Principles and constraints of nonrigid registration

Abstract

Medical image registration refers to the process of determining the correspondence of anatomical or functional features between two or more images collected at different time points (serial image registration) or acquired as part of different sequences or imaging modalities (multi-modal image registration). Registration and fusion applications are proved to be of great help to the medical experts due to their contribution to the diagnostic procedure. The aim of this paper is to provide a solid theoretical review of the principles and constraints around non-rigid registration methodologies that are implemented in the medical field. The core components of these methodologies such as the similarity measures, the transformation models and the optimisation, interpolation and validation techniques are introduced and described.