External Email - Use Caution

Hello Freesurfer Users,

I am running selxavg3 on functional MRI data from 2 runs, with a block design (paradigm files attached) and select external regressors originating from fMRIPrep (attached). There are 5 non-fixation conditions in the design.

I am having errors come up during this analysis.

_____

Without external regressors, there are no runtime errors and one contrast looks okay (corresponding to -a 3 -c 2), however:

- 2 contrasts yield t-stats that are the exact opposites of each other, which was not expected (corresponding to -a 2 -c 4 and -a 3 -c 5)

- Another contrast yields erroneous results, with t-stats on the order of 10^-11 ( corresponding to -a 3 -a 5 -c 2 -c 4).

_____

With external regressors, only one contrast will run (again, -a 3 -c 2) - the rest give a "design is ill-conditioned error" (log attached).

I have attached the corresponding Xtmp.mat file (zipped).

_____

I am wondering if there are any strategies someone can recommend for figuring out why these runtime / processing errors occur. Could this be a problem with the task design itself? Any advice is appreciated!

Thank you very much,

_______________________

Eli Bulger

Research Assistant

Lab in Multisensory Neuroscience

Carnegie Mellon University