Hi,

I want to investigate lGI differences between 4 groups (n=449) while controlling for sex, age and surface area using DODS design matrix on the command line stream of FreeSurfer. Will you please tell me which of the below two contrasts are correct for this purpose? 

I have ran both contrasts (separately) and although all steps have completed successfully, the permutation tests have been running for the past 2 days (now on poll ~14616) – I think this is normal given my previous experience with how long permutation tests take to complete with large numbers of participants such as this case.

 

However, I’d really appreciate it if you’d please inform me which contrast (1 or 2 below) is correct, and if I should just wait for the permutation tests to complete or if you think something may be wrong?

 

I have provided below information regarding my regressors:

 

In my data here, I have:

• 2 discrete factors: Diagnosis, sex
• 8 groups/levels/classes: TD M, TD F, NDD M, NDD F, ASD M, ASD F, ADHD M, ADHD F
• 2 continuous covariates: age, SA

 

• Number of classes: 8
• Number of variables: 2

 

• NregressorsDODS= Nclasses*(Nvariables+1)= 8*(2+1)= 24
• Regressor 1: ones for TDM subjects, 0 otherwise. Codes intercept for TDM
• Regressor 2: ones for TDF subjects, 0 otherwise. Codes intercept for TDF
• Regressor 3: ones for NDDM subjects, 0 otherwise. Codes intercept for NDDM
• Regressor 4: ones for NDDF subjects, 0 otherwise. Codes intercept for NDDF
• Regressor 5: ones for ASDM subjects, 0 otherwise. Codes intercept for ASDM
• Regressor 6: ones for ASDF subjects, 0 otherwise. Codes intercept for ASDF
• Regressor 7: ones for ADHDM subjects, 0 otherwise. Codes intercept for ADHDM
• Regressor 8: ones for ADHDF subjects, 0 otherwise. Codes intercept for ADHDF
• Regressor 9: age for TDM subjects, 0 othersie. Codes for TDM
• Regressor 10: age for TDF subjects, 0 otherwise. Codes for TDF
• Regressor 11: age for NDDM subjects, 0 otherwise. Codes for NDDM
• Regressor 12: age for NDDF subjects, 0 otherwise. Codes for NDDF
• Regressor 13: age for ASDM subjects, 0 othersie. Codes for ASDM
• Regressor 14: age for ASDF subjects, 0 otherwise. Codes for ASDF
• Regressor 15: age for ADHDM subjects, 0 othersie. Codes for ADHDM
• Regressor 16: age for ADHDF subjects, 0 otherwise. Codes for ADHDF
• Regressor 17: SA for TDM subjects, 0 otherwise. Codes for TDM
• Regressor 18: SA for TDF subjects, 0 otherwise. Codes for TDF
• Regressor 19: SA for NDDM subjects, 0 otherwise. Codes for NDDM
• Regressor 20: SA for NDDF subjects, 0 otherwise. Codes for NDDF
• Regressor 21: SA for ASDM subjects, 0 otherwise. Codes for ASDM
• Regressor 22: SA for ASDF subjects, 0 otherwise. Codes for ASDF
• Regressor 23: SA for ADHDM subjects, 0 otherwise. Codes for ADHDM
• Regressor 24: SA for ADHDF subjects, 0 otherwise. Codes for ADHDF

 

Contrast 1:

Null hypothesis: is there a difference in lGI between TD and any of the other 3 groups (NDD or ASD or ADHD groups) regressing out the effects of sex, age and SA?

0.5 0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

 

OR

 

Contrast 2:

Null hypothesis: is there a difference in lGI between TD and NDD or ASD or ADHD groups regressing out the effects of sex, age and SA?

0.5 0.5 -0.5 -0.5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0.5 0.5 0 0 -0.5 -0.5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0.5 0.5 0 0 0 0 -0.5 -0.5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

 

1^st line tests for a difference between TD and NDD

2^nd line tests for a difference between TD and ASD

3^rd line tests for a difference between TD and ADHD

 

The permutation command line I used:

mri_glmfit-sim --glmdir lh.lgi.TD_NDD_ASD_ADHD_age_sex_SA_FWHM0.glmdir --perm 1000 1.3 abs --perm-resid --cwp  0.05 --2spaces --bg 1

 

I’d appreciate your soon reply,

 

Thank you,

 

Avi