Abstract
Accurate estimation of myocardial motion can help to better understand the pathophysiological processes of ischemic heart diseases. However, because of partial and noisy image-derived measurements on the cardiac kinematics, the performance of model-based motion estimation relies heavily on the assumption of noise distribution on the measurement data. While existing studies of model-based motion estimation have often adopted the \(\mathcal {H}_{2}\) criteria (e.g. least square error) based on fixed model constraints from mathematical or mechanical nature, we present a robust estimation framework with an adaptive biomechanical model constraint using dual \(\mathcal {H}_{\infty }\) criteria for the first time. Comparing to the minimization of average gaussian error in \(\mathcal {H}_{2}\) criteria, our \(\mathcal {H}_{\infty }\) criteria aims to minimize the maximum error without regarding the noise distribution. In this work, our dual estimation framework consists of two iterative \(\mathcal {H}_{\infty }\) filters: One filter for the kinematics estimation and another one for the elasticity estimation. At each time step, heart kinematics is estimated with sub-optimal fixed material parameters, followed by an elasticity property recovering given these sub-optimal kinematic state estimates. Such coupled estimation processes are iteratively repeated as necessary until convergence. We evaluate the performance of dual estimation framework on synthetic data, cine image sequences, and human image sequence. Our dual estimation framework shows a higher tolerance of noise than the conventional extended Kalman filter. The results obtained by both synthetic data of varying noises and magnetic resonance image sequences demonstrate the accuracy and robustness of the proposed strategy.
Similar content being viewed by others
References
Abaqus homepage: https://www.3ds.com/products-services/simulia/products/abaqus/
Abd-Elmoniem K, Stuber M, Prince J (2008) Direct three-dimensional myocardial strain tensor quantification and tracking using zHARP. Med Image Anal 12(6):778–786
Axel L, Dougherty L (1989) MR imaging of motion with spatial modulation of magnetization. Radiology 171(3):841–845
Bathe K, Wilson E (1976) Numerical methods in finite element analysis. Prentice-Hall, New Jersey
Benayoun S, Ayache N (1998) Dense non-rigid motion estimation in sequences of medical images using differential constraints. Int J Comput Vis 26(1):25–40
Britton OJ, Bueno-Orovio A, Van Ammel K, Lu HR, Towart R, Gallacher DJ, Rodriguez B (2013) Experimentally calibrated population of models predicts and explains intersubject variability in cardiac cellular electrophysiology. Proc Nat Acad Sci 110(23):E2098–E2105
Burger M, Modersitzki J, Ruthotto L (2013) A hyperelastic regularization energy for image registration. SIAM J Sci Comput 35(1):B132–B148
Cheng TM, Savkin AV, Celler BG, Su SW, Wang L (2008) Nonlinear modeling and control of human heart rate response during exercise with various work load intensities. IEEE Trans Biomed Eng 55(11):2499–2508
Chiang P, Cai Y, Mak KH, Zheng J (2013) A B-spline approach to phase unwrapping in tagged cardiac mri for motion tracking. Magn Reson Med 69 (5):1297–1309
Cochet H, Bullier E, Gerbaud E, Durieux M, Godbert Y, Lederlin M, Coste P, Barat JL, Laurent F, Montaudon M (2013) Absolute quantification of left ventricular global and regional function at nuclear MPI using ultrafast CZT SPECT: initial validation versus cardiac MR. J Nuclear Med 54(4):556–563
Cox M, Driessen N, Bouten C, Baaijens F (2006) Mechanical characterization of anisotropic planar biological soft tissues using large indentation: a computational feasibility study. J Biomech Eng 128(3):428–436
Declerck J, Denney T, Ozturk C, O’Dell W, McVeigh E (2000) Left ventricular motion reconstruction from planar tagged MR images: a comparison. Phys Med Biol 45(6):1611–1632
