Abstract
This paper introduces a new parameterization of diffeomorphic deformations for the characterization of the variability in image ensembles. Dense diffeomorphic deformations are built by interpolating the motion of a finite set of control points that forms a Hamiltonian flow of self-interacting particles. The proposed approach estimates a template image representative of a given image set, an optimal set of control points that focuses on the most variable parts of the image, and template-to-image registrations that quantify the variability within the image set. The method automatically selects the most relevant control points for the characterization of the image variability and estimates their optimal positions in the template domain. The optimization in position is done during the estimation of the deformations without adding any computational cost at each step of the gradient descent. The selection of the control points is done by adding a L 1 prior to the objective function, which is optimized using the FISTA algorithm.
Similar content being viewed by others
Notes
If the image domain Ω is discretized into a regular lattice of N im voxels, then Y(t) could be seen as a dN im-dimensional vectors of the voxels positions that are mapped back via the inverse deformation.
Note that this is not exactly the elastic net paradigm, since we do not use the usual Euclidean norm on the momentum vectors for the L 2 penalty (α 0,i t α 0,i ) but the L 2 metric induced by the metric K g instead (α 0,i t K g α 0,i ).
References
Allassonnière, S., Trouvé, A., & Younes, L. (2005). Geodesic shooting and diffeomorphic matching via textured meshes. In Proc. of EMMCVPR (pp. 365–381).
Allassonnière, S., Amit, Y., & Trouvé, A. (2007). Towards a coherent statistical framework for dense deformable template estimation. Journal of the Royal Statistical Society. Series B. Statistical Methodology, 69(1), 3–29.
Allassonnière, S., Kuhn, E., & Trouvé, A. (2010). Construction of bayesian deformable models via a stochastic approximation algorithm: a convergence study. Bernoulli, 16(3), 641–678.
Arsigny, V., Commowick, O., Pennec, X., & Ayache, N. (2006). A log-euclidean framework for statistics on diffeomorphisms. In LNCS: Vol. 4190. Proc. MICCAI (pp. 924–931). Berlin: Springer.
Beck, A., & Teboulle, M. (2009). A fast iterative shrinkage-thresholding algorithm for linear inverse problems. SIAM Journal on Imaging Sciences, 2(1), 183–202.
Beg, M. F., Miller, M. I., Trouvé, A., & Younes, L. (2005). Computing large deformation metric mappings via geodesic flows of diffeomorphisms. International Journal of Computer Vision, 61, 139–157.
Cootes, T. F., Edwards, G. J., & Taylor, C. J. (2001). Active appearance models. IEEE Transactions on Pattern Analysis and Machine Intelligence, 23(6), 681–685.
Dupuis, P., Grenander, U., & Miller, M. (1998). Variational problems on flows of diffeomorphisms for image matching. Quarterly of Applied Mathematics, 56(3), 587–600.
Durrleman, S. (2010). Statistical models of currents for measuring the variability of anatomical curves, surfaces and their evolution. Thèse de sciences (phd thesis), Université de Nice-Sophia Antipolis.
Durrleman, S., Pennec, X., Trouvé, A., & Ayache, N. (2009). Statistical models of sets of curves and surfaces based on currents. Medical Image Analysis, 13(5), 793–808.
Durrleman, S., Fillard, P., Pennec, X., Trouvé, A., & Ayache, N. (2011a). Registration, atlas estimation and variability analysis of white matter fiber bundles modeled as currents. NeuroImage, 55(3), 1073–1090.
Durrleman, S., Prastawa, M., Gerig, G., & Joshi, S. (2011b). Optimal data-driven sparse parameterization of diffeomorphisms for population analysis. In G. Székely & H. Hahn (Eds.), LNCS: Vol. 6801. Information processing in medical imaging (IPMI) (pp. 123–134).
Glasbey, C. A., & Mardia, K. V. (2001). A penalised likelihood approach to image warping. Journal of the Royal Statistical Society. Series B. Statistical Methodology, 63, 465–492.
Glaunès, J., Qiu, A., Miller, M., & Younes, L. (2008). Large deformation diffeomorphic metric curve mapping. International Journal of Computer Vision, 80(3), 317–336. doi:10.1007/s11263-008-0141-9.
Grenander, U. (1994). General pattern theory: a mathematical theory of regular structures. London: Oxford University Press.
Grenander, U., & Miller, M. I. (1998). Computational anatomy: an emerging discipline. Quarterly of Applied Mathematics, LVI(4), 617–694.
