Image coregistration is generally defined as the alignment of two or more image volumes into the same geometric space. Coregistration can be applied to the same image modality— intramodality coregistration (e.g., magnetic resonance imaging (MRI) to MRI)—or across modalities — intermodality co-registration (e.g., positron emission tomography (PET) to MRI). Coregistration can also be applied to alignment of images from various subjects to a common space — intersubject coregistration (e.g., patient scan to an atlas of a mean normal or control data of combined images).

Numerous coregistration methods with various merits and limitations are employed for clinical purposes. The methods are classified as either rigid-body or nonrigid transformations. Rigid-body methods assume no change in shape or deformation in structure between images. This assumption is generally valid for most intrasubject brain coregistration applications, and this is why most clinical applications are emerging fastest in neuroimaging. Rigid-body techniques are the best-validated methods. They are also the simplest since only six degrees of freedom are involved — three translations along the x y, and z-axes, and three rotations around each of the respective axes. Nonrigid transformations, especially those based on nonlinear models for warping, are much more computationally complex and difficult to validate. Still, methods of image deformation or warping are expected to be increasingly used in clinical coregistration applications.

In this chapter, areas of present and some future clinical application of coregistration will be reviewed. The applications include only those in neuroimaging and cover in particular the examination of dementia, multiple sclerosis (MS), central nervous system (CNS) tumors, and epilepsy.

0 0

Post a comment