Generators of electric field which can be registered by scalp electrodes are groups of neurons with uniformly oriented dendrites. The neurons permanently receive impulses from other neurons. These signals affect dendritic synapses inducing excitatory and inhibitory postsynaptic potentials. Currents derived from synapses move through the dendrites and cell body to a trigger zone in the axon base and pass through the membrane to the extracellular space along the way. EEG is a result of summation of potentials derived from the mixture of extracellular currents generated by populations of neurons. Hereby the EEG depends on the cytoarchitectures of the neuronal populations, their connectivity, including the feedback loops, and the geometries of their extracellular fields (Freeman 1992). The main physical sources of the scalp potentials are the pyramidal cells of cortical layers III and V (Mitzdorf 1987).
Figure. Neuronal oscillators inside the cortex, discharging with their intrinsic frequencies (f1, f2, f3), produce extracellular currents summed on scalp surface as EEG signal. The spectral analysis decodes these oscillators activity out of EEG record. In the rectangle window the hypothetical scheme of neuronal oscillator is given. The axonal collateral of basic neuron activates the circuits with excitatory and inhibitory interneurons. The inhibitory neuron of the scheme is given in black.
The appearance of EEG rhythmic activity in scalp recordings is only possible as a result of the synchronized activation of massifs of neurons, the summed synaptic events of which become sufficiently large (Steriade et al 1990). The rhythmic activity may be generated by both pacemaker neurons having inner capability of rhythmic oscillations and neurons which can not generate a rhythm separately but can synchronize their activity through excitatory and inhibitory connections in such a manner that constitute a network with pacemaker properties. The latter may be designated as neuronal oscillators (Madler et al 1991; Kasanovich and Borisyuk 1994; Abarbanel et al 1996). The oscillators have their own discharge frequency, various among different oscillators and dependent on their internal connectivity (Figure 1), in spite of close intrinsic electrophysiological properties of single neurons which constitute different oscillators. The neuronal oscillators start to act in synchrony after application of external sensory stimulation (Lopes da Silva 1991; Basar 1992) or hidden signals from internal sources, for example, as a result of cognitive loading (Basar et al 1989).
The detailed circuitry of the neuronal oscillators underlying EEG rhythms was given in the Report of International Federation on Clinical Neurophysiology (IFCN) Committee on Basic Mechanisms (Steriade et al 1990).
The usual classification of the main EEG rhythms based on their frequency ranges is as follows: delta - 2-4 Hz, theta - 4-8 Hz, alpha - 8-13 Hz, beta - 13-30 Hz, gamma - higher than 30 Hz. But this classification only partially reflects the functional variation of rhythmic activities. For example, EEG rhythms within the alpha range may be distinguished by their dynamics, place of generation and relation to certain behavioral acts (Niedermayer 1997; Lutzenberger 1997; Pfurtscheller et al 1997).
Since the pioneering work of H.Berger (1929), the main EEG rhythm — alpha rhythm (Berger’s rhythm) has been known. This rhythm is typical of the resting condition and disappears when the subject perceives a sensory signal or when he makes mental efforts. It was shown that the alpha rhythm is generated by reverberating movement of nerve impulses between cortical neuronal groups and some thalamic nuclei, interconnected by a system of excitatory and inhibitory connections, resulting in rhythmic discharges of large populations of cortical neurons (Llinas 1988; Lopes da Silva 1991). In the visual cortex, however, the alpha rhythm could be generated by intracortical networks involving layer V pyramidal neurons, the latter being the main potential sources (Lopes da Silva and Storm van Leeuwen 1977; Steriade et al. 1990).
The theta-rhythm originates as a result of interactions between cortical and hippocampal neuronal groups (Miller 1991). The neuronal oscillators, which generate the beta rhythm, presumably are located inside the cortex (Lopes da Silva 1991). The basis for gamma oscillations is interneuronal feedback with quarter-cycle phase lags between neurons situated close to each other in local areas of cortex (Freeman 1992).
Most of the rhythms are rather widespread in brain structures. Induced gamma, theta and alpha rhythms were found in cortex, hippocampus, thalamus, and brain stem (Basar 1992). Freeman (1988) used the expression “common modes” for the existence of similar frequencies in various networks of the brain. This may play a role in the integration of activities of neuronal oscillators distributed over various brain structures. The candidate mechanism for such integration is synchronization of the distant neuronal oscillators’ activity on a fine temporal scale (this synchronization of spatially separated oscillators should be distinguished from a term synchronization usually applied to the enhancement of EEG rhythm amplitude due to the synchronized activity of large neuronal populations under one electrode).
