to flaws at nanoscale: Lessons from nature, Proceedings of the National Academy of Sciences 100 (10) (2003) 5597–5600. doi:10.1073/pnas. 0631609100. URLhttp://dx.doi.org/10.1073/pnas.0631609100

    2003 doi

  24. [32]

    Paraskevoulakos, S

    C. Paraskevoulakos, S. Ghosh, T. Andriollo, A. Michel, Sensitivity study using synthetic 3d image datasets to investigate the effect of noise artefacts on digital volume correlation, Experimental Mechanics 64 (5) (2024) 595–624

    2024

  25. [33]

    P. L. Gould, Introduction to Linear Elasticity, Springer New York, 2013. doi:10.1007/978-1-4614-4833-4 . URLhttp://dx.doi.org/10.1007/978-1-4614-4833-4

    2013 doi

  26. [34]

    S. I. Kabanikhin, Definitions and examples of inverse and ill-posed problems, Journal of Inverse and Ill-posed Problems 16 (4) (Jan. 2008).doi:10.1515/jiip.2008.019. URLhttp://dx.doi.org/10.1515/JIIP.2008.019

    2008 doi

  27. [35]

    P. E. Barbone, A. A. Oberai, Elastic modulus imaging: some exact solutions of the compressible elastography inverse problem, Physics in Medicine and Biology 52 (6) (2007) 1577–1593. doi:10.1088/0031-9155/52/6/ 003. URLhttp://dx.doi.org/10.1088/0031-9155/52/6/003 16 Inverse Ela...

    2007 doi

  28. [36]

    P. E. Barbone, C. E. Rivas, I. Harari, U. Albocher, A. A. Oberai, Y . Zhang, Adjoint-weighted variational formulation for the direct solution of inverse problems of general linear elasticity with full interior data, International Journal for Numerical Methods in Engineering 81...

    2009 doi

  29. [37]

    O. A. Babaniyi, A. A. Oberai, P. E. Barbone, Direct error in constitutive equation formulation for plane stress inverse elasticity problem, Computer Methods in Applied Mechanics and Engineering 314 (2017) 3–18. doi: 10.1016/j.cma.2016.10.026. URLhttp://dx.doi.org/10.1016/j.cma...

    2017 doi

  30. [38]

    M. M. Doyley, P. M. Meaney, J. C. Bamber, Evaluation of an iterative reconstruction method for quantitative elastography, Physics in Medicine and Biology 45 (6) (2000) 1521–1540.doi:10.1088/0031-9155/45/6/309. URLhttp://dx.doi.org/10.1088/0031-9155/45/6/309

    2000 doi

  31. [39]

    A. A. Oberai, N. H. Gokhale, G. R. Feij o, Solution of inverse problems in elasticity imaging using the adjoint method, Inverse Problems 19 (2) (2003) 297–313.doi:10.1088/0266-5611/19/2/304. URLhttp://dx.doi.org/10.1088/0266-5611/19/2/304

    2003 doi

  32. [40]

    A. A. Oberai, N. H. Gokhale, M. M. Doyley, J. C. Bamber, Evaluation of the adjoint equation based algorithm for elasticity imaging, Physics in Medicine and Biology 49 (13) (2004) 2955–2974. doi:10.1088/0031-9155/49/ 13/013. URLhttp://dx.doi.org/10.1088/0031-9155/49/13/013

    2004 doi

  33. [41]

    M. M. Doyley, S. Srinivasan, S. A. Pendergrass, Z. Wu, J. Ophir, Comparative evaluation of strain-based and model-based modulus elastography, Ultrasound in Medicine & Biology 31 (6) (2005) 787–802. doi: 10.1016/j.ultrasmedbio.2005.02.005. URLhttp://dx.doi.org/10.1016/j.ultrasm...

    2005 doi

  34. [42]

    D. Smyl, S. Bossuyt, D. Liu, Openqsei: A matlab package for quasi static elasticity imaging, SoftwareX 9 (2019) 73–76.doi:10.1016/j.softx.2019.01.004. URLhttp://dx.doi.org/10.1016/j.softx.2019.01.004

    2019 doi

  35. [43]

    Sutton, J

    M. Sutton, J. Yan, V . Tiwari, H. Schreier, J. Orteu, The effect of out-of-plane motion on 2d and 3d digital image correlation measurements, Optics and Lasers in Engineering 46 (10) (2008) 746–757. doi:10.1016/j. optlaseng.2008.05.005. URLhttp://dx.doi.org/10.1016/j.optlaseng....