Grenander, U., Srivastava, A., & Saini, S. (2007). A pattern-theoretic characterization of biological growth. IEEE Transactions on Medical Imaging, 26(5), 648–659.
Hansen, M. S., Larsen, R., Glocker, B., & Navab, N. (2008). Adaptive parametrization of multivariate b-splines for image registration. In Computer vision and pattern recognition (pp. 1–8). New York: IEEE Press.
Hastie, T., Tibshirani, R., & Friedman, J. (2009). The elements of statistical learning: data mining, inference, and prediction (2nd ed.). Berlin: Springer.
Joshi, S., & Miller, M. (2000). Landmark matching via large deformation diffeomorphisms. IEEE Transactions on Image Processing, 9(8), 1357–1370.
Lei, W., Beg, F., Ratnanather, T., Ceritoglu, C., Younes, L., Morris, J., Csernansky, J., & Miller, M. (2007). Large deformation diffeomorphism and momentum based hippocampal shape discrimination in dementia of the Alzheimer type. IEEE Transactions on Medical Imaging, 26, 462–470.
Lorenzen, P., Davis, B., & Joshi, S. C. (2005). Unbiased atlas formation via large deformations metric mapping. In Lecture notes in computer science: Vol. 3750. Medical image computing and computer-assisted intervention—MICCAI (pp. 411–418). Berlin: Springer.
Marsland, S., & McLachlan, R. I. (2007). A Hamiltonian particle method for diffeomorphic image registration. In LNCS: Vol. 4548. Proceedings of information processing in medical imaging (IPMI) (pp. 396–407). Berlin: Springer.
Meyer, Y. (2001). University lecture series: Vol. 22. Oscillating patterns in image processing and nonlinear evolution equations. Providence: Am. Math. Soc. The fifteenth Dean Jacqueline B. Lewis memorial lectures.
Miller, I. M., Trouvé, A., & Younes, L. (2002). On the metrics and Euler-Lagrange equations of computational anatomy. Annual Review of Biomedical Engineering, 4, 375–405.
Miller, M., & Younes, L. (2001). Group actions, homeomorphisms, and matching: A general framework. International Journal of Computer Vision, 41, 61–84.
Miller, M., Trouvé, A., & Younes, L. (2006). Geodesic shooting for computational anatomy. Journal of Mathematical Imaging and Vision, 24(2), 209–228.
Nesterov, Y. E. (1983). A method of solving a convex programming problem with convergence rate o(1/k 2). Soviet Mathematics. Doklady. 27(2). Translation by A. Rosa.
Pennec, X., Fillard, P., & Ayache, N. (2006). A Riemannian framework for tensor computing. International Journal of Computer Vision, 66(1), 41–66.
Risser, L., Vialard, F. X., Wolz, R., Murgasova, M., Holm, D. D., & Rueckert, D. (2011). Simultaneous multi-scale registration using large deformation diffeomorphic metric mapping. IEEE Transactions on Medical Imaging, 30, 1746–1759.
Rueckert, D., Aljabar, P., Heckemann, R. A., Hajnal, J., & Hammers, A. (2006). Diffeomorphic registration using B-splines. In Proc. MICCAI (pp. 702–709).
Singh, N., Fletcher, P., Preston, J., Ha, L., King, R., Marron, J., Wiener, M., & Joshi, S. (2010). Multivariate statistical analysis of deformation momenta relating anatomical shape to neuropsychological measures. In T. Jiang, N. Navab, J. Pluim, & M. Viergever (Eds.), Lecture notes in computer science: Vol. 6363. Proc. MICCAI’10 (pp. 529–537). Berlin: Springer.
Sommer, S., Nielsen, M., Darkner, S., & Pennec, X. (2012a). Higher order kernels and locally affine lddmm registration. arXiv:1112.3166v1.
Sommer, S., Nielsen, M., & Pennec, X. (2012b). Sparsity and scale: compact representations of deformation for diffeomorphic registration. In IEEE workshop on mathematical methods in biomedical image analysis (MMBIA 2012), Breckenridge, Colorado, USA. http://hal.inria.fr/hal-00641357/en/.
Trouvé, A. (1998). Diffeomorphisms groups and pattern matching in image analysis. International Journal of Computer Vision, 28(3), 213–221.
Trouvé, A., & Younes, L. (2005). Metamorphoses through lie group action. Foundations of Computational Mathematics, 5(2), 173–198.
Vaillant, M., & Glaunès, J. (2005). Surface matching via currents. In Lecture notes in computer science: Vol. 3565. Proceedings of information processing in medical imaging (pp. 381–392). Berlin: Springer.
