On 2/15/2023 5:35 PM, Antonio #30 wrote:
External Email - Use Caution
Hello again,
I would like to ask one last thing. How can the class averages computed from csdbase.y.ocn.dat not match the gamma.mgh values in the cluster?
e.g. How can gamma show +29 in a vertex that is part of the significantly positive cluster in an A-B contrast, while the average of all values taken from csdbase.y.ocn.dat referring to class A in that same cluster is lower than that of class B (average of values of input y for class A = 20.367; average for class B = 20.401)?
The cluster is defined after thresholding. So there might be some vertices that have a huge effect in one direction, but from the cluster perspective, they don't count anymore than a vertex that has a small but suprathreshold effect. The computed average will be affected though.
Are the values in csdbase.y.ocn.dat not corrected by the variables chosen as covariates?
The gamma values are corrected.
Thanks again!
Antonio Bonifanti
Il giorno mer 15 feb 2023 alle ore 18:36 Antonio #30 toni.bonif2511@gmail.com ha scritto:
Hi, 1. I was wondering why the --fsvol argument in -pctsurfcon takes rawavg.mgz by default, when most literature on grey/white contrast tends to compute it from nu.mgz. Is the intensity-corrected volume to be preferred in all cases or was there a reason for choosing rawavg as a default while writing the script?
I don't think it matters as the computation is very local (ie, wm a few mm from gm) so the bias field will be minimal.
2. I would also like to ask whether or not you believe the values of smoothing kernels seen in literature for grey-white contrast (which often are over 20mm) to be adequate for such a measure.
I don't know, I've never looked into it. 20mm seems like a lot.
3. As a final note, would you say it is appropriate, from a statistical standpoint, to compare the results of group analyses conducted on the two hemispheres with different covariates? (e.g. with mean left cortical thickness and mean right cortical thickness rather than mean whole cortical thickness)
It just leads to a different interpretation. If the interpretation is reasonable and you can explain it, then it is ok.
Thank you for your time! Antonio Bonifanti
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer