HCs always have 0 medicationload so it is not possible to measure a load slope for them. Is it the case that a BD with load=0 is the same as HC? If so, then just have one class. If not, do you expect the age, edu, and tiv slopes to differ between the groups? If not, then use DOSS (default was DODS).

On 9/11/2022 3:28 AM, Cao Yuan wrote:

External Email - Use Caution

Dear experts,

I would like to run mri_glmfit in a group analysis. However, I met the following error when I added the medicationload as variable in my fsgd file.

Since I have 305 subjects and only 75 of these have medication load, it seems to this variable includes too many zero. Just I guess.

The error:

ERROR: matrix is ill-conditioned or badly scaled, condno = 1e+08\

\

\

--------------------------------\

-------- ERROR: READ THIS -----------------\

--vvvvvvvvvvvvvvvvvvvvvvvvvvvvvv-----\

Possible problem with experimental design:\

Check for duplicate entries and/or lack of range of\

continuous variables within a class.\

If you seek help with this problem, make sure to send:\

1. Your command line:\

mri_glmfit --y lh.volume.BDHC3Study.10.mgh --fsgd FSGD/BDHC3Study.fsgd --C Contrasts/HC3-BD3.mtx --C Contrasts/BD3-HC3.mtx --surf fsaverage lh --cortex --glmdir lh.volume.BDHC3Study.10.glmdir \

2. The terminal output of this program (ie, everything it prints to the screen)\

3. The FSGD file (FSGD/BDHC3Study.fsgd)\

4. The design matrix lh.volume.BDHC3Study.10.glmdir/Xg.dat\

Attempting to diagnose further \

SumSq: Min=6.368664 (col 8), Max=16.261612 (col 7)\

... could not determine the cause of the problem\

--^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^----\

}

I attached my terminal output, fsgd file, and design matrix

Thanks

Yuan
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
MailScanner has detected a possible fraud attempt from "secure-web.cisco.com" claiming to be https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer