Elastic registration of brain cine-MRI sequences using MLSDO dynamic optimization algorithm

Abstract

In this chapter, we propose to use a dynamic optimization algorithm to assess the deformations of the wall of the third cerebral ventricle in the case of a brain cine-MR imaging. In this method, an elastic registration process is applied to a 2D+t cine-MRI sequence of a region of interest (i.e. lamina terminalis). This registration process consists in optimizing an objective function that can be considered as dynamic. Thus, a dynamic optimization algorithm based on multiple local searches, called MLSDO, is used to accomplish this task. The obtained results are compared to those of several well-known static optimization algorithms. This comparison shows the efficiency of MLSDO, and the relevance of using a dynamic optimization algorithm to solve this kind of problems. © Springer-Verlag Berlin Heidelberg 2013.