Dempster AP, Laird NM, Rubin DB (1977) Maximum likelihood from incomplete data via the em algorithm. J Royal Stat Soc 39(1):1–38
Denney T, Prince J (1994) 3D displacement field reconstruction from planar tagged cardiac MR images. In: Proceedings of the IEEE Workshop on Biomedical Image Analysis, pp 51–60
Didinsky G, Pan Z, Basar T (1995) Parameter identification for uncertain plants using \(\mathcal {H}_{\infty }\) methods. Automatica 31(9):1227–1250
Gao H, Allan A, McComb C, Luo X, Berry C (2014) Left ventricular strain and its pattern estimated from cine CMR and validation with DENSE. Phys Med Biol 59(13):3637–3656
Gao Z, Xiong H, Liu X, Zhang H, Ghista D, Wu W, Li S (2017) Robust estimation of carotid artery wall motion using the elasticity-based state-space approach. Med Image Anal 37:1–21
Gilliam A, Epstein F, Acton S (2009) Cardiac motion recovery via active trajectory field models. IEEE Trans Inf Technol Biomed 13:226–235
Glad T, Ljung L (2000) Control theory. Taylor & Francis, London
Glass L, Hunter P, McCulloch A (1991) Theory of heart. Springer-Verlag, New York
Gopal S, Otaki Y, Arsanjani R, Berman D, Terzopoulos D, Slomka P (2013) Combining active appearance and deformable superquadric models for LV segmentation in cardiac MRI. In: SPIE Medical Imaging, p. 86690G. International Society for Optics and Photonics
Halmos PR, Sunder VS (1978) Bounded integral operators on L 2 spaces. Springer, Berlin
Hameeteman K, Klooster R, Selwaness M, Lugt A, Witteman J, Niessen W, Klein S (2013) Carotid wall volume quantification from magnetic resonance images using deformable model fitting and learning-based correction of systematic errors. Phys Med Biol 58(5):1605–1623
Han L, Hipwel J, Tanner C, Taylor Z, Mertzanidou T, Cardoso J, Ourselin S, Hawkes D (2012) Development of patient-specific biomechanical models for predicting large breast deformation. Phys Med Biol 57(2):455–472
Honarvar M, Sahebjavaher R, Salcudean S, Rohling R (2012) Sparsity regularization in dynamic elastography. Phys Med Biol 57(19):5909–5927
Jahanzad Z, Liew Y, Bilgen M, McLaughlin R, Leong C, Chee K, Aziz Y, Ung N, Lai K, Ng S, Lim E (2015) Regional assessment of LV wall in infarcted heart using tagged MRI and cardiac modelling. Phys Med Biol 60 (10):4015–4031
Janzc RF, Kubertc BR, Moriarty TF, Grimma AF (1974) Deformation of the diastolic left ventricle —II. Nonlinear geometric effects. J Biomech 7(6):509–516
Kerwin W, Prince J (1999) The kriging update model and recursive space-time function estimation. IEEE Trans Signal Process 47(11):2942–2952
Klein G, Huesman R (2000) Elastic material model mismatch effects in deformable motion estimation. IEEE Trans Nuclear Sci 47(3):1000–1005
Liu H, Hu H, Sinusas A, Shi P (2012) An H ∞ approach for elasticity properties reconstruction. Med Phys 39(1):475–481
Liu H, Shi P (2004) Cardiac motion and elasticity characterization with iterative sequential H ∞ criteria. In: International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), pp 34–42
Liu H, Shi P (2009) Maximum a posteriori strategy for the simultaneous motion and material property estimation of the heart. IEEE Trans Biomed Eng 56(2):378–389
Marchesseau S, Delingette H, Sermesant M, Cabrera-Lozoya R, Tobon-Gomez C, Moireau P, i Ventura RMF, Lekadir K, Hernandez A, Garreau M, Donal E, Leclercq C, Duckett SG, Rhode K, Rinaldi CA, Frangi AF, Razavi R, Chapelle D, Ayacha N (2013) Personalization of a cardiac electromechanical model using reduced order unscented Kalman filtering from regional volumes. Med Image Anal 17(7):816– 829
McEachen J, Duncan J (1997) Shape-based tracking of left ventricular wall motion. IEEE Trans Med Imaging 16(3):270–283
McEachen J, Nehorai A, Duncan J (2000) Multiframe temporal estimation of cardiac nonrigid motion. IEEE Trans Image Process 9(4):651–665
