Hello Anders,
The efficiency calculations are made using an "average" of the voxelwise EV - it is not calculated for every voxel independently. For your design it may be possible that the average of the grey-matter is close to collinear with one of the other EVs. If the data values in the voxelwise EV are ( e.g.) much larger than the other EVs then this can also make the efficiency output harder to interpret.
Kind Regards
Matthew
> Dear FSL Experts,
>
>
> I am using a GLM with seven EVs for a higher-level FEAT analysis.
> When I add a voxel-dependent EV to the matrix (grey matter anatomy created using feat_gm_prepare), I get a warning message that the matrix is now rank deficient.
>
> Does this mean that the values of one (or more voxels) are collinear with one of the other EVs?
>
> Could this be specified in more detail (the values of what voxels are linearly related to what EVs)?
>
> What are the consequences of this?
>
> I have attached the GLM efficiency output.
>
> All the best,
> Anders
> <design_cov.png>
|