# Method for studying human's and animal's electroencephalograms

FIELD: medicine.

SUBSTANCE: real-time and time-delay EEGs are recorded and processed by continuous wavelet decomposition. Scalograms of a wavelet coefficient matrix are drawn. That is followed by the further analysis of chains of local minimums and maximums on the successively drawn scalograms. The chains of the local minimums and maximums are formed. Basic chain parameters required for the further analysis are determined. The chains are typed according both to frequency and energy parameters. The averaged chains of the local minimums and maximums are derived. The number of chains is used to construct frequency- and energy-based incidence cross tables to be statistically processed; the incidence can fall within the range of an absolute value, and be rated by the common number of chains in the tables, the common number in the respective line, and the common number in the respective column.

EFFECT: invention enables providing the higher accuracy and information value of the EEG analysis in various functional conditions by detecting the fine structure of the local minimums and maximums carrying the information of respective brain centres activity.

15 dwg, 2 ex

The invention relates to medicine, namely to functional diagnosis of normal and pathological physiology.

There is a method of studying the EEG of humans and animals (Patent No. 2332160 Registered in the Russian Agency for patents and trademarks on August 27, 2008), based on registration of EEG and further spectral analysis using the continuous wavelet transform, which determine the capacity of the frequency kardiointervalogrammy a at time b by the formula:

, a, b∈R, and>0,

where W(*a*b) - coefficient wavelet transform;

f (t) is the analyzing function;

ψ((t-b)/*a*) is the analyzing wavelet;

building on the basis of the matrix of wavelet coefficients scalogram on the interval [b_{i}b_{j}] by the formula:

, i, j<N, j>i

where V(*a*_{l}) - salagrama signal;

N - number of coefficients;

*a*_{l}- scale wavelet transform;

selection on salagrama physiologically relevant frequency bands on the basis of distances between neighboring local minima on the curve scalogram by the formula:

where Δ*a*- physiologically meaningful range,*a*_{m},*a*_{n}- neighboring local minima on the curve scalogram;

ODA is the division of the values of the wavelet power density (VPM) U in each of the frequency bands according to the formula:

,

the definition changes wavelet power density in time as U(t);

the determining the change in frequency bands in the time Δ*a*(t);

determining the value of the specific wavelet power density U' in time according to the formula:

U'=U(t)/Δ*a*(t)

which reflects the dynamics of change in activity of the various generators of the EEG for short periods of time.

Scalogram ("energy" diagrams) are based on a matrix of wavelet coefficients, defined as the average of the squares of the coefficients W(*a*b) for a fixed parameter*and*on the interval [b_{i}b_{j}]. As a function of scale, salagrama reflects the same information as the spectral power density of the Fourier transform, which is a function of the frequency. As you know, the wavelet transform has the advantage, first of all, due to the properties of time-frequency localization of wavelets. The wavelet transform, which represents the time base range, allows to obtain more localized in time energy information. Energy diagrams (scalogram) are based on short-term (on the order of 0.01 seconds with a sampling rate of 5000 Hz) intervals, allowing you to track the temporal dynamics of the process.

On salagrama allocate local spectra and f is zoologichesky significant frequency bands Δ*
a*that is calculated from the distance between the local minima*a*_{m},*a*_{n}associated with different types of regulatory mechanisms In CP man. When identifying the three most important ranges are determined by the two most pronounced minimum, four - three, and so on

The total value of the wavelet power density U reflects the total activity of the nerve centre and is defined in each of the frequency bands.

Specific wavelet power density U' characterizes the specific expression of active nerve centre and reflects the processes of change over time in the contribution of the nervous centres, generating certain frequencies in the overall picture of the EEG. The selection of physiologically significant range between the local minima on the curve scalogram associated with different EEG rhythms, the evaluation of this parameter allow to detect even small amplitude rhythmic activity at different stages of ontogenesis, in norm and pathology, both at rest and during transients that quality improves informative and accurate way of assessing EEG. Introduction temporal assessment of specific wavelet power density allows to describe the dynamics of activity generators of different EEG rhythms alone for short periods of time.

The disadvantages of this method include continuous assessment of the structure of local minima, so that the resulting frequency space "breaks" artificially combined. Indeed, on the basis of Fig.1 shows that the structure of local minima (light gray lines in the background of dark gray fill) in some cases artificially connected to other structures through the area local maxima (light-grey). In other words, there are no criteria by which to determine the occurrence of a chain of local minima, join this chain sequence of several local minima, and hence the calculated frequency ranges are not always correct.Known PikWave (registration No. 2006613500), providing analysis of local minima and local maxima, scalogram by the formula (2), built on the matrix W(*a*b). The program implements the calculations and build their chains of local maxima and minima and the calculation of the values of the wavelet power density in dynamic ranges. The disadvantages of this program should be considered, as in the case of a patent, as the presence of artificial compounds chains local minima, and a similar algorithm for chains of local maxima. Check that the abbreviation CL bude order to identify local maxima chain,
and chain local minima - Clmin, unless otherwise specified. However, the settings for this program allow you to get the value chains of the local maxima and minima in the form of their true coordinates on the matrix of wavelet coefficients, and their coordinates in the time-frequency plane, smoothed by the moving average method. This will be differences between estimated and true position of the local extrema of wavelet spectrum.

Also known article "Age-related changes in the time-frequency patterns of sleep spindles on EEG in rats with a genetic predisposition to absent-epilepsy (line wag/rij)" E. Y. Sitnikov, V. C. Grubb, A. E. Khramov, A. A. Koronovskii, Journal of higher nervous activity. 20/2, volume 62. No. 6, S. 733-744. In the article there was no mention of local minima, and in the case of local maxima obvious mistake is the lack of rules that define the beginning and the end of the chain, which they form. Indeed, as seen in figure 1 of this article, the calculation includes (and even shown on the graph) not all chains, but only that its time coordinate is in the area of sleep spindles. This will repeat: the rest of the chain is not even rendered. Thus, there may be situations where existing CLM, which analyze the authors, represents samoupravlenie started much earlier CLM. Similar situation with the termination CLM. It is obvious that not only the lack of rules, which is the interrupt CLM, but even primitive rendering all CLM this matrix of squared wavelet coefficients makes it impossible to evaluate the correctness of their construction. Obviously, the authors did not think of these rules, otherwise they would have been given. In addition, the article takes into account the average frequency and start and end frequency CLM. Since there are no algorithms for determining the beginning and end SLM, i.e., the criteria by which a local maximum is considered as the beginning point, and the criteria by which it is considered to be its end, then a substantial part in the assessment of the frequency of the local maxima allocated in this way, tends to zero. Not to mention that very little in the article there is mention of different types CLM. The basis for their classification, according to the authors, is only the frequency CLM, it is insensitive to the form CLM. However, the study of the distribution CLM form. Indeed, based on figure 2 of the aforesaid article, you can see that in some cases the distribution of types is studied only at one frequency, and the division into types is not on the minima of the histogram and captures higher values, which does not allow to evaluate this classification as held correctly.