    2008 doi

  36. [44]

    B. Pan, L. Yu, D. Wu, High-accuracy 2d digital image correlation measurements with bilateral telecentric lenses: Error analysis and experimental verification, Experimental Mechanics 53 (9) (2013) 1719–1733. doi: 10.1007/s11340-013-9774-x. URLhttp://dx.doi.org/10.1007/s11340-013-9774-x

    2013 doi

  37. [45]

    Freddi, G

    A. Freddi, G. Olmi, L. Cristofolini, Experimental Stress Analysis for Materials and Structures: Stress Anal- ysis Models for Developing Design Methodologies, Springer International Publishing, 2015. doi:10.1007/ 978-3-319-06086-6. URLhttp://dx.doi.org/10.1007/978-3-319-06086-6

    2015 doi

  38. [46]

    C.-T. Chen, G. X. Gu, Learning hidden elasticity with deep neural networks, Proceedings of the National Academy of Sciences 118 (31) (Jul. 2021).doi:10.1073/pnas.2102721118. URLhttp://dx.doi.org/10.1073/pnas.2102721118

    2021 doi

  39. [47]

    Haghighat, M

    E. Haghighat, M. Raissi, A. Moure, H. Gomez, R. Juanes, A physics-informed deep learning framework for inversion and surrogate modeling in solid mechanics, Computer Methods in Applied Mechanics and Engineering 379 (2021) 113741.doi:10.1016/j.cma.2021.113741. URLhttp://dx.doi.o...

    2021

  40. [48]

    H. Gao, M. J. Zahr, J.-X. Wang, Physics-informed graph neural galerkin networks: A unified framework for solving pde-governed forward and inverse problems, Computer Methods in Applied Mechanics and Engineering 390 (2022) 114502.doi:10.1016/j.cma.2021.114502. URLhttp://dx.doi.o...

    2022

  41. [49]

    L. T. P. Nguyen, B. H. Liu, Machine learning framework for determination of elastic modulus without contact model fitting, International Journal of Solids and Structures 256 (2022) 111976. doi:10.1016/j.ijsolstr. 2022.111976. URLhttp://dx.doi.org/10.1016/j.ijsolstr.2022.111976

    2022

  42. [50]

    B. Ni, H. Gao, A deep learning approach to the inverse problem of modulus identification in elasticity, MRS Bulletin 46 (1) (2021) 19–25.doi:10.1557/s43577-020-00006-y. URLhttp://dx.doi.org/10.1557/s43577-020-00006-y 17 Inverse Elasticity Physics-informed Neural Network (IE-PI...

    2021 doi

  43. [51]

    Mouloodi, H

    S. Mouloodi, H. Rahmanpanah, C. Burvill, H. M. Davies, Prediction of load in a long bone using an artificial neural network prediction algorithm, journal of the mechanical behavior of biomedical materials 102 (2020) 103527

    2020

  44. [52]

    Raissi, P

    M. Raissi, P. Perdikaris, G. Karniadakis, Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations, Journal of Computational Physics 378 (2019) 686–707.doi:10.1016/j.jcp.2018.10.045....

    2019 doi

  45. [53]

    Raissi, A

    M. Raissi, A. Yazdani, G. E. Karniadakis, Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations, Science 367 (6481) (2020) 1026–1030.doi:10.1126/science.aaw4741. URLhttp://dx.doi.org/10.1126/science.aaw4741

    2020 doi

  46. [54]

    C. Song, T. A. Alkhalifah, Wavefield reconstruction inversion via physics-informed neural networks, IEEE Transactions on Geoscience and Remote Sensing 60 (2022) 1–12.doi:10.1109/tgrs.2021.3123122. URLhttp://dx.doi.org/10.1109/TGRS.2021.3123122

    2022

  47. [55]

    Sahin, M

    T. Sahin, M. von Danwitz, A. Popp, Solving forward and inverse problems of contact mechanics using physics- informed neural networks, Advanced Modeling and Simulation in Engineering Sciences 11 (1) (May 2024). doi:10.1186/s40323-024-00265-3. URLhttp://dx.doi.org/10.1186/s40323...

    2024 doi

  48. [56]

    Mallampati, M

    A. Mallampati, M. Almekkawy, Measuring tissue elastic properties using physics based neural networks, in: 2021 IEEE UFFC Latin America Ultrasonics Symposium (LAUS), IEEE, 2021, p. 1–4. doi:10.1109/laus53676. 2021.9639231. URLhttp://dx.doi.org/10.1109/LAUS53676.2021.9639231

    2021

  49. [57]

    Z. Min, Z. M. C. Baum, S. U. Saeed, M. Emberton, D. C. Barratt, Z. A. Taylor, Y . Hu, Biomechanics-Informed Non-rigid Medical Image Registration and its Inverse Material Property Estimation with Linear and Nonlinear Elasticity, Springer Nature Switzerland, 2024, p. 564–574.doi...

    2024 doi

  50. [58]

    Kianian, S

    O. Kianian, S. Sarrami, B. Movahedian, M. Azhari, Pinn-based forward and inverse bending analysis of nanobeams on a three-parameter nonlinear elastic foundation including hardening and softening effect using nonlocal elasticity theory, Engineering with Computers (May 2024).doi...

