Biomedical Engineering Reference
In-Depth Information
13
Nonrigid Registration: Concepts, Algorithms,
and Applica
tions
Daniel Rueckert
CONTENTS
13.1 Introduction.............................................................................................. 281
13.2 Techniques ................................................................................................ 282
13.2.1 Registration Using Basis Functions ........................................ 283
13.2.2 Registration Using Splines....................................................... 284
13.2.2.1 Thin-Plate Splines.................................................... 285
13.2.2.2 B-Splines ................................................................... 286
13.2.3 Elastic Registration.................................................................... 287
13.2.4 Fluid Registration...................................................................... 288
13.2.5 Registration Using FEM and Mechanical Models ................ 288
13.2.6 Registration Using Optical Flow............................................. 289
13.2.7 Registration as an Optimization Problem ............................. 290
13.3 Applications ............................................................................................. 291
13.3.1 Intrasubject Registration .......................................................... 292
13.3.2 Intersubjection Registration..................................................... 294
13.3.3 Analysis of Motion and Deformation Using
Nonrigid Registration............................................................... 296
13.4 Conclusions .............................................................................................. 296
Acknowledgments .............................................................................................. 298
References ............................................................................................. 298
13.1
Introduction
The previous chapters in this topic have focused on rigid transformations for
image-to-image and image-to-physical space registration. In many applica-
tions a rigid transformation is sufficient to describe the spatial relationship
between two images. For example, brain images of the same subject can be
related by a rigid transformation since the motion of the brain is largely