Known is also the program Wavemax 1.0 No. 2012614720. The program in contrast to the above article allows the selection CLM based on rules that are set by the user. Inclusion criteria local maximum or minimum in Maximumof or Lminimum is the distance in frequency and time space, which can be specified as constants, and the dynamic index. However, in the program there are no algorithms for the computation of these values, which requires using suitably qualified user. The program is essentially only a tool and does not contain physiological interpretation of the data.

The task of the invention is to improve the evaluation of the electroencephalogram of man and animals in different functional States.

The technical result consists in increasing the accuracy and informative way to study the EEG of humans and animals in different functional States.

The technical result is achieved by means of investigating the EEG of humans and animals includes registration of EEG and further spectral analysis using the continuous wavelet transform, which determine the power frequency and the electroencephalogram in time b by the formula:

, a, b∈R, and>0,

DG is W(*
a*b) - coefficient wavelet transform;

f (t) is the analyzing function;

ψ((t-b)/*a*) is the analyzing wavelet.

Based on the matrix of wavelet coefficients W(*a*b) you can build an "energy" diagram scalogram V(*a*_{l}signals as the average of the squares of these coefficients at a fixed scale parameter*a*_{l}:

where N is the number of averaged wavelet coefficients W(*a*_{l}b_{k})

In Fig.2 is a plot of the magnitude of the wavelet power density W (*a*b) scale-free*a*wavelet transform for the patient at rest. The dependence of the wavelet power density W^{2}(*a*b) from the scale and the wavelet transform for the patient at rest. Used abstraction Oz, the sampling frequency of the signal is 1 kHz, the build time scalogram 10 MS in steps of 20 MS. Spectra constructed in the following order: solid line, dashed, dotted, and dash-dotted. In fact the signal is set scalogram, V(*a*_{l}).

Can be seen from Fig.2 that different local EEG spectra have different location of the extrema in the frequency space, that from a physiological point of view characterizes the change of the oscillation period of the studied parameters selected by the user. Therefore clicks the zoom, receiving a result of calculations, the number of local maxima on one scalogram (i.e. on the local spectrum), we can estimate the number of oscillators involved in the formation of the EEG signal in a particular abstraction (in fact, this region of the brain) and to determine the frequency of oscillation.

In most currently published work values of extrema W^{2}(*a*b) are calculated at constant scope and wavelet transform, which gives, for example, the ability to calculate the parameter d, which is interpreted N. M. Astafieva as "fractal dimension of the series. However, if calculations are not at a constant scale value*a*and at constant time (b), the distribution of local maxima and minima of the frequency spectrum will change. Obviously, in the General case, the coordinates of the local maxima and minima of W^{2}(*a*b) in space (*a*b) will not coincide with each other.

First of all, we are talking about the fact that using different types of wavelets leads to different frequency resolution of the investigated signal. From the relation (3) implies that if you are using one of the most popular types of wavelets for the analysis of biomedical data - wavelet Morlaix - can with a resolution of less than 1 Hz is sufficient in detail to investigate the EEG signal at frequency f is about sample rate of 1 kHz, from a frequency of 32 Hz, which covers a large part of the frequency EEG. But it is not possible to fully analyze the underexplored high-frequency γ-rhythm, ranging from 35 Hz and above. Along with this is the frequency resolution can be applied only when the computer offline study, because when using wavelets Morlaix impossible to carry out operational wavelet analysis of EEG with a delay of less than 2-3 with, for example, that is a problem when using these wavelets in technology neurocomputer interface or biofeedback.

Hence the necessity of providing better localization in time types of wavelets, however, when this happens the deterioration of the frequency resolution of the signal. So the application of wavelet-type WAVE allows you to reach the indicated level of resolution obtained using wavelet Morlaix, only for frequencies <13 Hz, wavelet MNT - <15 Hz, wavelet DOG-8 - <20 Hz. In other words, waves of electromagnetic oscillations of neural ensemble, the frequencies of which are in the range of the best highlight their specific wavelets, can significantly reduce their amplitude. Therefore, they can remain redetection above or will be assigned to one of the Central frequency, in the which is the wavelet analysis, that will distort the results.

Consider now the process of changing the position of the local maxima of the wavelet power density W^{2}(*a*b) apparent in the analysis of the recorded EEG signal.

In Fig.3,4, where the coordinates of the scale - time presents the picture of the location of the local maxima of the wavelet power density W^{2}(*a*b) when using the discharge of Oz, the sampling frequency of the signal is 1 kHz and the period of building scalogram 1 MS, visible characteristic tree-like structure of this dependence. While for Fig.3 local maxima was built at a fixed time, and Fig.4 - at a fixed frequency (scale).

In Fig.5 (used abstraction Oz, the sampling frequency of the signal is 1 kHz, the period of building scalogram 1 MS) presents an example of the dynamic picture of the behavior of the local maxima of the wavelet power density W^{2}(*a*b) for EEG in the rest state of the patient. As can be seen from the figure, we can distinguish two types of dynamic paintings frequency EEG: "activating" (region B is characterized by an increase in frequency and the emergence of new frequency peaks in scalogram) and "dying" (area And characterized by a reduction in the frequency of the selected oscillator and a decrease in the number of oscillators involved in the formation of the EEG signal). Interpretation of results the Tobit dynamic patterns of behavior of the local maxima of the wavelet power density W^{
2}(*a*b) allows to detect changes in the activity of neural ensembles, forming the EEG signal. An increase in the frequency of the selected oscillator and the emergence of new oscillators (or misalignment already existing ensembles) typical, for example, indicative of the reaction, and the synchronization of their work and reduce the frequency for sleep state.

Consider the highlighted block a in Fig.5. Well seen, that for a certain period of time (~ 20-50 MS) the period of oscillations of EEG waves in some cases remains almost constant, that is, the scale and the wavelet transform corresponding to the peak frequency in the given frequency range of EEG, only slightly drifting in the area of increasing the period of this wave. In the future, should a sharp transition (jump) frequency of the investigated EEG waves at the level of lower frequencies. This happens several times up until the oscillator, producing frequency data, will not disappear.