Vaillant, M., Miller, M., Younes, L., & Trouvé, A. (2004). Statistics on diffeomorphisms via tangent space representations. NeuroImage, 23, 161–169.
Vercauteren, T., Pennec, X., Perchant, A., & Ayache, N. (2009). Diffeomorphic demons: efficient non-parametric image registration. NeuroImage, 45(1, Supp. 1), S61–S72. doi:10.1016/j.neuroimage.2008.10.040.
Yu, G., Sapiro, G., & Mallat, S. (2010). Image modeling and enhancement via structured sparse model selection. In Proceedings of the international conference on image processing (ICIP) (pp. 1641–1644). New York: IEEE Press.
Zou, H., & Hastie, T. (2005). Regularization and variable selection via the elastic net. Journal of the Royal Statistical Society. Series B. Statistical Methodology, 67(2), 301–320.
Acknowledgements
We would like to thank Timothy O’Keefe and Paul Sanders for their kind proofreading of the manuscript. This work has been supported by ANR grant IRMGroup and NIH grants NIBIB (5R01EB007688) and NCRR (2P41 RR0112553-12).
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix A: Proof of Proposition 1
Let δ S 0 be a small perturbation of the deformation parameters. This perturbation induces a perturbation of the system of particles δ S(t), which induces a perturbation of the position of the pixels mapped back by the inverse deformation δ y(0), which in turn induces a perturbation of the criterion δE:
According to (21), the perturbations of the state of the system of particles δ S(t) and the pixel positions δ y(t) satisfy the linearized ODEs:
The first ODE is linear. Its solution is given by:
The second ODE is linear with source term. Its solution is given by:
Plugging (41) into (42) and then into (40) leads to:
where
and
Let us denote η(s)=−V s0 t∇ y(0) A, g(s)=∂ 2 G(s)t η(s) and \(\xi(t) = \int_{t}^{1} {R_{0s}}^{t} g(s)ds\), so that the gradient (43) can be re-written as:
Now, we need to make explicit the computation of the auxiliary variables η(t) and ξ(t). By definition of V s0, we have V 00=Id and dV s0/ds=V s0 ∂ 1 G(s), which implies that η(0)=−∇ y(0) A and \(\dot{\eta}(t) = -{\partial_{1} G(t)}^{t}\eta(t)\).
For ξ(t), we notice that \(R_{ts} = \text{Id} - \int_{t}^{s} \frac {dR_{us}}{du} du \) \(= \text{Id} + \int_{t}^{s} R_{us}d_{\mathbf {S}(u)}F(u) du\). Therefore, using Fubini’s theorem, we get:
This last equation is nothing but the integral form of the ODE given in (24).
Appendix B: Lemma: Soft Thresholding in ℝN
Lemma 1
Soft-thresholding in ℝN
Let X and X 0≠0 two vectors in ℝN and F the criterion:
Then F achieves its minimum for
where S β (x)=max(x−β,0)−min(x+β,0).
Proof
If X≠0, F is differentiable and \(\nabla_{X} F = \frac{X}{\Vert X\Vert }+(X - X_{0})/\beta\). This shows that if the minimum of F is reached for X≠0 then X is parallel to X 0 and we can write X=λX 0/∥X 0∥ where λ satisfies:
If λ>0, then the minimum is reached for λ +=∥X 0∥−β, which occurs only if ∥X 0∥>β. If λ<0, then the minimum is reached for λ −=∥X 0∥+β, which occurs only if ∥X 0∥<−β. In the other cases, namely ∥X 0∥∈[−β,β], X=0.
Since F(λ + X 0/∥X 0∥)−F(0)<0, then the minimum of F is reached for X=(∥X 0∥−β)X 0/∥X 0∥ if ∥X 0∥>β. Since F(λ − X 0/∥X 0∥)−F(0)>0, then the minimum of F is reached for X=(∥X 0∥+β)X 0/∥X 0∥ if ∥X 0∥<−β. If ∥X 0∥∈[−β,β], the minimum of F is reached for X=0. These three cases are combined in a single expression using the soft-thresholding function S β . □
Rights and permissions
About this article
Cite this article
Durrleman, S., Allassonnière, S. & Joshi, S. Sparse Adaptive Parameterization of Variability in Image Ensembles. Int J Comput Vis 101, 161–183 (2013). https://doi.org/10.1007/s11263-012-0556-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11263-012-0556-1