EEG is the main type of instrumental diagnostics in modern psychiatry
In 1929, Hans Berger was the first to record the electrical potentials of the brain from the surface of the human head as non-invasive, i.e. bloodless electrodes – this is how a new scientific and practical diagnostic direction appeared – electroencephalography.
Interestingly, both G. Berger and his followers immediately tried to use the EEG of the brain to search for the causes of various mental disorders – for this they registered the EEG in mentally ill and tried to find a particular mental illness in the features of the EEG “pattern”. The search for EEG signatures of schizophrenia, depression and many other diseases continues to this day, but back in 2013, the US National Institute of Mental Health (NIMH), the main grant holder of scientific research in psychiatry in the world, recognized such a “frontal” approach as inadequate from a scientific point of view view and announced the termination of funding for this kind of research.
In 2017, depression was recognized as a major factor in mental illness and disability – according to WHO estimates, up to 500 million people worldwide suffer from this disorder. Statistical reports cite colossal losses associated with disability due to depression. However, the problems with diagnosing and treating depression are not only not improving, but in general they are becoming more complex and confusing. These difficulties begin already in the diagnostic manuals and textbooks themselves (DSM-5, ICD-10), in which depression is present both in the form of an independent disorder, and in the form of “other, indefinite, comorbid, etc.” types of depression. One of the reasons for this state of affairs is the lack of clear objective diagnostic criteria. And this applies not only to depression, but to almost any mental disorder.
Despite the emergence of more and more powerful diagnostic methods, such as positron emission tomography (PET), nuclear magnetic tomography (MRI), functional MRI, torsion tomography, etc., interest in traditional electroencephalography (EEG) is not fading away, but more and more, and thanks to such properties as non-invasiveness, relative cheapness, high temporal resolution and “hybridization” with computer technologies, as a result of which a modern computer modification of the EEG has emerged, has become a leader in wide everyday practice. This was also facilitated by numerous scientific fundamental studies, as a result of which the modern understanding of abnormal patterns of electrical activity of the brain in various mental diseases – endophenotypes or EEG phenotypes – was formed.
What is a computer EEG technique, what procedures does it involve and how does it differ from the traditional paper EEG, with which it all began?
In a fundamental sense, an EEG is a recording of fluctuations in electrical potential from the surface of the head that a huge number of nerve cells in the brain generate during their work. These potentials are very weak; therefore, special equipment is required to register them. We are talking about special amplifiers of electrical signals, which are the main link of the so-called electroencephalographs. Electrodes are connected to the inputs of these amplifiers, which are installed on the scalp – it is they that capture the weak electrical signals that cortical neurons generate.
From the electrodes, the neural signal enters the amplifier, amplifies, and then its fate is different depending on the generation of encephalographs. In the pre-computer era, this signal was simply recorded on paper in the form of an ordinary EEG, and this was practically the end of it – the clinician simply looked through kilometers of such paper tapes from the recorded EEG and, based on his knowledge, made conclusions and formulated clinical conclusions.
With the advent of computers, the fate of the neural EEG signal has changed a lot – in order to enter it into a computer, it is first converted from analog to digital form (ADC or analog-to-digital converter), and then, in the form of a sequence of zeroes and ones, is entered into the computer for further processing. The types of subsequent computer processing of the digitized EEG are constantly developing and improving, and today, among them, one can distinguish several types of hierarchically built on top of each other techniques that ensure the completeness and depth of analysis.
First of all, these are the most basic computer techniques that essentially duplicate some of the important functions of analog electroencephalographs – for example, the output and drawing of an EEG curve on a computer monitor screen. But even in this case, the use of a computer significantly enhanced the capabilities of this stage – it became possible to measure various EEG parameters (amplitude, latency, etc.) almost instantly for any set of channels.
