Hello James,
We are just about to have a new version of FSL ( 5.0.9), which has a randomise with improved handling of voxelwise/overcomplete designs. I would advise running your analysis with this version, once it is released.
Kind Regards
Matthew
> Hi Anderson,
> Thanks very much for getting back to me and for your suggestions & advice.
>
> Unfortunately I don't think the problems are due to the overlapping nature of the lesions. The highest overlap I have is 6 subjects out of 63 who have lesions in the same voxel, so this shouldn't violate the assumption you mention.
>
> Further, I get the same error of Rows >= Columns when running a test on just 10 subjects, all of whom have no lesions, hence the lesion maps are just blank nifti files.
>
> Having discussed things with my colleagues, they have previously been able to run randomise on a setup_masks derived call with more columns that rows in the matrix file - but no longer. Could a bug have appeared in a recent update??
>
> Thanks again for the help,
> James
|