Hi Doug,
Thank you very much for your reply and my apologies for this belated response. The two approaches you mentioned were the only ones we could think that might work for a DV like cortical thickness.
That being said, we were originally thinking of a repeated measures anova. However, (and I may be mistaken here) I believe that such an approach would need to be implemented in the Freesurfer longitudinal stream. However, I recently saw a thread in the archives that suggested this approach is not recommended if the T1 images are from matched subjects (as opposed to multiple scans from the same subject over time). Is that correct? As an aside, as we're hoping to separately look at matched subjects in three groups a bit later using a 1:1:1 match. If we should avoid the longitudinal stream, is there any other way to implement an RM anova in Freesurfer?
Otherwise, I think we're hoping to move forward by averaging the 2 controls together with make_average_subject and then implement a paired analysis -- which was probably the more sound choice to begin with.
Thank you again!
Kind regards,
Kevin
Douglas N. Grevehttps://www.mail-archive.com/search?l=freesurfer@nmr.mgh.harvard.edu&q=from:%22Douglas+N.+Greve%22 Mon, 16 Jul 2018 10:49:28 -0700https://www.mail-archive.com/search?l=freesurfer@nmr.mgh.harvard.edu&q=date:20180716
There is nothing to plug and play. I'm not even sure about the theory, but I'd guess you'd use something like a repeated measures anova or simply average the two controls together and do a simple paired test. Do you know what the theory would be behind your statitical test?
On 07/09/2018 01:27 PM, Dowling, Kevin Francis wrote:
Hello Freesurfer Experts,
I'm writing with a quick question about matched analysis.
We have a cohort of MRI data collected such that there are two controls for each case, and these 2 controls are each matched to a specific case using a propensity score algorithm (~400 subjects total). For our analysis, we would like to try to implement a matched (or paired difference) analysis of cortical thickness (across the whole surface, not .aparcs) in Freesurfer. However, based all the documentation I have been able to find it seems that FS permits only paired difference analyses using matched pairs. Is this correct? Might there any way to implement such an analysis in Freesurfer using matched trios instead?
We are open to using any Freesurfer version for this analysis, including dev, though the scans are currently reconned with FS 5.3.
Thank you so much for your consideration!
With kind regards,
Kevin
*Kevin F. Dowling * Clinical Research Coordinator Brain Genomics Laboratory Division of Psychiatric Neuroimaging Martinos Center for Biomedical Imaging Massachusetts General Hospital 149 13th Street Charlestown, MA, 02129 (p) 617.643.3215 Pronouns: He / Him / His
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Kevin F. Dowling Clinical Research Coordinator Brain Genomics Laboratory Division of Psychiatric Neuroimaging Martinos Center for Biomedical Imaging Massachusetts General Hospital 149 13th Street Charlestown, MA, 02129 (p) 617.643.3215 Pronouns: He / Him / His