Hi Derek, We do not use the standard FS pipeline for the analysis of the age group of your interest. In general, we have seen it being successfully used on subjects down to 4.5yrs of age, but 25 mo is quite young to be analyzed by the adult template. I assume, without seeing your data, that your assumption is correct about the registration to the adult fsaverage significantly influencing your results. Wrt (3), yes, I believe a new pipeline with new templates would need to be created to be more suited to your dataset. Best, Lilla
On Thu, 9 Aug 2018, Derek Sayre Andrews wrote:
Dear Freesurfer Developers and Experts,
I have been exploring various vertex wise (mass univariate) effects using freesurfer measures of cortical thickness and surface area in a cohort of toddlers/children aged 25-70 months.
When looking at the effect of age on surface area using individuals’ measures mapped to fsaverage and smoothed at 15mm fwhm (area.pial. fwhm15.fsaverage.mgh) I observed a widespread negative effect of age (i.e. decreased surface area with age). Given the age group of the cohort this seems implausible. Plotting the sum of each individuals lh+rh.area.pialfsaverage.fwhm15.mgh against age shows the relationship:
cid:image001.png@01D42FCE.C197B360
However, plotting the lh+rh total surface area returned by mris_anatomical_stats reveals a much more plausible relationship between surface area and age:
cid:image002.png@01D42FCE.C197B360
My working hypothesis is that the significantly smaller brain sizes of our cohort compared to fsaverage, and the fact(?) that -qcache does not preserve global surface area, is causing an “over adjustment” in our cohorts smallest brains when mapped to fsaverage.
Thus, my questions are three fold
- Is the mapping of our significantly smaller brains to fsaverage a potential explanation for the results we are seeing in our mass univariate analyses of surface area?
- If this is the case, would this effect be limited to surface area or also extend to other freesurfer measures including cortical thickness, lGI, sulcal depth etc.
- How would you recommend approaching vertex wise analyses in a cohort of this age? Create a study specific average subject using make_average_subject to map onto?
Thank you for your insight and guidance!
Best, Derek
PS: All processing was performed using freesurfer v5.1
cid:image003.png@01D42FCE.C197B360
Derek Sayre Andrews, PhD
Postdoctoral Scholar
Department of Psychiatry and Behavioral Sciences
The M.I.N.D Institute
University of California Davis
Telephone: +1 916 703 0360 Email: dandrews@UCDavis.edu