[pymvpa] Using SurfaceQueryEngine for spatial smoothing

[Swaroop Guntupalli]

If you are ok with getting a smoothed data in surface nodes (not
voxels), it is straightforward with the current setup in PyMVPA. You
just supply a mean feature measure to Searchlight with a trained
SurfaceQueryEngine object (mimicking surface searchlight example but
replacing cross validation transfer error measure).

Best,
Swaroop

On Fri, Apr 25, 2014 at 3:40 PM, Chris Johnson <effigies at gmail.com> wrote:
> Hi all,
>
> Bjornsdotter et al. (2011)
> (http://dx.doi.org/10.1016/j.neuroimage.2010.07.044) used a Monte Carlo
> searchlight method where a non-exhaustive searchlight is performed, and
> each voxel is assigned the average of the the information metric for
> those searchlights in which it is involved. Taken to its limit, this is
> just a spatial smoothing of a traditional searchlight analysis.
>
> The method for operating in the individual subject space is
> straightforward (construct a searchlight QueryEngine, take an average in
> each neighborhood), but I want to perform this in the FreeSurfer average
> subject (fsaverage) space. I thought I'd try this with the new
> SurfaceQueryEngine, but I'm not entirely sure how to train it.
>
> It looks like it wants a .nii loaded with voxels holding vertex indices,
> but I don't have one for fsaverage.
>
> Hopefully this is really easy and I'm just missing something, but the
> tutorials don't seem to be keeping pace with all the new features. :-)
>
> Thanks!
>
> --
> Christopher Johnson
> Ph.D. Candidate, Quantitative Neuroscience Laboratory
> Boston University
>
>
> _______________________________________________
> Pkg-ExpPsy-PyMVPA mailing list
> Pkg-ExpPsy-PyMVPA at lists.alioth.debian.org
> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-exppsy-pymvpa