External Email - Use Caution
Dear FreeSurfer team,
We are using the Freesurfer longitudinal pipeline, and we had questions about accounting for different scanners and acquisition parameters.
We are looking at 3 different timepoints with 2 different scanners (scanner 1 was used for time points 1 and 2, scanner 2 was used for time point 3).
Timepoint 1 (scanner1, 512 x 512 x 244 or 512 x 512 x 180)
Timepoint 2 (scanner1, 512 x 512 x 180)
Timepoint 3 (scanner2, 448 x 448 x 192 or 448 x 448 x 208)
Q1: Does creating the subject template based on median and intensity scaling help address the issue of different scanners and acquisition parameters when conducting longitudinal analyses? If it is not sufficient to address the scanner/acquisition issue, do you have any recommendations as to next steps to account for this? We have seen others use LongCombat harmonization or linear mixed effects models to address this issue, is that something you would consider as well?
Q2: In a similar vein, when we received the warning below in the "base" step of longitudinal analysis, we were wondering how much it was affecting our results. Would it be helpful to resample the images? Do you have suggestions on other avenues we could consider to address this warning and have robust and reliable results?
************************************************************ ******************* WARNING: Image parameters differ across time, maybe due to aquisition changes? Consistent changes in, e.g., resolution can potentially bias a longitudinal study! You can check image parameters by running mri_info on each input image. Will continue in 10 seconds ... ************************************************************ *******************
Thank you
Sincerely,
Yoonho