Skip to main content
. Author manuscript; available in PMC: 2014 Aug 19.
Published in final edited form as: Neuroimage. 2014 Mar 18;94:65–78. doi: 10.1016/j.neuroimage.2014.03.026

Table 1.

Methodological differences between analysis pipelines.

Pipeline name/differences 1: Erosion
kernel		 2: Registration
algorithm		 3: Registration targets 4: Transforming to standard
space		 5: Masking of voxels 6: Skeleton
projection
FSL TBSS 3 × 3 × 3 FNIRT Most-representative subject Affine registration (FLIRT) Voxels in all subjects Yes
ANTS-GW TBSS 3 × 3 × 1 ANTS SyN ANTS groupwise Template Nonlinear registration Voxels in ≥ 50% of subjects Yes
FSL VBA 3 × 3 × 3 FNIRT Most-representative subject Affine registration (FLIRT) Voxels in all subjects No
ANTS-GW VBA 3 × 3 × 1 ANTS SyN ANTS groupwise Template Nonlinear registration Voxels in ≥ 50% of subjects No