 |
 |
Hi FSLers, I plan to use randomise for an fMRI analysis within several regions of interest. I have non-overlapping functional ROIs defined by previous studies and these vary in shape and size. I would like to use the TFCE thresholding option but want to verify the appropriateness of this approach for my purposes: Because of the cluster-like features of TFCE, is it appropriate to use this approach with several ROIs of different sizes? My plan is to run my analysis separately for each ROI so that the cluster distributions are not influenced by different ROI sizes/shapes, as might be the case if combining the different ROIs into a single mask. I would therefore correct for multiple comparisons by using FDR thresholding on the 1-p TFCE output for each ROI. Also, does it make sense to report the cluster-based information from the TFCE output if looking at results within a small and/or irregular shaped ROI, for example, the intraparietal sulcus? Thank you, Michael