Evaluation of a Symmetric Diffeomorphic Registration Algorithm for Analyzing Neurodegenerative Disease
Grossman M, Gee J, Avants B
University of Pennsylvania
Our recently developed algorithm, symmetric diffeomorphic image registration, allows rigorous quantification of neurodegenerative disease. We show the benefits of our method in comparison to a standard elastic deformation model (Demons) for discovering statistically significant differences in populations through normalization. A comparison to gold standard results confirms that the new method outperforms Demons in terms of reproducing manual labelings and in terms of giving significant results.