bci2000.org BBS

Dear BCI2000 community,

I would like to record EEG and EMG signals with TMSi amplifier. It is a fact that TMSi-amplifier used common average referenz (CAR) method for decrease the noise.
That means that the EMG signals are affected the EEG signals. Is it possible to turn off the CAR-filter in BCI2000 for EMG signals? If not, maybe it is possible record data and afterwards recalculate EEG signals and remove the EMG influence.
Let assume we have 2 EEG-signals and 1 EMG-signal
CAR-filter
x1 - (x1+x2+x3)/3 = a
x2 - (x1+x2+x3)/3 = b
x3 - (x1+x2+x3)/3 = c
I tried to solve system of equation, but...
B = 2/3 -1/3 -1/3 X = x1 A = a -values in .dat file B*X = A goal to calculate X
-1/3 2/3 -1/3 x2 b
-1/3 -1/3 2/3 x3 c
det(B) = 0-> B is singular matrix-> X = A/B can't solve
Maybe one of you coped with this problem.
Thank you

As you found, the CAR filter matrix removes one dimension from the signal, and is thus singular. You may still invert it by reducing the dimension of the input, which means losing one channel:

For one channel, connect both inputs to ground (if bipolar) resp. its single input to ref (if unipolar). In that channel, you will then measure the negative of the Common Average, such that you may revert the CAR filter by subtracting that channel from all other channels.

Regards,
Juergen

Many Thanks!!!

We always record signals unipolar. What means "its single input to ref "? TMSi amplifier don't record reference signal. Instead of reference it used CAR-approach to decrease the noise.

Thanks

Connect it to ground then.

"To connect the single input to ground" isn't the solution. The signal in this channel is always zero. I guess the amplifier can recognize that the input is zero. And this channel is excluded from CAR calculation.

So how do you know that the amplifier applies a CAR filter at all?

Because
sum(Ch_i(t = t0)) = 0
i = 1:numberOfChannels

Because
sum(Ch_i(t = t0)) = 0
i = 1:numberOfChannels

What is t0? Do you mean that the sum of channel signals is zero at some initial time point t0, or that it is constantly zero, independent of time?