The use of different variants of cluster analysis allowed us to confirm that the marked points in Fig.5 indeed lie closer to each other than points, marking the local maxima, which are included in other sequence of local maxima on the matrix W^{2}(*a*b) that allows you to interpret them as a single or group of closely related oscillators and not in view of the separate neural ensembles.
Here it is important to note that the process of change and jumps frequencies of EEG rhythms simultaneously captures and the range of β-rhythm, and high-frequency sub-range of the α-rhythm without their division, but this division is traditionally held in research directly on medical issues.

In block B of Fig.5 presents a process reverse to the process noted in block A. it is clear that in the same frequency band (namely, when the frequency change from ≈ 9 Hz to ≈ 35 Hz) is the drift of the peak frequency of the detected oscillator in the high frequency region of the β-rhythm. But the disappearance of the peak, i.e. the cessation of work on close frequencies of groups of nerve cells leads to the appearance of two peaks, the nature of which is currently unknown. The first is in the range of higher frequencies and, as a rule, in the field, owned by the confidence interval around the point of the regression line based on the coordinates of the local maxima of the previous oscillator. The second is in the same frequency range as the previous peak, and closely follows the dynamics: born at lower frequencies, drifting in the direction of high-frequency ranges and in the future disappear after a while. The question of which of the new oscillators is physiological (morph is functional) "successor" of the oscillator, already ceased its operation, requires more detailed biomedical research.

Special attention should be paid to the analysis of γ-rhythm. You may notice that the above changes the frequency coordinates of the peaks of the oscillators are observed much less frequently in the frequency range of γ-rhythm compared with the lower frequency ranges, because of the much lower resolution of the used wavelet in frequency at higher frequencies. However, the presence of frequency drift in the direction of decrease and increase shows that the frequency peaks of γ-rhythm associated with cortical brain activity and cognitive functions vary considerably stronger than for α - and β-rhythms. So, for γ-rhythm change in the peak frequency of the EEG within 1-2 MS is ≈ 5-15 Hz, while in the lower frequency bands such changes do not exceed ≈ 2-5 Hz.

Let's formalize the rules for determining the local maxima and minima. The key moment in the formation of such chains or interruption are the rules by which a new coordinate point (*a*b) corresponding to the local maximum/minimum at scalogram, is included or not in the existing CLM.

To determine how formed CLM, that is, in fact, to get a picture of clustering local Maxim the MOU/lows was created and tested a number of algorithms, using the values of scale and time wavelet transform, changing with a constant pitch, or dynamically, i.e. scale-dependent generation of these parameters.

In Fig.6 (a chain of local maxima are determined by the dynamic rules (round black markers), for xed values of the scale (diamond black markers) and time (square light markers) shows the dependence between the values of*a*and the ratio Δt/k (the difference between end time and start CLM or Clmin in the case of constructing chains of lows to the number of points) for wavelet Morlaix, obtained using the optimal algorithm for a decision on inclusion in CLM or Clmin specific points that characterize the time-frequency location of the local maximum. As can be seen from Fig.6, when the dynamic creation of conditions for inclusion of a point in CL (round black markers) or Clin analyzed with respect to the scale of the wavelet transform (frequency decreases) rapidly approaches unity, which reflects the lack of CLM "drop points". Such points occur when combining two different CLM or Clmin, between which there are no local maxima/minima, located in the vicinity of this value, the time scale is of tschechow.
Fixed values of the surroundings of the end point CLM, which should be the next point CLM presented square transparent markers (region scales of the wavelet transform ±10 values if scale*a*endpoint CLM was less than 10, then the range of scales of the wavelet transform were examined from 0 to*a*+10, area time: +10 values) and diamond-shaped black markers (region scales of the wavelet transform ±50 values, scope, time +50 values). Seen from Fig.6 that if in the first case, the dependence is only slightly inferior to the dependence obtained on the basis of dynamically generated rules enable local maxima in SLM that is somewhat large scale*a*where the ratio Δt/k reaches its maximum value, while in the second case CLM are incorrectly merged local maxima, i.e. CLM separated by "empty" areas that do not contain in the analyzed time points of local maxima that can be clearly seen at low values of the ratio Δt/k. It is important to note that similar dependencies were obtained for local minima.

During the analysis of the 64 EEG, registered on 21-th channel with a sampling rate of 1 kHz per channel both at rest and during photostimulation, when addressing a person clicks the knowledge and logical problems,
and the neurocomputer interfaces implemented technology SSVEP and R, received the rule of maximization of Δt/k. The essence of this rule is that it allows, on the one hand, to highlight CLM or Clmin, and to avoid excessive associations such CLM or Clmin, which between them are observed frequency-time region without local maxima/minima. Using this rule, the expression for the Boolean function f(*a*b) a decision on the inclusion (or exclusion) of the local extremum with coordinates (*a*_{i}b_{i}in particular CL or Clmin:

where*a*_{i-1}- scale local maximum/minimum end point CLM or Clmin, the nearest on the magnitude scale is intended for inclusion in CLM or Clmin local maximum/minimum; b_{j-1}the time (or sample number, number scalogram) end point CLM or Clmin nearest time (count number, number scalogram) to the same value proposed for inclusion in CLM or Clmin local maximum/minimum;*a*_{s}is the scale value of the first point emerging CLM or Clmin. Constants u and ν obtained by using the rule of maximization of Δt/k and at most for this scale range is meant the th k.
For wavelets Morlaix and WAVE u=3 and ν=0,05.

Received so many CLM or Clmin can be subjected to further processing in order to identify them in significant elements. It is advisable to use for this purpose the following characteristics CLM or Clmin: the scale of the wavelet transform*a*(or the corresponding frequency), which appeared (*a*_{s}) or interrupted (*a*_{f}CLM or Clmin; the time of occurrence of t_{s}and termination of t_{f}CLM or Clmin; duration CLM or Clmin Δt=t_{f}-t_{s}; "drift" scale Δ*a*=*a*_{s}-a_{f}; number k local maxima/minima in CLM or Clmin; ratio Δt/k, reflecting the density of local maxima/minima in CLM or Clmin.

Consider the results of further analysis of the obtained CLM or Clmin.

The method is based on the paradigm based on the use of ideology allocation evoked potentials (EP) of the brain. As you know, cars are low-amplitude oscillations of the electric field of the brain that occur in response to certain stimuli (flash of light, a sharp sound, a click, tactile irritation, etc. or prior to certain actions (e.g., muscle contraction). Since the amplitude of the EAP is significantly lower amplitudes for the background EEG, in the proposed method for lucchinetti signal/noise is used coherent accumulation of fragments of EEG, associated with the target stimulus, which allows a detailed analysis of the EAP.