Metaxas DN, Kakadiaris IA (2002) Elastically adaptive deformable models. IEEE Trans Pattern Anal Mach Intell 24(10):1310–1321
Meyer C, Bonnemains L, Marċon F, Marie PY, Felblinger J, Vuissoz PA (2014) Longitudinal myocardial peak velocities using high temporal resolution phase-contrast and simple averaging are comparable to tissue doppler echocardiography. Magn Reson Mater Phys Biol Med 27(3):211–218
Meyer F, Constable R, Sinusas A, Duncan J (1996) Tracking myocardial deformation using spatially constrained velocities. IEEE Trans Med Imaging 15(4):453–465
Moireau P, Chapelle D, Le Tallec P (2008) Joint state and parameter estimation for distributed mechanical systems. Comput Methods Appl Mech Eng 197:659–677
Moireau P, Chapelle D, Le Tallec P (2009) Filtering for distributed mechanical systems using position measurements: perspectives in medical imaging. Inverse Probl 25(3):35,010–35,035
Nash MP (1998) Mechanics and material properties of the heart using an anatomically accurate mathematical model. PhD thesis, The University of Auckland, New Zealand
O’Brien RT, Kiriakidis K (2003) Convergence and performance in H ∞ -based parameter identification. IFAC Proc 36(11):127–132
Ozturk C, McVeigh E (2000) Four-dimensional B-spline based motion analysis of tagged MR images: introduction and in vivo validation. Phys Med Biol 45(6):1683–702
Papademetris X, Sinusas A, Dione D, Constable R, Duncan J (2002) Estimation of 3-D left ventricular deformation from medical images using biomechanical models. IEEE Trans Med Imaging 21(7):786–799
Pavarino E, Neves L, Machado J, de Godoy M, Shiyou Y, Momente JC, Zafalon GF, Pinto A, Valêncio CR (2013) Free tools and strategies for the generation of 3d finite element meshes: modeling of the cardiac structures. Int J Biomed Imaging 2013:1–11
Pelc N, Herfkens R, Shimakawa A, Enzmann D (1991) Phase contrast cine magnetic resonance imaging. Magn Reson Q 7(4):229–254
Perreard I, Pattison A, Doyley M, McGarry M, Barani Z, Houten E, Weaver J, Paulsen K (2010) Effects of frequency- and direction-dependent elastic materials on linearly elastic MRE image reconstructions. Phys Med Biol 55(22):6801–15
Sermesant M, Delingette H, Ayache N (2006) An electromechanical model of the heart for image analysis and simulation. IEEE Trans Med Imaging 25(5):612–625
Sermesant M, Moireau P, Camara O, Sainte-Marie J, Andriantsimiavona R, Cimrman R, Hill D, Chapell D, Razavi R (2005) Cardiac function estimation from MRI using a heart model and data assimilation: Advances and difficulties. International Workshop on Functional Imaging and Modeling of the Heart, pp 325–337
Shen X, Deng L (1999) A dynamic system approach to speech enhancement using the \(\mathcal {H}_{\infty }\) filtering algorithm. IEEE Trans Speech Audio Process 7(4):391–399
Shi P, Liu H (2003) Stochastic finite element framework for simultaneous estimation of cardiac kinematics functions and material parameters. Med Image Anal 7(4):445–464
Shi P, Sinusas A, Constable RT, Duncan J (1999) Volumetric deformation analysis using mechanics-based data fusion: applications in cardiac motion recovery. Int J Comput Vis 35(1):87–107
Simon D (2006) Optimal state estimation: Kalman, H-infinity, and Nonlinear Approaches. Wiley, NJ
Tavakoli V, Amini AA (2013) A survey of shaped-based registration and segmentation techniques for cardiac images. Comput Vis Image Underst 117(9):966–989
Taylor JR Classical mechanics, University Science Books, USA
Tsap L, Goldgof D, Sarkar S (2000) Nonrigid motion analysis based on dynamic refinement of finite element models. IEEE Trans Pattern Anal Mach Intell 22 (5):526–543
Vaccarella A, Momi ED, Enquobahrie A, Ferrigno G (2013) Unscented Kalman filter based sensor fusion for robust optical and electromagnetic tracking in surgical navigation. IEEE Trans Instrum Measur 62(7):2067–2081
