Dear Michaƫl,
SPM itself relies on fitting exactly the same model for each and every voxel in a dataset. If you could parameterise your predictor in a way that fitting beta coefficients per voxel to a set of global "predictor basis functions", you can include these basis functions in your SPM design and use SPM to fit the betas. If not, you are most likely looking after something like this: https://www.nitrc.org/projects/rbpm/. The project seems to lag behind recent developments in MATLAB and SPM, but it might be a starting point to look for a useful set of tools.
Hope this helps
Volkmar
|