Robust Similarity Metrics for the Registration of 3D Multimodal Medical Images. Mesures de Similarité Robustes pour le Recalage D'images Médicales Volumiques Multimodales

Robust Similarity Metrics for the Registration of 3D Multimodal Medical Images

Mesures de Similarité Robustes pour le Recalage D'images Médicales Volumiques Multimodales

Christophoros Nikou Fabrice Heitz  Jean-Paul Armspach  Izzie-Jacques Namer 

Laboratoire des Sciences de l'Image, de l'Informatique et de la Télédétection UPRESA CNRS 7005 / Université Strasbourg 14, Bd . Sébastien Brant, 67400 Illkirch, France

institut de Physique Biologique UPRESA CNRS 7004 / Université Strasbourg 1 Faculté de Médecine 4 rue Kirschleger, 67085 Strasbourg, France

Page: 
255-272
|
Received: 
2 April 1998
| |
Accepted: 
N/A
| | Citation

OPEN ACCESS

Abstract: 

In this paper, we develop data driven registration algorithms, relying on pixel similarity metrics, that enable an accurate rigid registration of dissimilar single ormultimodal 2D/3D medical images. Gross dissimilarities are handled by considering similarity measures related to robust M-estimators. Fast stochastic multigrid optimization algorithms are used to minimize these similarity metrics. The proposed robust similarity metrics are compared to the most popular standard similarity metrics on real MRI/MRI and MRI/SPECT image pairs showing gross dissimilarities. A blinded evaluation of the algorithm was performed, using as gold standard a prospective, marker-based registration method, by participating in a registration evaluation project (Vanderbilt University) . Our robust similarity measures compare favourably with all standard (non robust) techniques. 

Résumé 

Le recalage non supervisé d'images médicales volumiques reste un problème difficile en raison de l'importante variabilité et des grandes différences d'information pouvant apparaître dans des séquences d'images de même modalité ou dans des couples d'imagesmultimodales.Nous présentons dans cet article des méthodes robustes de recalage rigide d'images 2D et 3D monomodales etmultimodales, reposant sur la minimisation de mesures de similarité inter-images. Les méthodes proposées s'appuient sur la théorie de l'estimation robuste et mettent en oeuvredes M-estimateurs associés à des techniques d'optimisation stochastique multigrilles rapides. Ces estimateurs robustes sont évalués à travers le recalage d'images médicales volumiques monomodales(IRM/IRM) etmultimodales (IRM/TEMP). Ils sont comparés aux autres fonctions de similarité classiques, proposées dans la littérature. Les méthodes de recalage robustes ont, en particulier, été validées dans le cadre d'un protocole comparatif mis en place par l'Université de Vanderbilt. Elles sont actuellement utilisées en routine clinique et conduisent, tant pour les images de même modalité que pour les images multimodalesà une précision sous-voxel, comparable aux meilleures méthodes actuelles. Elles permettent de plus de recaler des couples d'images sur lesquels les méthodes classiques échouent. 

Keywords: 

Single and multimodal image registration, dissimilar image registration, similarity metrics, robust estimation, stochastic optimization, registration accuracy, registration evaluation .

Mots clés 

Recalage d'images monomodales et multimodales, mesures de similarité, estimation robuste, optimisation stochastique, évaluation de la précision du recalage .

1. Introduction
2. Fonctions de Similarité Classiques
3. Recalage par Fonctions de Similarité Robustes
4. Résultats Expérimentaux
5. Discussion et Conclusion
6. Remerciements
  References

[1] M.E Alexander and R. L. Somorjai. The registration of MR images using multiscale robust methods. Magnetic Resonance Imaging, 14(5) :453-468, 1996. 

[2] N. M. Alpert, D. Berdichevsky, Z. Levin, E. D. Morris, and A. J. Fischman. Improved methods for image registration.Neuroimage,3 :10-17, 1996. 

[3] B. A. Ardekani, M. Braun, B. F. Hutton, I. Kanno, and H. lida. A fully automatic multimodality image registration algorithm. Journal ofComputer Assisted Tomography, 19(4) :615-623, 1995. 

[4] M. J. Black and A. Rangarajan. On the unification of line processes, outliers rejection and robust statistics in early vision. International Journal ofComputer Vision, 19(1) :57-91, 1996. 

