Thus, although the use of a concatenated transformation to parcellate the follow-up data does not eliminate subject-to-atlas registration errors, we ensure that the same errors apply to baseline and follow-up data consistently, which would not be the case if the follow-up data were registered to the atlas independently. Because we are primarily interested in measuring longitudinal effects, longitudinal consistency of registration errors should be beneficial. Finally, as a test of the robustness of the sequential approach we reversed the inter- and intra-subject chain and reformatted the SRI24 labels maps for the baseline images using the concatenation of baseline-to-followup and followup-to-atlas transformations.
