Nonrigid Registration: Concepts, Algorithms, and Applications - Medical Image Registration - page 298

could appear to match perfectly the source image to the target image, such
that all features overlap, but the transformation calculated may be com-
pletely wrong. For example, in intrasubject registration, a transformation
might stretch some parts of the image and compress others to make struc-
tures appear to line up even if the underlying tissue is incompressible. In
intersubject registration, an algorithm might warp structures in one image
so they appear to line up with corresponding structures in a second image
of a different subject, but the calculated transformation might be inappropri-
ate for comparing functional regions identified from the two subjects. At its
current stage of maturity, nonrigid registration algorithms should be used
with care, especially where it is desirable to use the calculated transforma-
tion for secondary purposes such as understanding tissue deformation or
studying variability between subjects.

Acknowledgments

Daniel Rueckert was a research fellow at the Computational Imaging Science
Group, Department of Radiological Sciences at King's College London when
most of this work was performed. He was supported by EPSRC project grant
GR

L08519.

References

1. H. Lester and S.R. Arridge. A survey of hierarchical non-linear medical image
registration. Pattern Recogn ., 32(1):129-149, 1999.
2. D.L.G. Hill, C.R. Maurer, Jr., C. Studholme, J.M. Fitzpatrick, and D.J. Hawkes.
Correcting scaling errors in tomographic images using a nine degree of freedom
registration algorithm. J. Comp. Assist. Tomogr ., 22(2):317-323, 1998.
3. C. Studholme, J.A. Little, G.P. Penney, D.L.G. Hill, and D.J. Hawkes. Automated
multimodality registration using the full affine transformation: application to
MR and CT guided skull base surgery. In Proc. 4th International Conference Visu-
alization in Biomedical Computing (VBC'96) , pages 601-606, 1996.
4. D.L. Collins, P. Neelin, T.M. Peters, and A.C. Evans. Automatic 3D intersubject
registration of MR volumetric data in standardized Talairach space. J. Comp.
Assist. Tomogr ., 18(2):192-205, 1994.
5. R.P. Woods, S.T. Grafton, J.D.G. Watson, N.L. Sicotte, and J.C. Mazziotta. Auto-
mated image registration: II. Intersubject validation of linear and nonlinear mod-
els. J. Comp. Assist. Tomogr ., 22(1):153-165, 1998.
6. R. Szeliski and S. Lavallée. Matching 3-D anatomical surfaces with non-rigid
deformations using octree-splines. In IEEE Workshop on Biomedical Image Analysis ,
pages 144-153, 1994.
7. Y. Amit, U. Grenander, and M. Piccioni. Structural image restoration through
deformable templates. J. Am. Statis. Ass ., 86(414):376-387, 1991.