Deformable Model-Based Image Registration
In this chapter we introduce the concept of deformable image registration and point out that interpolation effects, non-rigid body registration, and joint segmentation and registration frameworks are among the challenges that remain for this area. To address the interpolation effects challenge, we choose the Partial Volume Interpolator (PV) used in multimodality registration as an example, and quantitatively analyze the generation mechanism of the interpolation artifacts. We conclude that the combination of linear interpolation kernel and translation-only motion leads to generation of the artifact pattern. As a remedy we propose to use nonuniform interpolation functions in estimating the joint histogram. The cubic B-spline and Gaussian interpolators are compared, and we demonstrate improvements via experiments on misalignments between CT/MR brain scans. A segmentation-guided non-rigid registration framework is proposed to address the second and third challenges. Our approach integrates the available prior shape information as an extra force to lead to a noise-tolerant registration procedure, and it differs from other methods in that we use a unified segmentation + registration energy minimization formulation, and the optimization is carried out under the level-set framework. We show the improvement accomplished with our model by comparing the results with that of the Demons algorithm. To explore other similarity metrics under the same framework to handle more complicated inputs will be the focus of our future work.
KeywordsMutual Information Image Registration Registration Result Joint Entropy Deformable Image Registration
Unable to display preview. Download preview PDF.
- 2.Haacke E, Liang Z-P. 2000. Challenges of imaging structure and function with MRI. IEEE Eng Med Biology Magn 19(5):55-62.Google Scholar
- 5.Collignon A, Maes F, Delaere D, Vandermeulen D, Suetens P, Marchal G. 1995. Automated mul-timodality image registration using information theory. In Proceedings of the 14th international conference on information processing in medical imaging (IPMI), pp. 263-274. Ed YJC Bizais. Washington, DC: IEEE Computer Society.Google Scholar
- 9.Ji X, Pan H, Liang ZP. 1999. A region-based mutual information method for image registration. In Proceedings of the 7th international meeting of the international society for magnetic resonance in medicine, Vol. 3, p. 2193. Washington, DC: IEEE Computer Society.Google Scholar
- 12.Taso J. 2003. Interpolation artifacts in multimodality image registration based on maximization of mutual information. IEEE Trans Med Imaging 22(7):845-864.Google Scholar
- 14.Liu J, Wang Y, Liu J. 2006. A unified framework for segmentation-assisted image registration. In Proceedings of the 7th Asian conference on computer vision. Lecture notes in computer science, Vol. 3852, pp. 405-414. Berlin: Springer.Google Scholar
- 15.Liu J, Wei M, Liu J. 2004. Artifact reduction in mutual-information-based CT-MR image registra-tion. In Proceedings of the SPIE, Vol. 5370, pp. 1176-1186. Medical imaging 2004: physiology, function, and structure from medical images. Ed AA Amini, A Manduca. Bellingham, WA: In-ternational Society for Optical Engineering.Google Scholar
- 18.Unal G, Slabaugh G,Yezzi A, Tyan J. 2004. Joint segmentation and non-rigid registration without shape priors. Siemens Technical Report SCR-04-TR-7495.Google Scholar
- 19.Unal G, Slabaugh G. 2005. Coupled PDEs for non-rigid registration and segmentation. In Pro-ceedings of the IEEE computer society conference on computer vision and pattern recognition (CVPR’05), Vol. 1, pp. 168-175. Washington, DC: IEEE Computer Society.Google Scholar
- 20.Vemuri B, Chen Y, Wang Z. 2002. Registration-assisted image smoothing and segmentation. Proceedings of the 7th European Conference on Computer Vision, Part IV. Lecture notes in computer science, Vol. 2353, pp. 546-559.Google Scholar
- 25.Simulated brain database. Available at http://www.bic.mni.mcgill.ca/brainweb/.
- 28.Leventon M, Grimson WEL. 1999. Multi-modal volume registration using joint intensity dis-tributions. In Lecture notes in computer science, Vol. 1496, pp. 1057-1066. Berlin: Springer. Available at http://www.ai.mit.edu/people/leventon/Research/9810-MICCAI-Reg/.