The study is averaging in the time-frequency space frequency local maxima present in this CLM or Clmin, to obtain C_{i,k}- coordinate values averaged CLM or Clmin i-th type:

where*a*_{i,j,k}the frequency value of the local maximum; i=1, ..., 5 - type CLM or Clmin; j=1, ..., n_{i}- the number of chains in the array CLM or Clmin of this type i (non chains are installed in accordance with their length, the smaller number corresponds to a greater length, the total number of chains of type i is equal to n_{i}); k=1, ..., m_{i}- number of points in a particular CL or Clmin type i; m_{i}- the number of pixels in the longest CLM or Clmin i-th type.

An important aspect of using this method is the choice of the starting point for averaging CLM or Clmin both in frequency and in time.

To solve this problem in a temporary space was used two approaches.

In the first, the averaging is performed after combining the first points CLM or Clmin, so that the times of occurrence of the first local maxima of t_{s1}, t_{s2}, ..., t_{sn}will be equal to zero, where t_{s1}the time (or sample number) of the first local maximum of the first C is M or Clmin;
t_{s2}the time (or sample number) of the first local maximum of the second SLM or Clmin; t_{sn}the time (or sample number) of the occurrence of the local maximum wavelet spectrum, which form the first point of the n-th chain used in the calculations. Then synchronize all coordinates CLM or Clmin in a temporary space, which consists in the fact that the calculated average values of the frequencies for the average CLM or Clmin. If one CLM or Clmin missing the point with certain time coordinates (Fig.7 this case is indicated by the arrow), the total summation for a given temporal coordinates such CLM or Clmin not involved. Therefore, here the averaging is performed only on the points obtained during calculation CLM or Clmin, without regard to "artificially" constructed points, for example, in the approximation or interpolation, which can fill areas CLM or Clmin without local maximums.

The second approach uses the function U_{i}characterizing the degree of difference of the chains in the frequency space:

r_{i}- the number of points in the current CLM or Clmin i-th type. The other designations are the same as in the formula (4).

You want to find an integer l, where the value of the function U_{i}is the least. The minimum of the function U_{j}where l_{j}- the value of l corresponding to the minimum U_{i}chain of the i-th type with number j.

It is easy to see that using this approach, in which CLM or Clmin before subsequent averaging lining up, moving to a temporary space by changing all time coordinates CLM by adding thereto the constants l, allows to minimize in frequency space differences this CLM or Clmin from CLM or Clmin with the maximum number of local maxima/minima. It should be note that the minimization of these differences is the use of the scale values, measured in Hz, not in the scale of the wavelet transform, as in this case, the data obtained for different wavelets for the same signal can vary considerably. Moreover, using exactly the frequency space and knowing the resolution of different wavelets for certain scales of the wavelet transform of the signal, can be most correctly in order to ecretariat classified data.

The implementation of these methods of averaging CLM or Clmin in the above form faces, however, a number of problems. In particular, the length of CLM or Clmin as a temporary space, and characterized by the number of points depends on the frequency scale of the wavelet transform). Thus, the summation of the two CLM or Clmin, appeared on the EEG frequencies, such as 3 and 35 Hz, is meaningless not only from a physiological point of view, but also from the point of view of the numerical processing of the signal. To resolve this problem was used division into sub-bands, the whole investigated frequency range of the EEG: the lower boundary of the investigated range was set equal to the lower bound of θ-rhythm, and the upper half of the Nyquist frequency for the electroencephalograph. You can use this separation as in the framework of the "classical" variant of EEG analysis, and by building sub-bands, based on the dynamic characteristics of their borders, based on the frequency values of the local minima, which are similar to maxima CLM or lows Clmin also change their frequency characteristics.

Another problem that arises when receiving average CLM or Clmin method of coherent accumulation, associated with their very different dynamic behavior in time.

As can be seen from Fig.7 (local distribution is of maximumof in the matrix of the squares of the coefficients of the wavelet transform W^{
2}(*a*b) EEG signals based on scalogram; abstraction Oz, the sampling frequency of the signal is 1 kHz, the horizontal axis represents time in MS, the vertical axis is the scale of the wavelet transform; the arrow indicates the presence of a "gap" in CLM; region: A - CLM first; B second; third; fourth; D - the fifth type), we can distinguish five different types of behavior CLM or Clmin:

1) CLM or Clmin with steadily increasing frequency (Fig.7, area A). Such CLM or Clmin largely characteristic EEG at rest. The pattern of increased frequency of local maximum/minimum for a variety of occasions here is different, that may allow you to select multiple subtypes of this type of behavior CLM or Clmin. Behavior CLM or Clmin in frequency space varies as the speed ν of the drift frequencies (ν=ω/t, where ω is the frequency; t is time; v is measured in Hz/s), and form the resulting CLM in space time - scale wavelet transform. There are the following types of dynamics CLM or Clmin: (a) uniform linear (when the dependence of ν(t) is well approximated by a linear function); b) uneven (when trends built on the beginning and end CLM or Clmin differ from its Central part).

2) CLM or Clmin with steadily decreasing frequency (Fig.7, area B). As for CL or C the Min of the first type, the nature of the reduced frequency for CLM of this type are quite different. Perhaps the "steadily decreasing type" represents the reverse compared to the type 1 variant functioning oscillators for modulation of electrical activity in the brain that generate behavior CLM or Clmin of the first type.

3) CLM or Clmin without changes in frequency space (Fig.7, region). The oscillator or oscillators) the brain does not increase and does not reduce its frequency within the resolution of the used wavelet.

4) CLM or Clmin, in which first an increase and then a decrease, though CLM or Clmin can be completed both at higher and lower compared to the initial frequency (Fig.7, region D).

5) CLM or Clmin showing an initial decrease in frequency with subsequent growth. As for CLM or Clmin previous type, the ratio of the frequencies of commencement and completion CLM or Clmin may be different (Fig.7, region D).

The above typology of local maxima/minima of the matrix of wavelet coefficients for EEG is important for investigation of EEG in different functional States of the person.

Will now demonstrate the possibilities of the method as applied not to the frequency, and the energy is some component CLM or Clmin.

For correct averaging of the amplitude parameters CLM or Clmin proposed three approaches, using different selection criteria space points (W^{2}(*a*b)) for synchronization CLM (Fig.8 is a schematic illustration of the three approaches to the interposition CLM b-axis (time, local spectrum) when calculating the averaged data on energy CLM; axis OX deferred value b, reflecting the process, along the axis OY - W^{2}(*a*b) characterizing the energy of the process occurring in a given frequency range of the EEG; And - averaging when synchronizing for the first value CL; B - average in synchronization while minimizing the difference between the index W^{2}(*a*b); - averaging when the synchronization indicatorsfor two CLM).

