Hi FreeSurfer Experts,


I am trying to run a dods cortical thickness analysis examining the interaction of group on a continuous variable while controlling for other variables. 


Specifically, I have 6 classes (gender by three groups [zero, one, two]). I am interested in the group interaction with a continuous variable while controlling for gender (among 4 other continuous variables). I set up my fsgd file (see attached) and contrast file (36 dimensions- F test):

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 [1 -1 0 1 -1 0]/2

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 [1 -0 -1 1  0 -1]/2


and ran this command:


mri_glmfit --y  lh.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forfsgd.thickness.10.mgh --fsgd fsgd.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forFSGD.152.thickness.08082017.txt dods --C /GLM_CONTRAST.MAT/Contrast.Male.Female.Age.PC1.PC2.PC3.GGT.RORAbySex.mtx --surf fsaverage lh --cortex --glmdir lh.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forfsgd.thickness.10.glmdir 


But keep getting this error:


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

--------------------------------

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.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forfsgd.thickness.10.mgh --fsgd fsgd.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forFSGD.152.thickness.08082017.txt dods --C /Volumes/VA_Imaging/Projects/salat/2389_TRT/dmiller/GGT_CRP/GLM_CONTRAST.MAT/Contrast.Male.Female.Age.PC1.PC2.PC3.GGT.RORAbySex.mtx --surf fsaverage lh --cortex --glmdir lh.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forfsgd.thickness.10.glmdir 

  2. The FSGD file (if using one)

  3. And the design matrix above



Any idea on what I am doing wrong?


Danielle R. Miller, Ph.D.

National Center for PTSD

VA Boston Healthcare System Jamaica Plain

Boston University School of Medicine

OFFICE: (857) 364-4022