Signal generator and offline analysis

[Seung-Chan Lee]

Hello. all.

I am a new member in BBS and this is first question.
I have a question about using signal generator and offline analysis.

Currently, our lab is focused on cursor task in BCI2000.
To verify experiment setup, I used the signal generator and offline analysis and used follow modules.

Source ; signal generator,
Signal processing ; dummy signal processing,
Application ; stimulus presentation.

To acquiring signal feature, I use ch.5(X axis) and ch.7(Y axis).
And I used left hand(Y axis, ch7) and right hand(X axis, ch5).


When I run offline analysis, I wonder that the r2 graph is right.
here is r2 graph.

this is rest vs left.(activated ch7, left)

this is rest vs right.(activated ch5, right)

this is left vs right..(activated ch7 and ch5)

Even if other channels are noise, other channels r2 value are not low.
I think other channels should have very low r2 value.
But result is diffrent.
It seems that noise channels are correlated with mouse movement.
Why that are possible? [/img]

[gschalk]

You probably have the CAR filter turned on.

Gerv

[Seung-Chan Lee]

Hello.

Thank you for response of the previous question.
I have a another question about CAR filter.
If high r2 value of the other channels caused by CAR filter, why is that?
And what is the role of the CAR filter?
How does the CAR filter affect the signal?

Please reply again. Thanks.

[gschalk]

The CAR filter simply removes the average of all the channels from a particular channel in question. This way, a strong activation in one channel will "bleed" into the other channels.

Check McFarland, Wolpaw, et al., 2000 for a discussion of different spatial filters.

Gerv