    2024 doi

  51. [59]

    C. M. Hamel, K. N. Long, S. L. B. Kramer, Calibrating constitutive models with full-field data via physics informed neural networks, Strain 59 (2) (Nov. 2022).doi:10.1111/str.12431. URLhttp://dx.doi.org/10.1111/str.12431

    2022 doi

  52. [60]

    Shukla, A

    K. Shukla, A. D. Jagtap, J. L. Blackshire, D. Sparkman, G. Em Karniadakis, A physics-informed neural network for quantifying the microstructural properties of polycrystalline nickel using ultrasound data: A promising approach for solving inverse problems, IEEE Signal Processin...

    2022

  53. [61]

    Ragoza, K

    M. Ragoza, K. Batmanghelich, Physics-Informed Neural Networks for Tissue Elasticity Reconstruction in Magnetic Resonance Elastography, Springer Nature Switzerland, 2023, p. 333–343. doi:10.1007/ 978-3-031-43999-5\_32. URLhttp://dx.doi.org/10.1007/978-3-031-43999-5_32

    2023 doi

  54. [62]

    Kamali, M

    A. Kamali, M. Sarabian, K. Laksari, Elasticity imaging using physics-informed neural networks: Spatial discovery of elastic modulus and poisson’s ratio, Acta Biomaterialia 155 (2023) 400–409. doi:10.1016/j.actbio.2022. 11.024. URLhttp://dx.doi.org/10.1016/j.actbio.2022.11.024

    2023 doi

  55. [63]

    Kamali, K

    A. Kamali, K. Laksari, Physics-informed unets for discovering hidden elasticity in heterogeneous materials, Journal of the Mechanical Behavior of Biomedical Materials 150 (2024) 106228. doi:10.1016/j.jmbbm.2023. 106228. URLhttp://dx.doi.org/10.1016/j.jmbbm.2023.106228

    2024

  56. [64]

    C. Chen, G. X. Gu, Physics-informed deep-learning for elasticity: Forward, inverse, and mixed problems, Advanced Science 10 (18) (Apr. 2023).doi:10.1002/advs.202300439. URLhttp://dx.doi.org/10.1002/advs.202300439

    2023 doi

  57. [65]

    L.-P. Chen, Mehryar mohri, afshin rostamizadeh, and ameet talwalkar: Foundations of machine learning, second edition: The mit press, cambridge, ma, 2018, 504 pp., cdn $96.53 (hardback), isbn 9780262039406, Statistical 18 Inverse Elasticity Physics-informed Neural Network (IE-P...

    2019 doi

  58. [66]

    W. Cao, W. Zhang, An analysis and solution of ill-conditioning in physics-informed neural networks, Journal of Computational Physics 520 (2025) 113494.doi:10.1016/j.jcp.2024.113494. URLhttp://dx.doi.org/10.1016/j.jcp.2024.113494

    2025

  59. [67]

    J. Y . Rho, R. B. Ashman, C. H. Turner, Young’s modulus of trabecular and cortical bone material: Ultrasonic and microtensile measurements, Journal of Biomechanics 26 (2) (1993) 111–119. doi:10.1016/0021-9290(93) 90042-d. URLhttp://dx.doi.org/10.1016/0021-9290(93)90042-D

    1993 doi

  60. [68]

    K. Tai, M. Dao, S. Suresh, A. Palazoglu, C. Ortiz, Nanoscale heterogeneity promotes energy dissipation in bone, Nature materials 6 (6) (2007) 454–462

    2007

  61. [69]

    Y . Zhou, M. J. Kastner, T. B. Tighe, J. Du, Elastic modulus mapping for bovine cortical bone from submillimeter-to submicron-scales using peakforce tapping atomic force microscopy, Extreme mechanics letters 41 (2020) 101031

    2020

  62. [71]

    Mildenhall, P

    B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoorthi, R. Ng, Nerf: representing scenes as neural radiance fields for view synthesis, Communications of the ACM 65 (1) (2021) 99–106. doi:10.1145/3503250. URLhttp://dx.doi.org/10.1145/3503250

    2021 doi

  63. [72]

    Vaswani, N

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, I. Polosukhin, Attention is all you need (2017).doi:10.48550/ARXIV.1706.03762. URLhttps://arxiv.org/abs/1706.03762

    2017 doi arXiv

  64. [73]

    Sitzmann, J

    V . Sitzmann, J. N. P. Martel, A. W. Bergman, D. B. Lindell, G. Wetzstein, Implicit neural representations with periodic activation functions (2020).doi:10.48550/ARXIV.2006.09661. URLhttps://arxiv.org/abs/2006.09661

    2020 doi arXiv

  65. [74]

    J. R. Barber, Elasticity, 3rd Edition, Solid Mechanics and Its Applications, Springer, Dordrecht, Netherlands, 2009

    2009

  66. [75]

    Sadd, M. H. Sadd, Elasticity, 2nd Edition, Academic Press, 2014. 19 Inverse Elasticity Physics-informed Neural Network (IE-PINN)A PREPRINT Supplementary Figures Supplementary Figure S1: Comparison between automatic differentiation and finite differentiation.Both differentiatio...

    2014

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.