In the first approach (see Fig.8A) reference point for averaging is the first time the value in CL or Clmin, while the rest of the values W^{2}(*a*b) in CLM or Clmin are arranged along the b-axis so that their sequence numbers of samples in different CLM or Zlin match, and only after this is accomplished through averaging. This approach is similar to our approach to averaging CLM or Clmin the scale parameter (frequency) wavelet transform.

The second approach (Fig.8B) is based on minimizing the dif the th CLM or Clmin by the values of W^{
2}(*a*b) in CLM or Clmin. One CLM shifted relative to the other along the axis b until, while between them are minimal difference between the energies of two CLM or Clmin:

where W^{2}(*a*b) is the local energy of the signal at time b at a frequency of*a*; r_{i}- the number of points in the current CL i-th type; i=1, ..., 5 - type CLM or Clmin; j=1, ..., n_{i}- the number of chains in the array CLM or Clmin of this type i (non chains are installed in accordance with their length, the smaller number corresponds to a greater length, the total number of chains of type i is equal to n_{i}); k=1, ..., m_{i}- number of points in a particular CL or Clmin type i; m_{i}- the number of pixels in the longest CLM or Clmin i-th type.

The second approach, as well as the first, conceptually identical to the used in our studies when assessing the dynamics CLM or Clmin in frequency space. In this case, the characteristics of the W^{2}(*a*b) use coordinates*and*local maxima CLM or Clmin.

The third approach (Fig.8B) is based on the idea of the quantities W^{2}(*a*b) how about showing energy, and ultimately the signal amplitude at a given frequency, which is the frequency of the local maximum. This allows us to offer another rule of averaging: CLM or Clmin synchronized so what Braz,
differencesandwere minimal, where- the maximum energy of the first CLM or Clmin type i andthe maximum energy of the second SLM or Clmin type i.

Obviously, when the apparent similarity of the second and third approaches are fundamentally different. The second approach minimizes the difference between the energies of two CLM or Clmin during the whole period of their existence. Thus, it is searched such mutual arrangement CLM or Clmin on the axis b, in which the total energy of the two processes differ minimally. In the third approach, the differences are minimized only for segments CLM or Clmin with maximum energies, the difference in total energy of the processes may not be the lowest possible. Obviously, the second approach has the advantage to processes relatively little change over time. Such, for example, can be attributed to the dynamics of the EEG in a relaxed state of the person or for the state with eyes closed.

The third approach is the most optimal for the analysis of distinct nestazionarnosty on the curve EEG, when the analyzed signal abruptly changes its properties. Besides simple example epilepsy, generating the EEG, in some cases, a characteristic structure of peak - wave" can be mentioned about the phenomenon of synchronization and desi the chronicity of the EEG signal, at which frequency the painting process changes dramatically.

Obviously, as in the case of the analysis of the dynamics of the frequencies of the local maxima/minima in time, you can see that the dynamic energy of the process can also greatly vary. Summation CLM or Clmin significantly different temporal dynamics of energy, can lead to incorrect and uninterpreted results. In this regard, an attempt was made to divide many CLM or Clmin into several types, in which the variance values W^{2}(*a*b) calculated between CLM or Clmin related to this type, would be minimal, despite the fact that the types could be physiologically meaningful interpretation (according to the rule of minimizing the variance within a cluster).

In light of the above identified five types of dynamics of energy within CLM or Clmin:

1. Steadily rising energy CLM or Clmin. It can be assumed that the oscillator or group of oscillators that generate this type CLM or Clmin, significantly increases over time their activity, which is reflected in the increase in the signal amplitude of this frequency. This type CLM verified, for example, in the reactions of sync when switching from the high-frequency beta rhythm to the lower frequency of α-rhythm. In the range of the last energy signal naturally the military increases.

2. Steadily decreasing in energy CLM or Clmin. By analogy with type 1 character descending values W^{2}(*a*b) a local maximum is associated with the inverse process - the desynchronization of the EEG. In the moment of transition from the dominant α-rhythm is relatively synchronized in occipital leads to the dominance of desynchronizing β-rhythm is consistent with the fall of the signal energy in the range of α-rhythm.

3. CLM or Clmin, almost no izmeneniia energy. The oscillator or oscillators) in the brain does not increase and does not reduce its "power", i.e. does not change its contribution to the signal amplitude. Taking into account features of the wavelet transform of the EEG associated with high-resolution signal amplitude, this type CLM when processing native EEG was not recorded. He appears after the artificial introduction of so-called "zero-corridor - range of variance values W^{2}(*a*b) how in the direction of increasing and decreasing, it is assumed that the initial and final energy CLM has not changed.

4. CLM or Clmin, showing the first increase in the signal energy, and then its decrease, despite the fact that it can be completed both at higher and lower energy relative to the initial one.

5. CLM or Clmin, demonstrating the decrease in energy is its subsequent growth. As for the previous type, the ratio of the energies of commencement and completion CLM may be different.

Types SLM 4-5 more often met in the signals registered with the active mental work (when solving a logical or expressive tasks).

Separate type are CL or Clmin consisting of only one value that is not assigned to one of the existing CLM or Clmin.

Considering the interpretation of the dynamics CLM in amplitude space for these chains, it is necessary to take into account that not always the presence of high values of energy on this scale (frequency) indicates the presence in the signal contribution from operating on the frequency of the oscillator. By analogy with the Fourier transform, for example, for amplitude-modulated sinusoid possibly different distribution of the energy peaks of the local spectra, which can lead to incorrect interpretation of results.

It is important to determine how the distribution of the total number CLM or Clmin on the basis of classifications (both in frequency and in the energy space), by analyzing the reactions of the EEG to the above function.

To answer this question were obtained and used contingency table for each assignment and group experiments on the types CLM or Clmin often in the nome and energy spaces (see table.1).

Table 1. Cross-tabulation of the total number CLM. Use abstraction C3. The subject faces challenges in creative thinking, in the experiment he chose the answer that corresponds to click the right mouse button. Designating types CLM: table rows for the dynamics in the frequency space, the columns for the dynamics of the values W^{2}(*a*b)

Types | |||||

CLM | Type 1 | Type 2 | Type 3 | Type 4 | Type 5 |

Type 1 | 7 | 10 | 0 | 0 | 0 |

Type 2 | 12 | 8 | 0 | 0 | 2 |

Type 3 | 0 | 6 | 0 | 0 | 0 |

Type 4 | 2 | 0 | 0 | 0 | |

Type 5 | 0 | 0 | 0 | 1 | 0 |

Further summation of the tables was based on the study design.