Further, due to the fact that, in fact, a digital EEG recording is, in the mathematical sense, a time series, it became possible to apply a large number of different algorithms of mathematical analysis to the processing of this series. Firstly, this is an interpolation procedure, the application of which made it possible to obtain a more compact representation of the information contained in the EEG – these are the so-called topographic maps. Its essence is as follows – the program, using a movable cursor, measures the potential amplitudes synchronously across all EEG channels at a certain point in time, enters the measured values into the “nodes” of the map corresponding to the electrode overlap points, then, using interpolation algorithms, calculates the theoretical values of the potential amplitude for all others map points, then applies black-and-white or color coding to the resulting array of values, and we get topographic maps of the distribution of the amplitude of the electric potential over the surface of the head – top, side, bottom, whatever.
Topographic maps of the EEG amplitude 200 ms and further after the cursor positioned in the edit window.
The next improvement in computer analysis of EEG is associated with the use of the so-called. spectral analysis using fast Fourier transform – the EEG time series is fed to the input of this algorithm, and at the output we obtain the spectral density function for all frequencies of the EEG spectrum – usually from 1 to 100 Hz. Summing up the spectral density values in the standard ranges of EEG rhythms (1-4 Hz – delta, 4-8 Hz – theta, 8-12 Hz – alpha, 12-36 Hz – beta, 38-42 Hz – gamma) or in any individual, we can represent the severity of various EEG rhythms in the form of the same topographic map as in the case described above with the potential amplitude.
Power spectra for each of the 19 EEG derivations from 0 to 50 Hz.
In addition to spectral analysis, we can apply a large number of other algorithms to the initial EEG time series and obtain, for example, a topographic map of the coherence or synchronicity of potential fluctuations in different areas of the cerebral cortex for different EEG rhythms – on the basis of this indicator, it became possible to talk about the functional connections of various structures of the brain and areas of the cortex. In general, the types of processing of the initial time series are constantly being improved both in qualitative and quantitative aspects.
Topographic maps of the distribution of the spectrum amplitude indicator over the head surface for delta, theta, alpha and beta rhythms. Dots on the maps show the places where the EEG electrodes are applied.
The next stage in the development of computer analysis of EEG is associated with the creation of computer EEG databases both for the norm and for various forms of mental pathology. Key in this area are the names of Bob Thatcher, who developed one of the few currently recognized regulatory frameworks, i.e. an EEG database for almost all ages, and Roy John, who developed the neurometric method and a number of clinical EEG databases.
The essence of this stage is that carefully selected healthy individuals in the age range from 3 months to 83 years without a history of mental and neurological diseases, whose relatives also do not have such a history of diseases, record EEG according to a standardized protocol (leads, rhythms, parameters, functional tests, etc.), individual EEGs are “run” through all processing algorithms, the individual parameters obtained are summed up, averaged and a huge array of various statistical metrics is obtained, which allow statistically comparing the EEG of a single person with this base of norms and obtaining norms for this person topographic map of the statistical indicator of the reliability of the deviation from the norm of any EEG rhythm and any of its parameters. We can say that this is one of the most recent achievements of computer analysis of EEG and reliable programs in this area have been developed relatively recently.
Topographic maps of Z-ratios for absolute and relative powers of delta, theta, alpha, beta and high-frequency beta EEG of a patient with generalized anxiety disorder. The color shows the degree of statistical deviation from the norm (red – above the norm, blue – below the norm).
Currently, we are witnessing an even more complex stage in the development of computer analysis of EEG associated with the development of expert systems using artificial intelligence algorithms to automate the process of describing and diagnosing clinical EEG. However, this stage is still far from complete and requires joint action by a large number of scientific and clinical teams. One of the tasks of this stage is to identify stable abnormal patterns of brain electrical activity in various mental diseases – endophenotypes or EEG phenotypes.
It is the position on endophenotypes and biomarkers that underlies the future objective classification of mental disorders, the foundations of which are being actively developed at the present time. To date, 14 different EEG phenotypic patterns have been described in the literature, which occur in various mental illnesses and reflect disorders of certain brain systems. It is the endophenotypes, according to modern concepts, that are an intermediary link between the genotype and a specific mental illness. However, we will postpone a more detailed consideration of this issue until a future publication on the website of our Center.