The idea of brain potential synchrony as a leading mechanism for neuronal communications descends from some basic ideas of the Russian classic neurophysiological school of N.E. Vvedensky and A.A. Ukhtomsky (see Rusinov 1973). At the beginning of the century they postulated that the number of excitation cycles per time unit, i.e., discharge frequency, is a fundamental parameter, characterizing the neural structure functional state (the “functional lability” parameter). A.A. Ukhtomsky proposed that the coincidence of the functional lability of two structures promotes their functional connections. Developing these ideas, M.N. Livanov (1977) and V.S. Rusinov (1973) suggested that EEG rhythms reflect the parameter of functional lability. Subsequently, the EEG synchronization can promote and reflect for the functional connectivity between two or more cortical areas. The reason is that in this case signals from one neuronal oscillator repeatedly reach the other oscillator in one and the same phase of its excitation cycle. When this phase is the exaltatory one, the excitation threshold of the second oscillator is lowered, facilitating its neurons’ response and their recruitment in a concerted activity with the first oscillator neurons. On the contrary, when the phase of the second oscillator is the refractory one, the message can’t be received and this connection becomes silent. Thus both the frequency coincidence and appropriate phase relationship favor the neural communications. In this process the phase relationship controls the switch of connections from the active to the inactive state as well as its direction (both oscillators are like in a dialogue and can be sender and recipient of the message). The idea that synchrony of potentials promotes the neural connectivity was proven in a crucial experiment carried out by M.N. Livanov (1977). In this experiment a computer pursued the correlation coefficient between the EEGs in visual and motor cortical areas of the rabbit. It appeared that, if the correlation coefficient exceeded some level, the visual signal triggered paw movements, and if this coefficient was low, no motor reaction occurred.
The concept of EEG rhythm synchronization as a basic mechanism and a marker for cortical connections was later confirmed in a number of studies, including mathematical modeling of neural processes (Malsburg 1981; Abarbanel et al 1996). The study of EEG synchrony is now used as one of the main tools in the study of circuitry of neural communications both in animal experiments and in cognitive neuroscience (French and Beaumont 1984; Sviderskaya 1987; Gevins and Bressler 1988; Gray and Singer 1989; Ivanitsky 1990, 1993; Petsche et al. 1992; Petsche 1996; Bressler et al. 1993; Andrew and Pfurtscheller 1996).
EEG recording, as it was mentioned in the beginning of the chapter, is a rather routine procedure, particularly in clinics. Therefore the equipment for EEG is manufactured in almost all developed countries and its advertising and specification is presented in the journals of appropriate profile. All this equipment is supplied with detailed instructions for its use. Nevertheless it is worthwhile to present below some details of EEG recording procedure useful for researchers naive in this field.
The EEG recording usually include the follows steps:
- A subject is seated in comfortable chair in dimly illuminated room;
- Electrodes are placed on his head according to certain scheme;
- The reference electrodes are chosen;
- Parameters of electroencephalograph and software for EEG acquisition and storage are established;
- Calibration of electroencephalograph and data acquisition software is executed;
- EEG is recorded;
- Artifacts are removed.
EEG cabin. The EEG recordings is performed usually in a room shielded from outer electrical and magnetic fields. But modern amplifiers can reject these effects. During the recording procedure the subject should avoid movements, which can cause artifacts in a record.
Electrodes and their placement schemes. The most appropriate electrodes for the EEG scalp recording are Ag-AgCl which avoid potential shift due to electrode polarization. To get a good (i. e., with impedance below 5 Kilo-Ohms) contact between electrode and skin surface, the skin has to be cleaned with ether or alcohol for fat or dirt removal. Some abrasives were in practice earlier to lower the impedance, but it is unacceptable due to risk of bacterial, HIV and prion infection. An electrode gel or salt solutions are used to improve potential conduction between skin and electrode surface.
Figure. 10-20 electrodes placement scheme. According to this scheme three distances are measured: that between two preauricular points, that between the nasion (nose bridge) and inion (the occipital bone mount), both across vertex, and the circumference between the last two point of the skull. These distances are divided in proportion of 10-20-20-20-20-10% in both orthogonal axes and in circumference, and a net of imaging quadrates is built on head surface. The electrodes are placed in a quadrates angles.
