Hi Lizanne,

PALM can be used in the same place where one would use randomise. If the input files are .nii or .nii.gz, the option -s shouldn't be used then, as the inputs are volumes. Otherwise it's fine.

That said, PALM could in fact take the -s, for FIRST vertex analysis, if the input (-i) were converted to either (a) a .csv file that had one subject be row and one vertex per column (transposed would also work with the proper option) or (b) were converted to the FreeSurfer .mgh/.mgz format (4D). In this case, the statistic would be truly computed in the surface, and TFCE and cluster inference would benefit from the actual areas of the surface of the subcortical structures. Although this can be done, conversion between these formats is currently a bit difficult, such that it might be simpler to just stay with the volume-converted NIFTI files. Maybe in the future there could be a file format that would help to deal with issues as these.

All the best,

Anderson

On 28 December 2015 at 23:58, Schweren, Lizanne <[log in to unmask]> wrote:

Dear FSL experts,

I am using PALM to perform vertex-wise group-level statistics, on FSL FIRST's output for surface-based vertex analysis. I run the following steps before PALM:

- run_first_all
- concat_bvars
- first_utils

The outputs from first_utils are images (.nii.gz) suitable for statistical testing in randomise. Am I right to conclude that I should not use the -s (surface) option if I chose to run the group-level stats in PALM?

If I do need to use the -s option, which file(s) should I specify here?

Thank you for your help, all my best,
Lizanne