Thus, using the chain analysis of local maxima and minima of different variations greatly expanded opportunities for the study of the electroencephalogram. Increased information content of EEG analysis. The task is fully completed.

EXAMPLE 1

Consider the process of mapping the averaged values of the dynamics of energy Maximumof(hereinafter in example 1 under CLM refers only local maxima) in dependence on one of the 4 described types.

The third type CLM almost never met in our research, for the reasons described above, and therefore it was excluded from the mapping.

For mapping, we have taken the results of registration of EEG in 19 healthy right-handed volunteers. Used EEG "Neuron-Spectrum-4 EP" produced by LLC "Neurosoft". 21 electrode for EEG registration was located according to the standard scheme 10-20. The sampling frequency of 250 Hz was chosen with consideration of the necessary processing with the test data in real-time. As the e load samples were solution imaginative and logical problems, in the course of which the subject by pressing the right or left mouse button was selected one of the two solutions, which are the same as the tasks were displayed on the screen of the LCD monitor 17 inches, located at a distance not closer than 1.5 m from the eyes of the examinee. Analyzed the frequency range from 3 Hz and above. The dynamics of energy CLM was defined as:

where the index n describes the initial, and e is a finite maximum CLM.

It is evident from Fig.9 (averaged dynamics of the energy density ln ΔW^{2}(*a*b) the examinee K. (100 series) depending on the lead (channel) EEG state obtained when showing black screen after solving a logical or imaginative task; axis OH pending lead (channels) EEG, the OY axis - averaged values of the 4 identified types CLM) shows that the highest values of ΔW^{2}(*a*b) correspond to the electrodes Fp1, Fp2, Fpz and reflect artifacts from eye movements - electroculogram. However, when removed from the calculations of these derivations using the non-parametric analogue of the Fisher criterion - criterion Kruskal Wallace showed significant differences (p<0.001) between the mean energy indicators CLM as depending on the type of assignment (channel EEG) and, therefore, from the region of the brain with which to register what was balsa signal,
and the type carried out in the course of the experiment activities. The use of non-parametric criteria, primarily due to the fact that the energy values of the local maxima demonstrated in all series of experiments, the nature of the distribution other than normal. All distribution had a significant excess (>5 and sometimes reaching values of ~ 200) and a pronounced asymmetry, which does not allow to use the methods of parametric statistics even after taking the logarithm of the obtained values of ΔW^{2}(*a*b). It should be noted that CLM with statistically significant lower values were observed in leads T6 and P4, which probably reflects the high proportion of low-amplitude desynchronizing rhythms in these areas, traditionally associated with cognitive human activity.

In Fig.10 (Distribution of ln (ΔW^{2}(*a*b)) CL F. the test leads to C3. Axis OH pending values ln (ΔW^{2}(*a*b)), the OY axis is the percentage of observations CLM included in this column of the histogram of the sum of the total number CLM of all types. The frequency range investigated CLM from 3 Hz to 125 Hz) and Fig.11 (Distribution of ln (ΔW^{2}(*a*b)) for CLM K. for the test leads of OZ. Axis OH pending values ln (ΔW^{2}(*a*b)), the OY axis is the percentage of observations CLM included in this column histogram, from sums the total number CLM of all types.
The frequency range investigated CLM from 3 Hz to 125 Hz) shows the dynamics of behavior CLM for different types depending on the assignment. The results of the evaluation of the distributions showed that for most EEG is characterized by the presence of the distributions of ln (ΔW^{2}(*a*b)), different from normal (p<<0.01 for criteria Shapiro-Wilke, Kolmogorov-Smirnov), primarily due to the high values of kurtosis (much larger 3), reflecting ontroversial distributions. This form of distribution is characteristic for most of the subjects in the Central leads of the international system, the location of EEG electrodes 10-20. This observation can be interpreted as the presence of certain fixed changes in the energy spectra CLM that more often took place in a limited range of values of signal amplitude. Perhaps this mechanism is associated with the activity of the nervous centres, recorded by EEG.

Thus, the dynamics of behavior in time and energy highs, forming a chain of local maxima W^{2}on the plane (*a*b). Similarly averaged over frequency to obtain the characteristic ZLM the proposed smoothing algorithms CLM energy. Shows three ways similar averages, enabling the physiological the automatic meaningful interpretation averaged CLM.
Preliminary mapping of the identified types CLM in relation to a number of problems to be solved subject, it allows to conclude that there is a dependence of nature CLM from the brain and this relationship changes when doing mental work.

EXAMPLE 2

Was formed by a group of 19 subjects of both sexes aged from 19 to 23 years, which was recorded electroencephalogram device "Neuron-spectrum-4 EAP" standard abstraction of 10-20 with a sampling rate of 250 Hz and bit ADC 16 bit. Low sampling rate is explained by the necessity of processing results in real time with minimal time delay, which was implemented on the coprocessor cards for CUDA technology. The subject during the experiment was in a darkened noise dampening chamber. At a distance of 1 to 1.5 m from the eyes of the examinee comfortable housed LCD monitor, describing a set of 100 randomly selected tasks, 50 of which were shaped (for example, selection of two flat pieces by shape and texture), and 50 - logical (finding solutions to equations and inequalities by choosing from 2 options). The average brightness of all the requirements of the test images were equal. The subject, solving the problem, chose one of the two proposed answers, they layout is gales on the same form to the left or right of it. The choice of response was made by pressing mouse button: to answer positioned on the left - the left button; to answer, to the right, - the right button of the mouse when it was in his right hand (all subjects were right-handed). After solving the immediate tasks of the subject within 5 demonstrated a "black screen" monitor, then he faced a new task.

The resulting implementation of the EEG, not containing artifacts were divided into three groups: corresponding to the decision task and pressing the right mouse button; problem solving and pressing the left mouse button; the demonstration of the "black screen" monitor. From the time series were removed U-shaped trends, after which they were subjected to the continuous wavelet transform using the Morlet wavelet. As is known, features of localization in the temporal wavelet domain of this family do not permit their use for analysis of EEG signals in real-time. However, in the described case, the time delay does not exceed the time of the decision subject to the following tasks when there is an accumulation of EEG signal for further processing, which allowed us to analyze the experiment with the time step, not to exceed 2 seconds.

