Dear FSL experts,

I am running Melodic with task fMRI data (Multi-session Tensor ICA). I have three doubts.

1. I have noticed that one of the participants had problems with registration. I have fixed this, and now I have to run...only this participant, or all the participants (n = 80). I am in the individual estimation of the IC (timeseries and spatial maps).

2. I have been visually inspecting the output from Melodic. I have 26 IC and lot of them are just noise I have to reject. However, I have seen that even the ones that seem task-related, when I give a look to the temporal mode graphic, the IC time course and the full model fit do not match. I wonder if it could be the case that I am using a highpass filter of only 90.

3. Finally, I have a doubt: to run the the group comparisons (anova 2*2) should I run dual_regression or in this case Glm? If Glm, should I include this last part in my design.fsf, the part of post-stats?

Thank yoiu for you valuable feedback.

Best regards,

Rosalia