Computer methods of analysis of an electroencephalogram

The main methods of computer EEG analysis used in the clinic include spectral analysis using the Fast Fourier Transform algorithm, instantaneous amplitude mapping, spike detection, and 3D localization of an equivalent dipole in the brain space.

The most frequently used spectral analysis. This method allows to determine the absolute power expressed in μV ² for each frequency. The power spectrum diagram for a given epoch represents a two-dimensional image on which the EEG frequencies are plotted along the abscissa axis and the power at the corresponding frequencies along the ordinate. Presented in the form of successive spectra, EEG spectral power data give a pseudo-three-dimensional graph, where the direction along the imaginary axis inward of the figure represents the temporal dynamics of changes in the EEG. Such images are convenient in tracking changes in the EEG in cases of mental disorders or the impact of any factors in time.

Coding color distribution of powers or average amplitudes on the basic ranges on the conditional image of a head or a brain, receive an evident image of their topical representation. It should be emphasized that the mapping method does not provide new information, but only presents it in a different, more visual form.

The definition of the three-dimensional localization of an equivalent dipole is that the mathematical simulation shows the location of a virtual source of potential, which could supposedly create a distribution of electrical fields on the surface of the brain that corresponds to the observed if it is assumed that they are not generated by cortical neurons throughout the brain, the result of passive propagation of the electric field from single sources. In some special cases, these calculated "equivalent sources" coincide with the real ones, which allows, if certain physical and clinical conditions are met, to use this method to clarify the localization of epileptogenic foci in epilepsy.

It should be borne in mind that computer EEG maps display the distribution of electric fields on abstracted head models and therefore can not be perceived as direct images similar to MRI. It is necessary to interpret them intellectually by an EEG specialist in the context of the clinical picture and data from the analysis of the "raw" EEG. Therefore, computer topographic maps sometimes applied to the EEG conclusion are completely useless for the neurologist, and sometimes dangerous in the course of his own attempts at their direct interpretation. According to the recommendations of the International Federation of EEG Societies and Clinical Neurophysiology, all the necessary diagnostic information, obtained mainly on the basis of a direct analysis of the "raw" EEG, should be presented by an EEG specialist in a text that is understandable to the clinician in a textual conclusion. It is inadmissible to provide as a clinical-electroencephalographic conclusion texts that are formulated automatically by computer programs of some electroencephalographs.

To obtain not only illustrative material, but also additional specific diagnostic or prognostic information, it is necessary to use more sophisticated algorithms for research and computer processing of the EEG, statistical methods for estimating data with a set of relevant control groups developed to solve narrowly specialized problems, the presentation of which goes beyond the standard use EEG in a neurological clinic.

[1], [2], [3], [4], [5], [6], [7]

General patterns

EEG tasks in neurological practice are as follows:

1. a statement of brain damage,
2. determination of the nature and localization of pathological changes,
3. evaluation of the dynamics of the state.

Explicit pathological activity on the EEG is a reliable evidence of the pathological functioning of the brain. Pathological fluctuations are associated with the current pathological process. With residual disorders, changes in the EEG may be absent, despite a significant clinical deficit. One of the main aspects of the diagnostic use of the EEG is the determination of the localization of the pathological process.

• Diffuse brain damage caused by inflammatory disease, discirculatory, metabolic, toxic disorders leads to diffuse EEG changes, respectively. They are manifested by polyrhythmia, disorganization and diffuse pathological activity. Polyrhythmia is the absence of a regular dominant rhythm and the predominance of polymorphic activity. Disorganization of the EEG - the disappearance of the characteristic gradient of the amplitudes of normal rhythms, the violation of symmetry. Diffuse pathological activity is represented by delta, theta, epileptiform activity. The picture of polyrhythmia is caused by a random combination of different types of normal and pathological activity. The main feature of diffuse changes, in contrast to focal ones, is the lack of constant locality and stable asymmetry of activity in the EEG.
• The defeat or dysfunction of the midbrain structures of the large brain involving non-specific ascending projections is manifested by bilateral-synchronous bursts of slow waves or epileptiform activity, with the probability of appearance and severity of slow pathological bilateral-synchronous activity being greater the higher the neural axis is the lesion. So, even with a severe lesion of bulbopontin structures, the EEG in most cases remains within the norm. In some cases, due to damage at this level of the nonspecific synchronizing reticular formation, desynchronization occurs and, accordingly, a low-amplitude EEG. Since such EEG are observed in 5-15% of healthy adults, they should be considered as conditionally pathological. Only a small number of patients with lesions at the lower barrel level observe outbreaks of bilaterally synchronous high-amplitude alpha or slow waves. In the case of lesions on the mesencephalic and diencephalic level, as well as higher lying median structures of the large brain: the cingulate gyrus, corpus callosum, orbital cortex, bilaterally synchronous high-amplitude delta and theta waves are observed on the EEG.
• With lateral lesions in the depth of the hemisphere due to the wide projection of deep structures on the vast regions of the brain, the pathological delta and theta activity, respectively, is widely spread along the hemisphere. Because of the direct influence of the medial pathological process on the median structures and the involvement of symmetrical structures in the healthy hemisphere, bilateral and synchronous slow oscillations, predominant in amplitude on the side of the lesion, also appear.
• Surface location of the lesion causes a local change in electrical activity, limited to a zone of neurons immediately adjacent to the focal point of destruction. The changes are manifested by slow activity, the severity of which depends on the severity of the lesion. Epileptic excitation is manifested by local epileptiform activity.