[5] L. G. Brown. A survey of image registration techniques. ACM Computing Surveys, 24(4) :325-376, 1992. 

[6] A. Collignon, F. Maes, D. Delaere, D. Vandermeulen, P. Suettens, and G. Marchal. Automated multimodality image registration using information theory. InProceedingsofthe International Conference on Information Processing in Medical Imaging,pages 263-274, 1995 .

[7] S. Geman and D. Geman. Stochastic relaxation, Gibbs distribution and the bayesian restoration of images. IEEE Transactions on Pattern Analysis and Machine Intelligence, 24(6) :721-741, 1984.

[8] J. Hajnal, N. Saeed, E. J. Soar, A. Oatridge, 1. R. Young, and G. M. Blyder. A registration and interpolation procedure for subvoxel matching of serially acquired MR images.JournalofComputer Assisted Tomography, 19(2):289296, 1995.

[9] F. Heitz, H. Maître, and C. de Couessin. Event detection in multisource imaging : application to fine arts painting analysis . IEEE Transactions on  Acoustic, Speech and Signal Processing, 38(4) :695-704, 1990.

[10] F. Heitz, P. Perez, and P. Bouthemy. Multiscale minimization of global energy functions in some visual recovery problems. Computer Vision, Graphics and Image Processing : Image Understanding,59(1) :125-134, 1994. 

[11] P. F. Hemler, T. S. Sumanaweera, P. van den Elsen, S. Napel, and J. R. Adler. A versatile system for multimodality image fusion.JournalofImage Guided Surgery, 1 :35-45, 1995.

[12] M. Herbin, A. Venot, J. Y Devaux, E. Walter, F. Lebruchec, L. Dubertet, and J. C. Roucayrol. Automated registration of dissimilar images : application to medical imagery. Computer Vision, Graphics and Image Processing, 47 :7788, 1989. 

[13] A. Hill, D. Hawkes, N. Harrison, and C. Ruff. A strategy for automated multimodality registration incorporating anatomical knowledge and imager characteristics. InProceedingsofthe International Conference on Information Processing in Medical Imaging, pages 182-196. Springer-Verlag, Berlin, 1993. 

[14] H. Jiang,A. Robb, and K. S. Holton. A new approach to 3D registration of multimodality medical images by surface matching. InSPIEProceedings of Visualization in Biomedical Imaging, volume 1808, pages 196-213, 1992. 

[15] D. Lemoine, D. Liegeard, E. Lussot, and C. Barillot. Multimodal registration system for the fusion of MRI, CT, MEG and 3D or stereotactic angiographie data. InProceedings ofSPIEInternational Conference on Medical Imaging, volume 2164, pages 46-56, 1994. 

[16] F. Maes, A. Collignon, D. Vandermeulen, G. Marchai, and P. Suetens. Multimodality image registration by maximization of mutual information. IEEE Transactions on Medical Imaging, 16(2) :187-198, 1997. 

[17] G. Q. Maguire Jr, M. E. Noz, H. Rusinek, J. Jaeger, E. L. Kramer, J. J. Sanger, and G. Smith. Graphics applied to medical image registration.IEEE Transactions on Computer Graphics and Applications, 11 :20-29, 1991. 

[18] J. B. A. Maintz, P. A. Van den Eisen, and M. Viergever. Evaluation of ridge seeking operators for multimodality medical image matching. IEEE Transactions on Pattern Analysis and Machine Intelligence, 18(4) :353-365, 1996. 

[19] J. B. A. Maintz and M. A. Viergever. A survey of medical image registration techniques.Medical Image Analysis, 2(1) :1-36, 1998. 

[20] G. Malandain, S. Fernândez-Vidal, and J. M. Rocchisani. Improving registration of 3D medical images using a mechanical-based method. In Lecture Notes in Computer Science. Proceedings ofthe 3rdEuropean Conference on Computer Vision (ECCV'94), volume 801, pages 131-136, 1994. 

[21] P.Meer,D. Mintz, A. Rosenfeld,and D. Y. Kim. Robust regression methods for computer vision : a review. International Journal of Computer Vision, 6(l):59-70, 1990. 

