SSVEP and Motor Imagery with Emotiv

Working with OpenViBE signal processing scenarios and doing scenario/BCI design
Kevin
Posts: 3
Joined: Sun May 13, 2012 2:33 pm

SSVEP and Motor Imagery with Emotiv

Post by Kevin »

Hello all,

I try to perform some BCI applications using OpenVibe and Emotiv headset. In this case, I have succeeded to reproduce P300 speller with xDAWN Spatial Filter. But I have some problems on running motor imagery and SSVEP applications.

For motor imagery, my acquired data are very bad that the training process does not stop (even until 100h training).

In SSVEP, I succeeded to reach the classifier performance of about 81%. But, when I run the online scenario, the result is not good and I have a little control on the "ship". Even, when I remove the headset the ship moves for itself. It may need some configurations for Emotiv headset.

I would be very thankful, if you can help through better performance of these two applications.

Kind Regards,
Kevin

jlegeny
Posts: 239
Joined: Tue Nov 02, 2010 8:51 am
Location: Mensia Technologies Paris FR
Contact:

Re: SSVEP and Motor Imagery with Emotiv

Post by jlegeny »

Hello Kevin,

for the SSVEP scenarios try using lower frequencies. You can try 7.5Hz 6.666Hz and 6Hz for example. Emotiv headset has quite low acquisition frequency, also usage of lower frequencies will help to eliminate jitter of the display.

Cheers
Jozef

Kevin
Posts: 3
Joined: Sun May 13, 2012 2:33 pm

Re: SSVEP and Motor Imagery with Emotiv

Post by Kevin »

Dear Jozef,

Thank you. This time I used the frequencies (7.5, 6.666, 6) and I could reach to 86% performance. Now, I have more control over the ship. Also, I tried the white color but it isn't very effective. Is there any specific relations among the frequencies? for example is (6 , 5, 4) valid? Also, is there any other configurations or tips that can result in better performance? (for example during the experiment, does "focus on the boxes" just mean normally look at them? can I blink during the session?)

Do you have any ideas on improving the performance of motor imagery application? I think the main problem is related to the data acquisition phase.

Thank you,
Kevin

Nattaya
Posts: 4
Joined: Mon Jun 11, 2012 10:00 pm

Re: SSVEP and Motor Imagery with Emotiv

Post by Nattaya »

Hi! Kevin
I have just started to work with Openvibe.
You mentioned that you have succeeded connect and using the classified scenario with Emotiv.
However, I failed doing so. Therefore, I would like to know if there is any configuration I need to change?


The 3 errors that I got are something like this.

[ ERROR ] At time 0.125 sec <Box algorithm::Spatial Filter> Bad matrix size - Expected 4 input channels and received 6 input channels
[WARNING] Box algorithm <Spatial Filter> has been deactivated because process phase returned bad status


Thank you

jlegeny
Posts: 239
Joined: Tue Nov 02, 2010 8:51 am
Location: Mensia Technologies Paris FR
Contact:

Re: SSVEP and Motor Imagery with Emotiv

Post by jlegeny »

Hello,

this error means that a Spatial Filter box has received incorrect number of channels. This usually happens when a CSP filter is badly trained/configured.

Can you please precise the exact procedure you have followed before you got the error? Which scenarios have you launched, in which order and so on.

Cheers
Jozef

Nattaya
Posts: 4
Joined: Mon Jun 11, 2012 10:00 pm

Re: SSVEP and Motor Imagery with Emotiv

Post by Nattaya »

Hi! Jozef
I followed the instruction in this page

http://openvibe.inria.fr/steady-state-v ... otentials/

up until classifier training second Scenario to use : ssvep-bci-4-classifier-training.xml

When I press to run the scenario it gave me the errors.

