Skip to main content
. Author manuscript; available in PMC: 2012 Feb 1.
Published in final edited form as: Neuroimage. 2010 Sep 17;54(3):2033–2044. doi: 10.1016/j.neuroimage.2010.09.025

Table 1.

Transformations and similarity metrics available in ANTs.

Category Transformation, ϕ Similarity Measures Brief Description
Linear Rigid† MSQ, CC, MI translation and rotation
Affine† MSQ, CC, MI rigid, scaling, and shear
Elastic Deformable CC, PR, MI, MSQ, JHCT, PSE Demons-like algorithm
DMFFD CC, PR, MI, MSQ, JHCT, PSE FFD variant
Diffeomorphic Exponential CC,PR, MI, MSQ, JHCT, PSE minimizes υ(x)
Greedy SyN† CC, PR, MI, MSQ, JHCT, PSE minimizes υ(x, t) locally in time
Geodesic SyN† CC, PR, MI, MSQ, JHCT, PSE minimizes υ(x, t) over all time

Similarity metric acronyms: MSQ = mean squared difference, CC = cross correlation, PR = CC after subtraction of local mean from the image, MI = mutual information, JHCT = Jensen-Havrda-Charvat-Tsallis divergence, PSE = point-set expectation.

ANTs also provides the inverse of those transformations denoted by the ‘†’ symbol. Only the MSQ, CC and MI metrics are available for both affine and deformable registration and are evaluated here with the Greedy SyN transformation model.