Comparing Methods: Registering an External Atlas to fsnative vs. Registering fsnative to an External Atlas

In many situations, I have noticed that most researchers register an external Atlas (not native to FreeSurfer) onto the subject’s original space (fsnative), rather than vice versa. What are the fundamental differences or specific considerations between these two registration methods in practical applications?

If we register an external template onto the subject’s fsnative space, will there be discrepancies in the number of vertices encompassed within different ROIs? Conversely, can this be circumvented if done the other way around? This is especially relevant when conducting group analyses. For example, if I aim to investigate the cortical thickness differences in specific atlas ROIs between patients with depression and healthy controls, should I:

  1. Register the Atlas onto each subject’s original fsnative space first, calculate the ROI’s cortical thickness, and then perform between-group comparisons? or
  2. Register the subject’s original fsnative space to the external atlas, then calculate the ROI’s cortical thickness, followed by between-group comparisons?

What are the distinctions between these two registration approaches?