Wang D, Wang B, Zhao S, Yao H, Liu H (2017) View-based 3D object retrieval with discriminative views. Neurocomputing 252:58–66
Wong A, Liu H, Sinusas A, Shi P (2003) Spatiotemporal active region model for simultaneous segmentation and motion estimation of the whole heart. In: IEEE workshop on Variational, Geometric and Level Set Methods in Computer Vision, pp 193–200
Young A, Kraitchman D, Dougherty L, Axel L (1995) Tracking and finite element analysis of stripe deformation in magnetic resonance tagging. IEEE Trans Med Imaging 14(3):413–421
Young A, Prince J (2013) Cardiovascular magnetic resonance: Deeper insights through bioengineering. Ann Rev Biomed Eng 15:433–461
Zerhouni E, Parish D, Rogers W, Yang A, Shapiro E (1988) Human heart: tagging with MR imaging – a method for noninvasive assessment of myocardial motion. Radiology 169(1):59–63
Zhang Z, Sahn DJ, Song X (2013) Right ventricular strain analysis from 3D echocardiography by using temporally diffeomorphic motion estimation. In: International Conference on Functional Imaging and Modeling of the Heart. Springer, pp 474–482
Zhao S, Chen L, Yao H, Zhang Y, Sun X (2015) Strategy for dynamic 3D depth data matching towards robust action retrieval. Neurocomputing 151:533–543
Zhao S, Yao H, Zhang Y, Wang Y, Liu S (2015) View-based 3D object retrieval via multi-modal graph learning. Signal Process 112:110–118
Zhu Y, Pelc N (1999) A spatiotemporal model of cyclic kinematics and its application to analyzing nonrigid motion with MR velocity images. IEEE Trans Med Imaging 18(7):557–569
Acknowledgements
This work was supported by National Key Research and Development Program of China (2016YFC1300302), Natural Science Foundation of China (No. 61427807, 61525106, 61771464), and Science Technology and Innovation Committee of Shenzhen for Research Projects (SGLH20150213143207911, JCYJ20151030151431727).
Author information
Authors and Affiliations
Corresponding authors
Appendix
Appendix
The term \(\mathcal {H}_{\infty }\) originates from the name of the mathematical space over which the optimization takes place(more details can be found in [53]). \(\mathcal {H}\) stands for ”Hardy space”. ∞ implies that it is designed to accomplish min-max restrictions. The motivation of the \(\mathcal {H}_{\infty }\) filter is to solve problems where the uncertainties of the model and inputs are problematic. For an \(\mathcal {H}_{\infty }\) criteria, the transfer function (the measure of how good of the estimator is), from the input disturbances to the estimation error, shall be required to have a system gain that conforms to an upper bound (See (16) and (30) This criteria represents a family of solutions where the peak energy gain of the transfer function from the input disturbances to the estimate error are less than an upper bound γ.
It is observed that the smaller the γ value, the smaller the estimation error. On the other hand, the Riccati equation has to have a positive definite solution. In our practical implementation and the following deviations, matrix Q is chosen to be an identity matrix. However, depending on the confidence on the inputs, Q can be of any form and it would not change the fundamental deduction of the procedure. Thus
where m a x{e i g(A)} denotes the maximum eigenvalue of the matrix A, and ξ is a constant larger than 1 to ensure that γ is always greater than a certain optimal performance level. If the γ value is too close to the optimal performance level, i.e. ξ ≈ 1, it might lead to numerical errors because the matrix \(\mathcal {R}(t)\) is now close to a singular matrix.
Rights and permissions
About this article
Cite this article
Gao, Z., Zhang, H., Wang, D. et al. Robust recovery of myocardial kinematics using dual \(\mathcal {H}_{\infty }\) criteria. Multimed Tools Appl 77, 23043–23071 (2018). https://doi.org/10.1007/s11042-017-5395-1
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11042-017-5395-1