FAQ

 

What is the difference between the REF and the GND inputs?

The GND input is used to establish electrical ground continuity between the subject and the amplifier ground plane. This plays a role in electrical safety as it provides a safe return path for any leakage currents. But its main role is to reduce the effect of electromagnetic interference from the power line. The REF input on the other hand is actually used by the input amplifier in deriving the EEG signal. The signal at the REF input is subtracted from the signal at the Active input to create the final EEG output. In order to function optimally, the REF input signal should have the same "common mode" signal as the active input leads, but not have any actual brain signal. For this reason the REF electrode is most commonly located on the earlobe. The earlobes tend to have higher impedance levels, and also tend to pick up an EKG signal. That is why the Alpha400 amplifier features the ability to LINK the two REF inputs together. By linking the REF inputs a lower level of impedance is achieved and because the polarity of the EKG signal is opposite on each ear, the EKG artifact tends to be cancelled out.

 

Can I perform bipolar recordings with the Alpha amps?

Yes, it is possible (if you insist). You can put an active scalp electrode into a REF jack and obatin a bipolar recording. For example, if you wanted to record F3-C3 then you could put F3 into Ch. 1 active and C3 into the REF. This would give you F3-C3 on channel 1. Since in a bipolar input when the EEG under each electrode is the same, the signal will be canceled out, it used to be, back in the day when single channel amplifiers were the norm, that doing a bipolar input was considered to be a "poor mans phase feedback" protocol. But these days it is rare to see anything less than a 2-channel amplifier, and there is almost nothing that can be accomplished with a bipolar input that would not be better accomplished with a pair of monopolar inputs.

Here is the problem: with a bipolar input you are literally blind as to the nature of the EEG activity that actually occured at each of the two electrode sites. In the above example for instance, say you had some protocol in which you fedback some parameter derived from F3-C3. In some cases there may be almost no signal at F3, and large amounts of activity occuring at C3 .... In other cases there may be large amounts of activity occuring at F3 and almost nothing happening at C3. In both cases, the EEG ouput signal from F3-C3 is going to be exactly the same! In fact there is literally an infinite number of possible patterns of activity that could have occured under each electrode position that would result in the identical output signal. You ONLY know the output, you are totally blind as to what the contribution to that output was from either electrode site. So, what if, just as a hypothetical example, when predominant activity is at the C3 site you consistently obtain good results with your protocol .... but, when predominant activity is at the F3 site you do not see any results? If you are performing bipolar recording you only know that sometimes your protocol works, and sometimes it doesn't, you won't have a clue as to why that is the case. But, if you had recorded both F3 and C3 with a monopolar input, and then in software derived your F3-C3 signal .... Now, you can veiw the actual activity at F3 and at C3 along with the F3-C3 signal. Now you have the opportunity for significant clinical insights as to the differences that existed in the EEG. That capability to see the actual EEG at each site would have been lost by using a bipolar input.

This discussion regarding monopolar vs. bipolar recording is as old as the field of EEG itself. In what is considered to be one of the "Bibles" of EEG recording, "Medical Electroencephalography" by Dr. Gibbs in 1967, Dr. Gibbs states: "Bipolar recordings scramble the wave forms by superminposing the activity of one brain area on the activity of another, and in such recordings the true electrical signal is not immediately identifiable. As a consequence, bipolarists commonly overlook, or disregard, patterns that have been shown to have clinical significance. Bipolar recordings are cumbersome and confusing and should be avoided if possible." There are other papers and studies that have been written on the subject, all of which have concluded that unless there is some other compelling specialized requirement, monopolar recording should be used. Specifically this is due to the fact that with bipolar recodring you are literally blind to what is actually happening at each electrode position, whereas with monopolar recording you see the actual true EEG signal at each of the electrode sites.

 

What is the normal impedance range for a good electrode connection?

The standard answer has always been that electrode impedance should be less than 10K-ohms. In fact, that probably is a good rule of thumb in normal practice, and impedance values below 10K-ohms can pretty easilly be achieved after a little practice. But the real answer though is a little more involved. In reality our amplifiers will routinely produce high-quality, clean EEG signals with electrode impedance values that can be as high as 30K-ohms, or even more. The more important factor is how well the electrodes are matched. If one electrode is 5K-ohms and the other is 30K-ohms, the resulting signal will actually be worse than it would if both electrodes were 30K-ohms.

 

Wouldn't a 24-bit A-D converter produce better EEG resolution than your 16-bits?

This is a very common misconception, and the answer is that, relative to the way the 24-bit converter is used in virtually all amplifiers that use one ... absolutely not. When you consider the resolution of the EEG that the amplifier is providing you need to consider how many bits there are over what input range. For example, consider if a 24-bit A-D converter is used with an amplifier input range that is 100 millivolts. Remember that 1 millivolt is equal to 1,000 microvolts. So, an input range of 100 millivolts means that the span the A-D converter is digitizing is actually 100,000 microvolts big. Now, when you realize that most EEG is in the range of maybe 25 to 50 microvolts ... that means the A-D converter is never utilized over anything other than just a very very tiny sliver of the bottom end of its range. The Alpha series EEG amplifiers are one of (if not THE) only amplifiers to feature a true analog gain control. The input range of the Alpha amplifiers can be selected to be 50, 100, or 500 microvolts, and the full resolution of our 16-bit A-D converter is applied to that range. The result is an effective input resolution of the actual digitized EEG that is many orders of magnitude greater than results from applying a 24-bit A-D conveter to input ranges that are hundreds of thousands of microvolts.

 

Can the Alpha Series EEG amplifiers be used for ILF low-frequency training?

That is a difficult question because the definition of exactly what the ILF signal is remains so obscure. The frequency ranges that are being considered are so slow that they require on the order of tens of minutes, if not even hours, just to complete one single cycle. In order to evaluate whether the "energy" at that frequency is increasing or decreasing over time would seem to imply that you would need to measure for at least two complete cycles. Given the very long timeframe involved in determing if the energy level at those cycles is increasing or decreasing, the idea that "moment-to-moment feedback" can somehow be generated really stretches the imagination.

In any case, without getting into the discussion about the practical considerations of whether such training makes any logical sense, the question still remains whether the Alpha series amplifiers could be used to implement such a protocol. Again, we get back to the question of exactly what the ILF signal actually is. Normally, this protocol uses a bipolar electrode input that is placed across the hemishperes. The result is that to the extent the EEG signal is the same under both electrodes the output will be cancelled out. This bipolar EEG signal is then fed into an ILF bandpass filter. The output from this filter is then the ILF signal that is used to drive the feedback. It would seem in this case that the "ultra-low frequency energy" is not actually going into the amplifier inputs through the electrodes. The ILF signal is rather a "derived" parameter that is created by filtering the bipolar derivation of an EEG of normal bandwidth. IF that is the case, then an ultra-low frequency response at the front end of the amplifeir should not be required in order to use it for ILF feedback. But, again, this is still very much a gray area and the parameters involved are not understood.