What Are the Different Types of EEG Electrodes?

In the practice of EEG, evoked potentials (EP) and self-generating potentials (EEG) are components that reflect the activity of different neural sources. Each component can be described by characteristics such as polarity, scalp area, frequency spectrum, latency, and voltage amplitude. In order to obtain accurate values of these characteristics, the potential with the inactive electrode as the reference is the raw data we most want to get. However, it is well known that in fact only the potential difference can be measured, so setting a reference electrode in a non-invasive scalp record is unavoidable. In the current EEG practice, people have used or are using head and non-head reference electrodes such as earlobe, neck and average reference electrode, etc., each kind of reference electrode has a certain impact on the record.

Zero reference background for EEG records

In the practice of EEG, evoked potentials (EP) and self-generating potentials (EEG) are components that reflect the activity of different neural sources. Each component can be described by characteristics such as polarity, scalp area, frequency spectrum, latency, and voltage amplitude. In order to obtain accurate values of these characteristics, the potential with the inactive electrode as the reference is the raw data we most want to get. However, it is well known that in fact only the potential difference can be measured, so setting a reference electrode in a non-invasive scalp record is unavoidable. In the current EEG practice, people have used or are using head and non-head reference electrodes such as earlobe, neck and average reference electrode, etc., each kind of reference electrode has a certain impact on the record.

Since the neural electrical activity is a spatiotemporal process, the influence of active reference electrodes exists in both space and time. The spatial effect of the reference electrode is shown by adding or subtracting a constant to all electrode records, like raising or lowering the horizontal plane, but does not change the surface state. Therefore, it can be considered that the reference electrode has a certain effect on the spatial information , But this impact is not substantial.

The effect of the reference electrode in the time domain is due to the electrical activity of the position of the reference electrode. If a point or average on the surface of the human body is active, its potential changes over time. When it is used as a reference, it is equivalent to introducing an unknown and time-varying signal into all electrode records. It can be seen that the active reference electrode will affect the time domain dynamic analysis and spectrum analysis of the EEG. In order to solve this problem, people naturally hope to have a neutral potential point as the reference electrode. Since the point in space is far away from the source of neural activity, such a point will not have any effect on the EEG recording, so its The potential can be assumed to be zero and is therefore our ideal reference electrode point. In fact, the infinity point has always been the default reference potential point when analyzing field problems in electromagnetics.

In the EEG literature, finding approximate neutral reference electrodes or zero-potential references has been a controversial topic. The usual approach is to subjectively determine that the change in electrical activity at a location is relatively small and choose one based on the specific situation. Another common practice is to use an average reference, which is based on the theory that when the electrodes are densely distributed throughout the brain, the theoretical sum of the potentials of all electrodes is zero. However, the actual situation is often that only the upper hemisphere has electrodes distributed and the number of electrodes Limited, so the average reference also introduces significant non-zero errors.

In order to solve or alleviate the problem of non-zero reference that has long plagued the field of EEG, two new technologies have appeared in recent years, one is Laplacian (SL) (Hjorth, 1975), and the other is infinite zero reference (REST) (Yao, 2001). What SL gets is an estimate of the radial current density flowing to the head meter. The process from the head meter potential to SL is irreversible, and the calculation process is sensitive to noise. The data of EEG processed by REST is still potential, so it is suitable for all current EEG research based on the concept of potential.

REST Zero Reference for EEG Records Zero Reference for EEG Records (REST)

EEG zero reference technology (REST: Reference electrode standardization technique), instead of looking for a neutral reference electrode point or potential zero point on the scalp, it will approximately refer to a record on the scalp or the average potential as a reference , Which is converted to a record that uses an infinite point in space as a reference electrode. The physical basis of this conversion is that the potentials before and after the conversion are all generated by the actual neural activity source in the brain or their equivalent sources, so the potentials before and after the conversion can be linked by a common physical source.

The physical principle of REST is: Assume that the scalp potential V based on the zero reference can be expressed as V = GX, X is the source of electrical activity in the head, and G is the head model. In the actual EEG recording, we can only use non-zero reference (a) for EEG recording, so we get the potential v _a which is a non-zero reference, which can be expressed as

Where Ga is the head model in the case of non-zero reference. Based on this, we can invert the latter equation to get X, and then bring it into the previous equation to get V in the case of zero reference.

. Obviously we don't need to know the real source information X here, it is only used as a bridge. Therefore, we can assume that X is a set of equivalent distributed sources of real sources, thereby greatly simplifying the operation of completing the above calculation steps. . The same head model can also be an equivalent model in the equivalent sense. In the existing literature, the head model is generally a three-layer concentric sphere model. The source model is generally a discrete dipole layer model, but it can also be an equivalent distributed point charge model or an equivalent multipole model. In principle, these models should get similar effects, but due to the different mathematical nature of the equations involved in the specific problem, there will be slight differences in actual results, but there will be no substantial differences.

Regarding the effect of REST, there have been many literatures for detailed analysis and research. The results show that REST is particularly effective for superficial cortical areas where advanced brain function is particularly important. A large number of studies on different electrode arrays, different conductor models, and the effects of noise further show that the effect of REST has a certain relationship with these parameters, but REST is always more than the commonly used average reference and connected ear reference The fact of small reference errors is consistent. Figure 1 is an example of a simulation.

Application of Zero Reference in EEG Recording

REST approximately reconstructs the true potential of the reference electrode in the EEG record, rather than the question of whether there is a true zero potential point on the head surface that was previously debated. Because the problem of non-zero references is widespread in current EEG records and is a historical issue since the emergence of EEG, with the application of REST, some previous research conclusions may be revised or overturned. For example, it makes us rethink the following question: Are the spectral characteristics of different brain states consistent with the conclusions we have obtained from previous records using a point on the head table as a reference electrode? Is the amplitude difference between different states, the latency difference of the induced EEG under different tasks, and the correlation between the electrode points and the brain network significantly different? All these require new experimental research to answer. At the same time, in order to further reduce the REST error, people will adopt more multi-channel EEG systems, which may have a significant impact on related industries.

It is important to note that many researchers in the field of electrophysiology have been hoping to find an inactive free reference electrode or zero reference point. However, nearly a century of EEG research and practice shows that this is just an unrealistic fantasy. However, computer-based reconstruction of REST illustrates the fact that the influence of the reference electrode can be significantly reduced, thereby improving the time domain analysis and spectrum analysis of the EEG. The REST method and its application are just beginning, and we look forward to more colleagues joining this research.