[22] O. Musse, J. P. Armspach, I. J. Namer, F. Heitz, F. Hennel, and D. Grucker. Data-driven curvilinear reconstruction of 3D MRI :application to cryptogenic extratemporal epilepsy. Magnetic Resonance Imaging, 16(10) :1227-1235, 1998. 

[23] C. Nikou, F. Heitz, and J. P. Armspach. Robust registration of dissimilar single and multimodalimages. InLecture Notes in Computer Science. Proceedings ofthe 5th European Conference on Computer Vision (ECCV'98), volume 2, pages 51-65,Freiburg, Germany, 2-6 June 1998. 

[24] C. A. Pelizzari, G. T. Y Chen, D. R. Spelbring, R. R. Weischelbaum, and C. T. Chen. Accurate three-dimensional registration of CT, PET and/or MR images of the brain.JournalofComputerAssisted Tomography, 13(l):20-26, 1989.

[25] U. Pietrzyk, K. Herholz, A. Jacobs, R. Mielke, I. Slansky, M. Wurker, and W. D. Heiss. An interactive technique for three-dimensional image registration :validation for PET, SPECT, MRI and CT brain images.Journal ofComputer Assisted Tomography, 35 :2011-2018, 1994.

[26] L. Rumbach, E. Racadot, J. P. Armspach, I. J. Namer, J. F. Bonneville, J. Wijdenes, C. Marescaux, P. Herve, and J. Chambron. Biological assessment and MRI monitoring of the therapeutic efficacy of a monoclonal anti-T CD4 antibody in multiple sclerosis patients. Multiple Sclerosis, pages 207-212, 1996.

[27] H. Sawhney, S. Ayer, and M. Gorkani. Model-based 2D-3D dominant motion estimation for mosaicing and video representation . In Proceedings ofthe International Conference on ComputerVision, pages 583-590, Cambridge, USA, June 1995. 

[28] C. Stewart. Bias in robust estimation caused by discontinuities and multiple structures.IEEE Transactions on Pattern Analysis and Machine Intelligence, 19(8) :818-833, 1997. 

[29] C. Studholme, D. Hill, and D. Hawkes. Automated three-dimensional registration of magnetic resonance and positron emission tomography brain images by multiresolution optimization of voxel similarity measures.Medical Physics, 24(1) :25-35, 1997. 

[30] P. Van den Elsen, E. J. D. Paul, and M. A. Viergever. Medical image matching - a review with classification. IEEE Engineering in Medicine and Biology, 12(1) :26-39, 1993. 

[31] P. Van den Elsen, E. J. D.Pol,T. S. Sumanaweera, P. F. Hemler, S. Napel, and J. R. Adler. Grey value correlation techniques used for automatic matching of CT and MR brain and spine images. In SPIEInternational Conference on Visualization in Biomedical Computing, volume 2359, pages 227-237, 1994.

[32] A. Venot, J. F. Lebruchec, and J. C. Roucayrol. A new class of similarity measures for robust image registration. Computer Vision, Graphics and Image Processing, 28 :176-184, 1984. 

[33] W. Wells 111, P. Viola, H. Atsumi, S. Nakajima, and R. Kikinis. Multimodal volume registration by maximization of mutual information .Medical Image Analysis, 1(1) :33-51, 1996. 

[34] J. West, M. Fitzpatrick, M. Wang, B. Dawan, C. Maurer Jr., R. Kessler, R. Maciunas, C. Barillot, D. Lemoine, A. Collignon, F. Maes, P. Suetens, D. Vandermeulen, P. Van den Elsen, S. Napel, T. Sumanaweera, B. Harkness, P. Hemler, D. Hill, D. Hawkes, C. Studholme, A. Mainz, M. Viergever, G. Malandain, X. Pennec, M. Noz, G. Maguire, Jr. M. Pollack, C. Pelizzari, R. Robb, D. Hanson, and R. Woods. Comparison and evaluation of retrospective intermodality brain image registration techniques . Journal of Computer Assisted Tomography, 21(4) :554-566, 1997. 

[35] R. P. Woods, J. C. Mazziota, and S. R. Cherry. MRI-PET registration with automated algorithm.Journal of ComputerAssisted Tomography, 17(4):536546, 1993.