Hello
Background:
I am trying to run a one-way ANOVA, repeated measures, 2nd-level analysis. I have done some internet scouring, and found this link (which was useful) https://neurostars.org/t/fmri-group-analysis-2nd-level-which-1st-level-contrasts-for-which-model/625, but I'm still confused.
I have 13 subjects, who performed an fMRI task with 4 conditions (Red, Gold, Green, Control). They were scanned 4 times, and each time was 1 of the 4 conditions, thus it is a repeated measure design (same subjects scanned 4 times).
I have performed a 1st level analysis and it ran fine. I have contrast maps and spmT stat maps from each of the conditions. Now I would like to perform a 2nd level analysis, to see if there is a difference in the task response based on which condition the subject was under.
What I've done:
1. I tried a 1-way ANOVA, with 4 cells, and took the relevant contrast map (con_0001.nii) from each of the 4 conditions, and put 1 in each cell. This design ran, however, when I tried to estimate it, it threw an error "Please check your data. There are no significant voxels." Now, this might be true, i.e. there aren't actually any voxels that survive, but I wanted to check my design was right. The myriad of PPTs and help online refers to setting up complicated contrasts, however I can't even get to the Results tab if Estimate isn't running.
2. When I realised I should be using the "ANOVA Within-subjects" design, I was further confused, because I wasn't sure what to input in the "scan" section or the "conditions" section.
Questions:
1. Which design should I be using?
2. Am I correct, in that I should be using the con_000X.nii files in the 2nd-level analysis?
3. If I use the within-subjects design, what should be input into the scan and conditions sections?
Thank you!
Michael
|