EEG Signal Quality

There are many parameters of the EEG measurement system that affect the recorded EEG signal quality. While a gold standard for the estimation of signal quality for wet electrodes is skin-electrode impedance, there are many more aspects to consider, particularly, for dry-contact electrode measurements. In this section, we will overview major influences on EEG signal quality.

Skin-Electrode Impedance

EEG devices measure the potential difference between the electrodes placed on a person’s scalp. However, the currents associated with neural activity have to pass through a couple of layers/components each having its own impedance.

The skin has two layers known as dermis and epidermis. Dermis can be electrically modeled as a simple resistor and hence has frequency-independent impedance and is typically lower than the epidermis layer.  

Skin-to-electrode impedance is typically the largest contributing factor for dry-contact electrodes. Their contact area is smaller when compared with wet gel-based electrodes increasing their impedance. Impedance is also frequency-dependent and decreases with increasing frequency. If someone is interested only in higher frequencies of EEG recording this impedance is less important, however, this is rarely the case.

Finally, the material or the composition that the electrode is made of may have significant impedance as well. Nonetheless, this is typically lower when compared to other parts.

If the impedance of these parts is large then a couple of issues may arise. The impedance should be much smaller than the input impedance of the EEG measurement device otherwise there will be a voltage drop and the measured signal will be attenuated. However, as the electronics have significantly improved over the years, this is not an issue for many new devices. 

Another problem with increasing impedance is that all the parts that carry the signal up until the amplification are more susceptible to external noise such as mains 50/60Hz noise as they essentially act as antenna. However, shielding techniques and, for example, the use of active electrodes can help to overcome these problems.

Traditionally, a standard was established for wet, gel-based electrodes that the impedance should be below 5-10 kOhm to ensure good-quality recordings. However, as explained here signal quality depends on the particulars of the EEG measurement device, and in many cases, good-quality EEG signals can be achieved with much larger skin-electrode impedances. Therefore, impedance alone should not be treated as a main indication of signal quality.

Noise and Artifacts

Mains noise. Mains also known as line noise is typically present in most everyday environments. Therefore, unless a specific environment is set up for EEG measurements, the devices should be capable of dealing with this type of noise. Effective methods include shielding, active-shielding, using active electrodes, and others.

Flicker and thermal noise. These types of noise sources affect the electronics of measurement devices. Thermal noise is constant across all frequency components while flicker noise decreases with increasing frequency.

Offset and drift. These phenomena apply to measurements with dry-contact electrodes. The offset is a constant voltage also called DC voltage that arises due to half-cell potential, which in turn is related to the electrode material and skin’s chemical composition. This offset either has to be removed with extra analog input filters or the signal digitization part should have a large range and high resolution in order to be able to record the relatively small time-varying EEG signal together with a constant voltage offset which is typically large.

Drift, unlike constant offset voltage, is a slowly varying voltage which is generally present in dry-contact electrode measurements. This is typically removed in post-processing using detrending and filtering techniques. However, these techniques may also remove some useful content from EEG recording.

Electrode movement artifacts. Dry contact electrodes have to be kept very stable at their location. Even very small ‘micro’ movements will result in large voltage spikes in EEG recordings. Typically EEG caps or straps are used to keep them in place. However, the stronger the force holding the electrodes in place the less comfortable it becomes for the user.

Quality Estimation from EEG signal itself

Due to the mentioned drawbacks of using impedance as an indicator for EEG signal quality, we, at Neurotechnology, are advocating for quality evaluation from the EEG signal itself. Ultimately, it is the EEG signal that is of interest and it can be affected by many contributing factors, hence we believe quality determination from the signal parameters is a more appropriate and representative approach. In addition to this, impedance could typically only be measured before EEG recordings start as it is made over the frequencies that EEG signals are acquired. Hence, it is not possible to observe impedance changes during the measurements. On the contrary, the quality of the signals can be estimated during experiments, which allows observing if the quality degrades during the acquisition in real-time and responding accordingly.

Therefore, many parts of BrainAccess software evaluate EEG signal quality either for indication or for some process automation. For example, the BrainAccess Viewer has a tri-state quality indication, with the ‘red’ level meaning that the electrodes are not fitted properly, ‘yellow’ – that there is an EEG signal but the electrodes are not fully fitted or there are artifacts present such as blinks and ‘green’ – that signal quality is good.