It's probably the case that it's badly scaled (eg, the age regressor is 10 or 100 times the other regressors). Try dividing the age by 10 or 100.

doug

Lusineh Gharapetian wrote:

Hello,

I'm trying to use the mri_glmfit function but keep running into the same error.

I am using the FSGD format and have set up my fsgd file according to the tutorial instructions:

" GroupDescriptorFile 1 Title mci_age_BNTchange Class Male Class Female Variables Age BNTchange Input Subject1 Male 70 2 Input Subject2 Female 75 -3 Input Subject3 Female 72 0 etc."

Our .mat file states:

" 0 0 0 1" indicating that "BNTchange" is our variable of interest.

We are attempting to run a DOSS analysis.

The line of code I type into the terminal is:

" mri_glmfit --glmdir mci_age_bntchange_lh --y dv-visit2-lh.mgz --fgd mci_age_bntchange.fsgd doss --C mci_age_bntchange.mat"

This results in the following error:

"ERROR: matrix is illconditioned or badly scaled, condno = 1e+08"

We have tried runninf this off of different versions of Freesurfer (R302 and R305) and cannot figure out how to fix this error.

Any help would be appreciated!

Thanks!

• Lusineh Gharapetian

---

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer