[pymvpa] Obtaining max responsive masks via MVPA

Yaroslav Halchenko debian at onerussian.com
Wed Oct 6 16:48:53 UTC 2010

On Wed, 06 Oct 2010, MS Al-Rawi wrote:
>    First, I have a question about the data
>    in [1]http://data.pymvpa.org/datasets/haxby2001/
>    Was it preprocessed for;  motion correction, slice-timing, normalized
>    to MNI, or smoothed?  or any other preprocessing?

AFAIK nothing was done to it

that subj1 was taken for the tutorial_data and there it was slightly

>    The max-responsive masks it contains were obtained per subject via GLM
>    contrast based localizer maps, this goes for mask4_vt.nii too. Why
>    there isn't one mask (per category or vt) for all subjects via group
>    analysis in order to use it in MVPA analysis?

because no one created such I guess ;-) and each subject VT is different
(if you want to be more or less precise)

>  Also, is it possible to
>    obtain such masks using MVPA via searchlight or any other alternative?

sure ;) but, be warned that to avoid the independence and circularity in
the data analysis, if you are to derive functional localizers masks
(e.g. FFA etc) you better use a separate dataset (run), or do it along
with your MVPA analysis within cross-validation folds (thus possibly
having different masks across folds)

=------------------------------   /v\  ----------------------------=
Keep in touch                    // \\     (yoh@|www.)onerussian.com
Yaroslav Halchenko              /(   )\               ICQ#: 60653192
                   Linux User    ^^-^^    [175555]

More information about the Pkg-ExpPsy-PyMVPA mailing list