design is ill-conditioned - Freesurfer

31 Oct 2022


      External Email - Use Caution
Dear experts,
I’m running into an error when running selxavg3. It says the design is ill-conditioned. I’m a bit stuck, because the same pipeline works in most subjects, but not all. Also, if I exchange the functional scan (f.nii.gz) with a scan from a working session of the same subject (same number of volumes etc.) I don’t get any errors. I’m grateful for any advice that helps to track down the mistake I’m doing.
Here the content of the selxavg3.log file:
cd /home/hector/SENS_MAPPING_BLOCK/sessions
/opt/software/freesurfer/fsfast/bin/selxavg3-sessbc -s HEC_035_1 -analysis MPBDandefs_orig_lhrh_revd11.5_sm2.lh -no-preproc -fwhm -force
Found 42449/139130 (30.5) voxels in mask
Creating Design Matrix
... creation time =  0.004 sec
DoMCFit = 1
ntptot = 500, nX = 33, DOF = 467
XCond = 2.63784e+16 (normalized)
ERROR: design is ill-conditioned
Before the error it gives a warning that the design matrix is singular or badly scaled:
[Warning: Matrix is close to singular or badly scaled. Results may be
inaccurate. RCOND =  1.271035e-17.]
[> In fast_selxavg3bc (line 318)]
norm(inv(XtX),2) = [14601308845709730.0]
Saving X matrix to /home/hector/SENS_MAPPING_BLOCK/sessions/HEC_035_1/bold/MPBDandefs_orig_lhrh_revd11.5_sm2.lh/Xtmp.mat
XCond = 2.63784e+16 (normalized)
ERROR: design is ill-conditioned
...
...

ERROR: fast_selxavg3bc() failed\n
Many thanks in advance,
Martin