Hi Desmond, There are two different things that are slightly mixed up. Let's clarify: A) Multi-modal analysis using NPC: this will seek areas where there is a joint effect, across modalities, that can be explained by the model. Two or more input files are supplied with "-i", and NPC combines across them for a single, joint inference. This is useful, for instance, to see if there are group differences jointly significant on a set of modalities. B) Voxel-wise (or vertexwise) EVs: this will run a GLM with one or more regressor that varies across space, that is, each voxel in the input file (entered with "-i") uses a different set of regressors, the latter being supplied with "-evperdat". This is useful, for instance, to see if Modality 1 is correlated with Modality 2 (on each voxel). It is possible to do both things together, e.g., combine modalities 1 and 2 via NPC, using a design matrix that has some voxelwise regressor 3. Before answering the question, I'd like to know whether you'd like to combine (jointly analyse) Modalities 1 and 2, or if you'd like to investigate things like correlations between Modalities 1 and 2. Thanks. All the best, Anderson On 9 January 2017 at 21:10, Desmond Oathes <[log in to unmask]> wrote: > Hello Anderson and FSL listserve, > > I’m using palm for a ’multi-modal’ analysis (voxelwise within subjects) > that I hope to use to explore what brain areas show convergence across two > sets of images. The outputs are looking a little funny to me so I would be > grateful for feedback on my steps: > > In the GLM set up, put in ‘1’ for all entries under ‘Group,’ and ‘1’ for > all entries under EV1. For voxelwise EV2, I put in the 4D dataset for > imaging modality #2. Under contrasts, entered ‘1’ for EV1 and ‘1’ for EV2. > > Then in octave, I used this XX model and contrast file to do: > > palm -i Modality1_4D.nii.gz -i Modality2_4D.nii.gz -d XX.mat -t XX.con -n > 10000 -T -npc -noniiclass -tonly -savemask -o Output_Image > > Then I looked to see if anything was significant in the output by: > fslmaths on the TFCE corrected output with -uthr 0.049999 > > Did I mess up anywhere up there? Would it matter if the inputs were > zscores from one or both modalities? > > Thanks lots! > Desmond >