Page 1 of 1

No experimental setup description for BCI MI dataset

Posted: Mon Feb 11, 2013 4:14 pm
by Jay
Hi. I'm looking for pre-recorded motor imagery datasets for use in testing classifiers.

I came across the dataset on the website

http://openvibe.inria.fr/datasets/

but when it comes to describing the experiment it says
The experiment followed the Graz University protocol as described in [ref].
What was "ref" meant to be? There's this

http://openvibe.inria.fr/motor-imagery-bci/

on the main site but it has no info.

Re: No experimental setup description for BCI MI dataset

Posted: Tue Feb 12, 2013 8:56 am
by fabien.lotte
Hi Jay,

If I am not wrong, these data were collected using the standard protocol designed by the Graz BCI lab, which is described in most their publications, e.g., the following one:

Pfurtscheller, G. & Neuper, C.
Motor Imagery and Direct Brain-Computer Communication
proceedings of the IEEE, 2001, 89, 1123-1134

You can also have a look at the OpenViBE scenarios which implement this protocol (the ones presented in the page you mentioned: http://openvibe.inria.fr/motor-imagery-bci/). Open this scenario and look at "Graz Motor Imagery BCI stimulator" box (a Lua stimulator box) . The exact timing of the protocol is there, together with which stimulation is used to indicate which event (instructions, type of motor imagery, etc.).

Finally, the data set you mentioned has been used in the following paper :

Exploring Two Novel Features for EEG-based Brain-Computer Interfaces: Multifractal Cumulants and Predictive Complexity, Nicolas Brodu, Fabien Lotte, Anatole Lécuyer. Neurocomputing 79: 87-94 (2012).
The pdf is available here: http://nicolas.brodu.net/common/recherc ... fac_pc.pdf
The source code associated with the paper is available here: http://nicolas.brodu.net/common/recherc ... src.tar.gz

I hope this helps!

Best regards,
Fabien

Re: No experimental setup description for BCI MI dataset

Posted: Tue Feb 12, 2013 9:44 am
by jtlindgren
I took the liberty to add those links to the web page in question. :) Thank you Fabien.


Cheers,
Jussi