ABSTRACT
This chapter explores plenty of applications ranging from brain development study to early disease diagnosis. To meet the booming requirement for the unbiased group analysis, a relatively new family of registration methods, called as groupwise registration has come to the stage. As demonstrated in various applications, groupwise registration algorithms generally demonstrate superior performance over their pairwise counterparts, and thus becomes a hot topic in the field of medical image analysis. The majority of the current groupwise registration algorithms can be classified into three classes: first, pairwise registration derived groupwise registration; second, population center-guided groupwise registration; and third, hidden common space-based groupwise registration. The first class of methods directly applies pairwise registration to achieve the goal of groupwise registration. In the literature, several groupwise registration methods have been proposed to better estimate and utilize the population center. In general, hidden common space-based groupwise registration can be formulated as an optimization problem with the goal of minimizing variations within the image population.