So, during a series of experiments, data were obtained about the structure CLM in the frequency range from 3 to 125 Hz. In the future, may have occurred with maze results of experiments
in crosstabs, based on the design of the study, the comparison of data obtained in the tables is carried out using, for example, the criterion χ^{2}. The data obtained were subjected to classification using hierarchical methods of cluster analysis, with the coordinate axes of the space in which occurred classification, served in the frequency of occurrence of all 25 types CLM (Fig.12. The results of cluster analysis vectors CLM. The abscissa axis is the distance between the centers of the clusters on the y - axis information in the form XXY where XX is the abstraction of the EEG 10-20 system, Y=R - clicking on the right mouse button, Y=L - clicking the left mouse button while solving the problem).

To confirm the stability of the clustering used different rules for evaluating the distance between clusters (near and far neighbor, weighted and unweighted averages of the cluster centers) and different approaches to the assessment measures the distance (Euclidean distance, the method blocks, the Chebyshev measure).

It is evident from Fig.12 shows that when performing tasks requiring choice of two alternatives, vectors types CLM form distinct groups associated not only topographical proximity, but also the electrical activity recorded over sometimes quite far removed from each other by parts of the brain: just look at the part the key tree clustering, marked on the figure by arrows. If you select "right" from definitive answers, diversion, OZ, P4, O2, T4 formed a fairly compact cluster, in the case of the "left" response to the aforesaid leads added leads C3 and T6.

It is important to note that although the absolute values of the frequencies of occurrence CLM different types were differences depending on the assignment, the structure of these frequencies were very close. So most of the time (p<<0.001) were types 1 and 2 CLM as for frequency and energy components CLM, and their combinations (Fig.13. Cumulated frequency of occurrence CLM different types of 19 subjects (1811 experiments) for the discharge of C3 when selecting the "left" of the two proposed answers. Types CLM axis OX represented as "type frequency/type dynamics ΔW^{2}(*a*b)"). Least frequently met combination of the third energy type, because of the high resolution method for this indicator, and therefore, the difficulty of obtaining CLM, in which the start and end values are consistent with each other. Intermediate values between the frequencies of occurrence of the described types are combinations of the dynamics in the frequency and energy spaces, in which at least one element patterns are attributed to chetvert is the fifth type.

It should be noted that in the absence of differences in the duration of the reaction in choosing the "right" or "left" response to the presented task (p>>0.05) the number CLM were significantly different: in the case of choosing the "right" answer number CLM was more than selecting the "left" ((p<<0.01), taking into account the effect of multiple comparisons), for all types of CLM. However, differences in the absolute frequencies of occurrence of different types CLM for different leads formed individual picture for each participant. Analysis of the functional asymmetry of the EEG, which consists in the difference of the indicators to be registered with the symmetric parts of the brain showed using the criterion χ^{2}that for all pairs of leads (C3-C4, O1-O2, Fp1-Fp2, F3-F4, F7-F8, P3-P4, T3-T4, T5-T6) obtained statistically significant differences in the distribution of frequencies of occurrence CLM different types (p<<0.01).

In Fig.14 presents the difference between the number CLM obtained during the processing of EEG with symmetric parts of the skull: the number CLM for the left hemisphere was subtracted number CLM the appropriate type for the right hemisphere. The results obtained are shown in Fig. 14, and the asymmetry for the case of selection of the "left" response is shown in Fig.14a(the Difference in the number CLM registered with the symmetric parts of the brain, when choosing test response, is located in the left part shown the CSO field), and, accordingly, the "right" (Fig.14b. The difference in the number CLM registered with the symmetric parts of the brain, when choosing test response, is located in the right part of the required fields).

The method of study of the electroencephalogram of man and animals, namely, that provide registration of EEG with subsequent processing in real time, and in modes with a time delay by means of the continuous wavelet transform by the formula:

, a, b∈R, a>0

where W(a, b) - coefficient wavelet transform; f(t) is the analyzing function; ψ((t-b)/a) is the analyzing wavelet;

building on the basis of the matrix of wavelet coefficients scalogram on the interval [b_{i}b_{j}] by the formula:

, i, j<N, j>i

where V(a_{l}) - salagrama signal; N is the number of coefficients; a_{l}- scale wavelet conversion;

characterized in that the analysis focuses on the chain of local minima and maxima consistently built scalogram V(a),

the definition of criteria for the formation of chains of local minima (Clmin) and/or maximums (CLM) scalogram based on the criterion of proximity to each other local maxima and/or minima in time and/or the proximity of local maxima and/or local minima by frequency or the scale of the BU wavelet transform and/or "energy" W^{
2}(a, b), taking into account that the maximum is searched for the nearest maximum and the minimum of the local minimum; the criteria can be as constant for the whole matrix W^{2}(a, b) and dynamically change their properties depending on time and/or frequency of observation, following the rule of

maximize Δt/k, i.e. the difference between end time and start CLM and/or Clmin to the number of points obtained for chains of local minima and maxima parameters: the scale of the wavelet transform*a*or the corresponding frequency, which appeared (*a*_{s}) or interrupted (*a*_{f}this CLM and/or Clmin; the time of occurrence of t_{s}and termination of t_{f}CLM and/or Clmin; duration CLM and/or Clin Δt=t_{f}-t_{s}; "drift" scale Δ*a*=*a*_{s}-*a*_{f}; number k local maxima in CLM and/or Clmin; ratio Δt/k, reflecting the density of local maxima in CLM and/or Clmin;

chain local maxima and/or minima typologists, i.e., belong to one of the types for types CLM and/or Clmin, analyzed by frequency or scale, it is:

CLM and/or Clmin with steadily increasing frequency,

CLM and/or Clmin with steadily decreasing,

CLM and/or Clmin without changes in frequency space,

CLM and/or Clmin, in which first happens is an increase in the frequency,
and then its decrease,

CLM and/or Clin showing an initial decrease in frequency with subsequent growth;

for types according to the energy values W^{2}(a, b)

steadily rising energy CLM and/or Clmin,

steadily decreasing in energy CLM and/or Clmin,

CLM and/or Clmin, almost constant energy,

CLM and/or Clmin, showing the first increase in the signal energy, and then its decrease,

