OpenVibe + EPOC noise reduction question

[cbr0wn]

I am trying to grasp the concept of how OpenVibe reduces noise from EPOC headset. I have a scenario where I use the temporal filter to get just beta waves (about 15Hz to 30Hz) in order for users to have to controls (left and right) by having a user think about left and right hand motions. How does OpenVibe filter these brain signals so that I can be as accurate as possible in ensuring the correct control was intended to be executed?

As a side question are some brain signals that are easier to detect and use then others? For example are eye blinks easier to detect then foot movements?

Thank you for all your help!

[lbonnet]

Hi cbr0wn,

I am trying to grasp the concept of how OpenVibe reduces noise from EPOC headset.

The EPOC already does some noise reduction filtering inside the headset. The signal you get in openvibe is already filtered a bit.
Maybe there is more information in the EPOC documentation.

How does OpenVibe filter these brain signals so that I can be as accurate as possible in ensuring the correct control was intended to be executed?

If you take a look at the motor-imagery scenarios given with openvibe, you will have a concrete example of how the signal is processed. In this case, we compute a band power after temporal filtering, and classify this feature (left vs right hand MI).

However I must warn you, these scenarios are not adapted to the EPOC. The headset electrodes are not placed on the motor cortices, which makes the motor imagery detection harder...

As a side question are some brain signals that are easier to detect and use then others? For example are eye blinks easier to detect then foot movements?

Eye blinks are not brain signals, but motor activation of the muscles, same as jaw clenching. It's easy to detect in the EEG though.
Motor imagery (hands/foot/tongue) are quite hard to detect and requires training (from my personal experience).
P300 signal benefits from many years of study and we can detect it with good accuarcy, but I'm afraid it will be harder with an EPOC (noise and electrode placement).

You can simply extract some frequencies (alpha, beta, etc.) These frequencies, when picked on certain area of the brain, can be indicator of stress/concentration levels. Ongoing research... Check recent papers to know more.

Hope this helps,

Laurent

[AntonioTrotta]

Hi lbonnet,

I was reading your answer and I found it very helpful. I have a very quick question. The Emotiv Control Panel (Research SDK efition) has an interface (Affective Suite) that allows to display some sort of emotional reaction while recording. Elements are computed and displayed in real time.

Is it possible to capture the same emotional dimensions through OpenVibe?

Thanks in advance for your answer.

Regards,
Antonio

[lbonnet]

Hi Antonio,

Is it possible to capture the same emotional dimensions through OpenVibe?

We don't know what is the signal processing pipeline behind this "emotion detection" claimed by Emotiv.
In my opinion it may be EEG, EMG (muscular activity of the face) or a combination of both activities.
I'm not sure if they communicated on this, and how they really compute the Affective Suite values...

However, once you find precisely how to do it (frequency bands to select, filters, etc.) I'm pretty sure an OpenViBE scenario can do the job !

Laurent-