The most popular scheme for electrode placement is the so-called 10/20 scheme (Jasper 1958) (Figure 2). Additional electrodes may be placed between the basic ones. According to “IFCN Standards for digital recording of clinical EEG“ (Nuwer et al 1998), amplification and channel acquisition must be available for at least 24 EEG channels. For artifact removal electrooculogram records are used. Now the most common way to place the electrode array on the scalp is the use of a cap with the electrodes fixed on it. These caps (or helmets) are available with different numbers of electrodes (19, 32, 64, up to 256 electrodes) and in several sizes, including ones for children (see, for example, the catalogues of “Electro Cap”, “Geodesic Sensor Net” and “NeuroScan”). Such devices can be placed and removed rapidly and cause a minimal unpleasant feeling. The latter is especially important for psychophysiological experiments, when a rather long recording is required. These caps automatically provide the electrode placement with appropriate, usually equal, interelectrode distance.
Reference electrodes. One of the important questions in EEG recording is the site of reference electrodes, relative to which the electric brain potentials in all other electrodes is measured. The reference electrodes should be placed on a presumed “inactive” zone. Frequently, this is the left or right earlobe or both of them. If one earlobe electrode is used as a reference, the topography of EEG rhythms is rather close to true, but there is the systematic decrease of EEG amplitude in the electrodes which are closer to the reference side. If “linked” earlobes are used, this kind of asymmetry is avoided but this distorts the EEG picture since the electric current flows inside the linking wire. This affects the intracranial currents that form the EEG potentials. Besides this, low-amplitude EEG is observed in both temporal areas. Alternatively, the EEG may be recorded with any scalp electrode as a reference, and then the average reference is computed as a mean of all electrodes. It avoids all kind of asymmetry and makes the EEG recorded in various laboratories comparable. But in some cases using the common reference may reveal rhythms not at their actual location. Sometimes the so called bipolar recording is used when the potential is measured between two active electrodes. This sheme is good for exact location of some locale potential changes, i.e., pathological activity focus. The comprehensive review of reference problem may be found in Lehmann (1987).
Parameters for computerized EEG acquisition and storage. For acquisition and storage of EEG data “IFCN Standards” recommend a minimum sampling rate of the analog to digital conversion (ADC) of 200 samples/second (Nuwer et al 1998). This rate allows to analyze frequencies up to 100 Hz, as the maximal allowed frequency of the input signal (the Nyquist frequency) should be the half of sampling rate. If the signal is sampled at too low rate, aliasing (falsification of the signal) may occur with unpredictable errors in the digital waveform compared to the original one. Prior to sampling an anti-aliasing low-pass filter must be used. ADC should be done at a resolution of at least 12 bits in order for the EEG to be resolved down to 0.5 µV. Whenever possible, the low-pass filter should be set to 0.16 Hz or less for recording. Routine use of higher settings of this frequency for recording are discouraged, as they should be reserved for specific or difficult clinical recordings only. A 50-60 Hz notch filter should be available, but not routinely used. Interchannel cross-talk must be less than 1%, i.e., 40 dB down or better.
Calibration. The calibration is needed to determine the exact amplitude of EEG signal and to evaluate the amplifier noise and other possible artifacts produced by it as well as by connection wires. Usually sine, triangle, and rectangle impulses of known amplitude are generated for this purpose by a special circuitry on the input of the main amplifier of the electroencephalograph. The calibration signal thus passes through the large part of the EEG signal’s path in the recording system. The calibration impulses should be recorded and then used to measure the true EEG amplitude and to evaluate the equipment noise in quantitative EEG analysis. The modern software usually includes automatical comparison of EEG and calibration signals showing the actual brain electrical potential values.
Artifacts. EEG artifacts appear due to external electrical or magnetic fields and subjects movements during recording procedure. The last are caused both by muscle electrical potentials fields and electrode displacement. Visual and automatic search of high amplitude artifacts usually is not difficult. For example, eye movement artifacts can be eliminated via special algorithms (Gratton et al 1983). Search and rejection of low amplitude artifacts is possible only by collation of results of frequency analysis, topographical mapping, and original EEG records. Topographical distribution of main artifacts discussed by Lee and Buchsbaum (1987). Eye movements are mainly reflected in frontal sites. Muscle activity is high-frequency and has lateral topography. Artifacts due to bad electrode placement have simple forms and are restricted by given EEG derivation.