P300 and CMS/DRL

Post Reply
Uli
Posts: 3
Joined: Wed Nov 17, 2004 6:37 pm
Location: EPFL Switzerland
Contact:

P300 and CMS/DRL

Post by Uli » Fri Nov 19, 2004 11:43 am

Hi Biosemi users,

I am recording P300 event related potentials with a 32 channel Biosemi Active Two system. Strangely enough I'm seeing an inverted P300 (i.e. negative peaks) at some occipital and temporal electrodes in my data (referencing to average reference does not change this effect).

To my understanding this occurs because CMS/DRL were located near to Cz during the recording. Now the CMS/DRL loop tries to compensate the positive voltage caused by a P300 via a negative voltage at DRL. This voltage spreads to more or less neutral electrodes and causes a negative P300. In addition the P300 at central sites is weakened.

My questions are:
Is the above view of things more or less correct or am I missing something? Is there a better position for CMS/DRL for P300?

Best Regards

Ulrich Hoffmann

Coen
Site Admin
Posts: 986
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Post by Coen » Fri Nov 19, 2004 3:09 pm

The above view of the operation of the CMS/DRL loop is fully incorrect. The CMS/DRL circuit cannot change the current distribution caused by physiological voltage sources. No extra current is sourced/sinked via the DRL (there is nowhere any current could flow, all other electrodes are connected to high impedance inputs). Only the leakage currents caused by the capacitive coupling of the subject and AD-box with interference sources flow via the DRL electrode (see http://www.biosemi.com/publications/pdf ... uction.pdf ). The CMS/DRL feedback loop generates a voltage at the DRL electrode so as to keep the potential difference between CMS and the amplifier Common (amp ground, ADC reference) as small as possible. In other words: the CMS/DRL circuit does not influence the potential differences between locations on the body, it only influences the potential differences between the CMS location and the amplifier Common.

Subtraction of two signals measured with respect to CMS cancels the influences of the CMS location, see http://www.biosemi.com/faq/cms&drl.htm and viewtopic.php?t=33. So, the location of the CMS and DRL electrodes do not influence the referenced signals measured by the ActiveTwo.

Finally, please note that ActiView displays and save the signals according to a "positive = up" convention.

Best regards, Coen (BioSemi)

Uli
Posts: 3
Joined: Wed Nov 17, 2004 6:37 pm
Location: EPFL Switzerland
Contact:

Post by Uli » Mon Nov 22, 2004 1:46 pm

Thanks for your answer Coen. Indeed my previous post was way too fuzzy, but not fully incorrect :). I thought more about it and here's what I think is happening:

1) Let the potentials of the electrodes Oz, Cz, CMS and the Amplifier Common (AC) be Oz = 1, Cz = 10, CMS = 10, AC = 0 (all with respect to a hypothetical perfect zero). Now CMS/DRL drives CMS a close as possible to AC. This results in the raw data Oz = -9, Cz = 0. Referencing to average reference gives Oz = -4.5, Cz = 4.5.

2) We change the position of CMS to somewhere near Oz: Oz = 1, Cz = 10, CMS = 1, AC = 0. With the CMS/DRL loop we have Oz = 0, Cz = 9. Referencing to average reference gives Oz = -4.5 , Cz = 4.5.


As you said the position of CMS/DRL changes nothing in the referenced signals. One can see negative peaks at Oz before and after referencing. Before referencing these result from the signals being measured with respect to CMS, and after referencing from taking the average reference.

Best Regards

Ulrich Hoffmann

Coen
Site Admin
Posts: 986
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Post by Coen » Mon Nov 22, 2004 2:13 pm

Your explanation is now clear and correct, but your conclusion is still (partly) wrong.

In the unreferenced (raw) signals you cannot see activity in the electrodes close to CMS. So, in your situation 2 (Oz close to CMS), it is not true that "one can see negative peaks at Oz before (..) referencing". Only after referencing, you can see the correct potential distribution.

Best regards, Coen (BioSemi)

Post Reply