I am running a searchlight analysis with nilearn.decoding.SearchLight.
I have 2 sets of data in input : data preprocessed with SPM (normalized and smoothed) and the same data preprocessed with fmriPrep (not normalized and not smoothed).
The input files are similar (matrix of 73x87x73 (or close when not normalized) with a resolution of 2.5x2.5x2.5).
However the outputs are different :
- when the input files are preprocessed with SPM, the output file has the same resolution and size as the input files (matrix of 73x87x73 with a resolution of 2.5x2.5x2.5)
- when the input files are preprocessed with fmriPrep, the output file has a resolution of 1x1x1 and a size of 192x256x256.
Furthemore, this resolution (1x1x1) and this size (192x256x256) are the same of the mask given to run the searchlight with fmriPrep preprocessed data. The mask given to run the searchlight with SPM preprocessed data is 2x2x2 and 91x109x91.
So I wonder why, with the fmriPrep preprocessed data, the voxel resolution and matrix size change in the output file?
Thank you very much!