Hi, FSL experts:

 

I am trying to make a design matrix for  the timeseries model  for the  Post-Stats in MElODIC, so I could get the Z-statistics and p-value for the defined contrasts with the IC maps.

I have two conditions in each run and I want to make three contrasts, cond1-fixation, cond2-fixation,cond1-cond2.

The problem is that the order of the two conditions are different acoss total 7 runs of each subject, in other words, I have an unique experimental design for each run.

When I specified the two conditions as two EVs and try to use the 3 column format in the Glm Gui to automatically generate the design.mat and design.con for Post-Stats of Single-session ICA for only one run , it seems work perfectly. But when I want to conduct Multi-session temporal concatenation for total 7runs, I don't know how to write the design matrix for the concatenated data derived from 7runs. It seems not reasonable to combine all the time points of 7 runs together since they belong to different runs, right? Does any one know how to define such a design matrix for Multi-session temporal concatenation ICA, or if the MELODIC can only concatenate input datas of same experimental design?

 

 

Best Wishes!