MVPA for Subject Label Classification

#1

Hi everyone,

I have a subject cohort with and without reading disabilities, and want to use a searchlight+SVM approach to see if there are a set of voxels during a reading task that may contain information to delineate between the two groups. The Nilearn tutorial on searchlight is focused on within-subject classification of different tasks, which is not what I want. Does anyone know of a good place to look for code for this? I’d prefer to not use MATLAB, but can if needed to. For a similar example of what I am trying to do, but for a different clinical population, see this paper on schizophrenia (Bleich-Cohen et al., 2014). Happy to provide more info as needed.

Thanks,
Steven

#2

Hi,
There is nothing in the Nilearn example that ties it to within-subject analysis. It works as well for inter-subject analysis.
This being said, I would advise you to have a look at brain-wide, instead of local models. See e.g.
https://nilearn.github.io/auto_examples/02_decoding/plot_haxby_frem.html#sphx-glr-auto-examples-02-decoding-plot-haxby-frem-py
Best,
Bertrand

1 Like
#3

Thanks for the reply, I’ll check that out!