Thank you
Nattaya

Nattaya
Posts: 4
Joined: Mon Jun 11, 2012 10:00 pm

Re: SSVEP and Motor Imagery with Emotiv

Post by Nattaya »

I finally got it working :D
The performance is 84%

wuwu622
Posts: 5
Joined: Tue May 29, 2012 4:35 pm

Re: SSVEP and Motor Imagery with Emotiv

Post by wuwu622 »

Hi nattya, I am trying to run the same scenarios as you and am running into similar problems.

Could you please give some insight on how you resolved the incorrect # of channels error??

Nattaya
Posts: 4
Joined: Mon Jun 11, 2012 10:00 pm

Re: SSVEP and Motor Imagery with Emotiv

Post by Nattaya »

I actually did not do many things. I just didn't know that when I run "ssvep-bci-3-CSP-training" scenario I had to wait until it actually finish running. I rushed to the next scenarios. :)

jlegeny
Posts: 239
Joined: Tue Nov 02, 2010 8:51 am
Location: Mensia Technologies Paris FR
Contact:

Re: SSVEP and Motor Imagery with Emotiv

Post by jlegeny »

Glad to see that it is working.

ipyd
Posts: 5
Joined: Sat May 05, 2012 1:50 pm

Re: SSVEP and Motor Imagery with Emotiv

Post by ipyd »

Hi all,

first of all, i would like to discuss a general topic regarding SSVEP using Emotiv EPOC.

in my experiment, i am using frequency stimulus from openViBE and then save EEG data using 'Testbench' software provided by Emotiv SDK.
from the captured data, i calculated FFT of signal O1 and O2.

however, i didnot see any signals other than noise.

actually, i just want to check whether or not the EEG signal has the same frequency as the generated signal.

so, is there any step to acquire SSVEP signal from Emotiv EPOC ? can we just take the FFT of O1 and O2 signal to see SSVEP signal ?

thank you all for the kindly support.

jlegeny
Posts: 239
Joined: Tue Nov 02, 2010 8:51 am
Location: Mensia Technologies Paris FR
Contact:

Re: SSVEP and Motor Imagery with Emotiv

Post by jlegeny »

Hello,

you will have to filter the signal before being able to see anything. The EEG from the Epoc headset is quite noisy, so you will have to filter at least the 50Hz and low (<4Hz) frequencies.

Also, in order to "see" the SSVEP response in the EEG, you will have to calculate an average over a longer period of time (7 seconds for example). Try to look at the SSVEP scenarios included in OpenViBE to see how the signal processing is made there.

Cheers
Jozef

razano1
Posts: 18
Joined: Wed Jun 20, 2012 1:56 pm

Re: SSVEP and Motor Imagery with Emotiv

Post by razano1 »

Hi jozef...

i have mentioned in post viewtopic.php?f=4&t=413 that i am working on a customized SSVEP app...
i wana mention one thing.. i find jitter in trainer app stimulus... missing frames and lags ... few but observable .. i am not exactly sure how much it will effect the performance... online app works well in this regard..

it might be due to the constant tcp/ip communication .. any suggestions to improve this ...

In the trainer App, What exactly is the impact of target sequence, is it in some random order or what ??? i am using four frequencies btw..

Best Regards

Asim

jlegeny
Posts: 239
Joined: Tue Nov 02, 2010 8:51 am
Location: Mensia Technologies Paris FR
Contact:

Re: SSVEP and Motor Imagery with Emotiv

Post by jlegeny »

Hello,

the training sequence is indeed in randomized order (by bulks of 0 1 2 3) in order to remove the possible ordering effect. It should not really matter though, as long as each frequency is stimulated for the same amount of time.

Regards
Jozef

razano1
Posts: 18
Joined: Wed Jun 20, 2012 1:56 pm

Re: SSVEP and Motor Imagery with Emotiv

Post by razano1 »

Dear Jozef...

Did you consider the harmonics/sub harmonics of stimulating frequencies in this demo ssvep app... ???
Is it possible to involve them in Openvibe environment ??? if yes any clue regarding that

Thanks and Best Regards

Asim

Post Reply