How to do mvpa with afni?

AFNI Message Board
1

Hi,

This is a very naive question, I’m using afni for fMRI anlaysis and planning to use support vector machine for mvpa.
I downloaded LIBSVM in matlab.

How can I do mvpa with afni and matlab version of LIBSVM?
Also, if I download matlab library for afni, can I write a script for fMRI analysis as like shell script?

Thank you!

2

Hello,

The answer depends on the type of data/variable you want to input to MVPA and LibSVM. If you want to input the effect size (i.e. the betas) for each stimuli, you can compute them with 3dDeconvolve using --stim_times_IM. It might be a good idea to run 3dREMLfit to account for serial correlations. However, if the trials were highly correlated (e.g. in a fast event-related design), the program 3dLSS might be a better option as shown in the paper that described this approach.

http://dx.doi.org/10.1016/j.neuroimage.2011.08.076

You can also have a look at 3dSVM for the implementation of the support vector machines in AFNI.

Best,
Cesar

3

Thank you for your help.
I’ll look at the paper first. Thank you

4

vd.mpark Wrote:

Hi,

This is a very naive question, I'm using afni for
fMRI anlaysis and planning to use support vector
machine for mvpa.
I downloaded LIBSVM in matlab.

How can I do mvpa with afni and matlab version of
LIBSVM?

If you are familiar with Matlab, you can also consider using CoSMoMVPA (disclaimer: I'm the main contributor to the project).
It provides input/output support for AFNI through the AFNI Matlab library, so you can easily get AFNI datasets into a CoSMoMVPA struct, do analyses and CoSMoMVPA, and then map results back to an AFNI dataset.

website: http://cosmomvpa.org
code: GitHub - CoSMoMVPA/CoSMoMVPA: A lightweight multivariate pattern analysis (MVPA) toolbox in Matlab / Octave