Hi,
I have about 50 participants with DTI data collected across two time points (Pre and POST) each. Each subject got one of the two treatments (A or B). I want to compare the change in FA across the 2 time points for each of the 2 treatments and if there was an interaction between the time point and treatment, which fits the 2-factor 2-levels analysis example in the Glm manual. I want to know if I have it setup correctly for doing this:
1) We have pre-processed the DTI data for both the timepoints as suggested in the TBSS manual.
2) We have the all_FA_skeletonised file created for each of the two time points separately. For the purpose of the randomise command, I have merged the all_FA_skeletonised file for Pre and Post into a single all_FA_skeletonised file.
3) Design matrix:
Number of Inputs = total number of subjects in the merged all_FA_skeletonised file
EVs: Pre_A Pre_B Post_A Post_B
0 1 0 0
1 0 0 0
:
:
0 0 0 1
0 0 1 0
:
:
4) Contrast matrix:
To see main effect of time point ie if mean(Pre) = mean (POST), contrast would be [1 1 -1 -1];
To see main effect of treatment ie if mean(A) = mean (B), my contrast would be [1 -1 1 -1];
To see an interaction ie if Pre_A-Pre_B = Post_A-Post_B, my contrast would be [1 -1 -1 1];
Also, if the interaction reached significance for some FA tracks, how do I determine which contrast is contributing for the interaction to be significant?
5) Would fixed effects(cell means approach) answer all my questions or do I need to use random effects approach?
Thank you in advance
-- Regards,
Jyoti
|