Dear SPM experts,
I have resting state data from healthy subjects and patients. I found that one resting state ICA component is more strongly connected with an area of the brain in the patient population. Next, I thought to investigate the parameter estimates of effective connectivity between nodes of the network and that region, so I ran a fully connected model with all of the major nodes of the network and with the region.
I have two questions:
1) Do I have to specify different models and run BMS, chose the best model and compare parameter estimates between the two groups, or is it enough to compare them with the fully connected model.
2) Can/should I optimize the model for each subject before comparing parameter estimates?
I would appreciate your help.
Best regards,
George
|