CLM and/or Clmin, showing a reduction of energy, with subsequent growth;

table is created cross-tabulate frequency of occurrence types CLM and Clmin frequency and "energy", while the frequency of occurrence can be

both in absolute values and normalized by the total number CLM and Clmin in the table, the total number in the appropriate line, the total number in the appropriate column;

is determined by the average chain, and separately for classes on the frequency or scale, and "energy" W^{2}(a, b), and for chains of local maxima and minima defined by class frequency, using two approaches:

the averaging is performed after combining the first point, so that the times of occurrence of the first local maxima of t_{s1}, t_{s2}, ...t_{sn}will be equal to zero, where t_{s1}the time or sample number of the first local maximum of the first SLM, t_{s2}the time or number is the count of occurrence of the first local maximum of the second SLM;
t_{sn}the time or count number of occurrence of the local maximum wavelet spectrum, which form the first point of the n-th chain used in the calculations;

the second approach uses the function U_{i}characterizing the degree of difference of the chains in the frequency space:

when l=1, 2, ..., m_{i}-r_{i}; j=2, 3, ..., n_{i};

r_{i}- the number of points in the current CL i-th type;

for CLM, defined as classes of energy W^{2}(*a*b) use three approaches:

in the first approach reference point for averaging is the first time the value in CL, while the rest of the values W^{2}(*a*b) in CLM are arranged along the b-axis so that their sequence numbers of samples in different CL match, and only after that is averaging;

the second approach minimizes differences CLM by the values of W^{2}(*a*b) in CLM, with one CLM shifted relative to the other along the axis b until, while between them are minimal difference between the energies of two CLM:_{i}-r_{i}; j=2, 3, ..., n_{i},

where W^{2}(*a*b) is the local energy of the signal at time b at a frequency of*a*; r_{i}- the number of points in the current CL i-th type; i=1, ..., 5 is the IP CLM;
j=1, ..., n_{i}- the number of chains in the array SLM of this type i when this thread numbers are established according to their length, the smaller number corresponds to a greater length, the total number of chains of type i is equal to n_{i}; k=1, ..., m_{i}- number of points in a particular CLM type i; m_{i}- the number of pixels in the longest CL i-th type;

third approach: CLM synchronized so that differencesandwere minimal, where- the maximum energy of the first CLM type i andthe maximum energy of the second SLM type i.

**Same patents:**

FIELD: medicine.

SUBSTANCE: polysomnography is conducted. Slow sleep phases (SSP) and fast sleep phases (FSP) are determined. A maturity index of integrative sleeping apparatuses (MIS) is calculated by formula MIS = SSP/FSP. If the MIS is less than 1.5, a physiologically optimum structure of the nocturnal sleep is stated in a healthy child.

EFFECT: method enables assessing the quality of the nocturnal sleep in the children.

1 tbl, 2 ex

FIELD: medicine.

SUBSTANCE: invention refers to medicine and can be used in labour hygiene and occupational health problems. A digital camera is fixed on a driver's head in front of his eyes and a blinding light source. Crossing coordinates of a display plane and straight lines connecting an eye centre with each blinding source are determined. Blackout areas comparably sized with a head lamp light of the oncoming car are displayed. A maximum contrast negative image of the blinding light sources read out from the camera is presented on a transparent display.

EFFECT: method provides the more effective driver's eyes protection if blinded by the light of the oncoming car that is ensured by displaying the maximum contrast negative image of the segments corresponding to the blinding head lamp light.

4 cl, 2 dwg

FIELD: medicine.

SUBSTANCE: group of inventions refers to medicine and medical equipment. A distance of the upper to lower eyelids of at least one eye is measured over a period of time, Eye openness coefficients varying within the value of wide open eye, through the value of partially open eye to the value of completely closed eye are determined. The eye openness coefficients are graphed. The eye openness coefficients variations over the period of time are compared to a reference eye closure model indicating the microsleep cases. Besides, the method is implemented according to the version, which provides notifying an operator if the microsleep has been detected, by signalling. The method is also implemented by comparing the microsleep models with the eye openness coefficients variations as shown by EEG and EOG. That is ensured by using a device comprising an infrared emitter, which is connected to an image selector. A microprocessor with an electronic procedure of microsleep detection configured to detect face, eyes and eyelids images in a digital image and to calculate the eye openness coefficient with determining a microsleep-specific coefficient, and presenting the obtained information in the form of graphical presentation of the eye openness coefficients at the selected moments of time. A memory unit connected to the microprocess and comprising the reference eye closure models to be compared to the eye openness coefficients at the selected moments of time.

EFFECT: invention enables providing the more reliable assessment of microsleep that is ensured by microsleep detection at the early stages of falling asleep.

28 cl, 6 dwg

FIELD: medicine.

SUBSTANCE: pulse electric activity of sensorimotor central neurons is recorded in experimental animals adapted to hypoxia. The recorded activity frequency is modulated by a multivibrator and an electroacoustic transducer; the signals are copied and transferred onto a carrier. The patient is exposed to distant acoustic signals with the use of the laser generator. The exposure is sequential and starts with sessions at frequency 5-8 Hz for 5-7 minutes and follows with sessions at frequency 10-15 Hz for 5-8 minutes. The sessions are daily, one session a day; the therapeutic course is 10-14 sessions.

EFFECT: method enables using the drug-free modalities and normalise the blood pressure that is ensured by providing the mode and sequence of acoustic signal flow.

2 tbl, 3 ex

FIELD: medicine.

SUBSTANCE: EEG frequency response is pre-determined by a period analysis in a highly-watchful live operator being involved in active visual-motor activities and in actual activities. Theta-, alpha- and gamma-wave count per second is measured. The derived values are compared. If the theta-, alpha- and gamma-wave count per second is stated to fall outside the values specific for high watchfulness, an EEG analyser converts a sequence of the values into a sensory signal and feeds a biofeedback-like warning signal automatically to state the control circuit logoff by the live operator.

EFFECT: method improves the accuracy of the control measurements enabling feeding the sensory signal in due time that is ensured by pre-determining a combination of highly-watchful EEG theta-, alpha- and gamma-wave frequencies.

1 dwg, 2 ex

FIELD: medicine.

SUBSTANCE: recent comatose condition is detected; magnetic resonance imaging shows centres of cerebral structural changes; electroencephalography shows epileptiform activity, diffuse sharp waves, spikes, reduced complexes, high-amplitude low activity paroxysmal events, frequent paroxysmal events of 'peak-slow wave' and 'spike-slow wave' complexes. The EEG also shows irritant activity, local transient high-frequency beta-activity, advanced transient low-amplitude and advance continuous long-term high-amplitude activity. The presence of an aetiological agent of the disease caused by tick-borne encephalitis or the presence of encephalitis of other and uncertain aetiology is stated. The derived data are scored depending on the presence, absence and manifestations thereof. The derived data are used to calculate linear classification functions and to detect the favourable (LCF1) outcome of encephalitis without symptomatic epilepsy (SE) progression and the unfavourable (LCF2) outcome of encephalitis with SE progression. If LCF1>LCF2, the encephalitis outcome without SE progression is predicted, while LCF2>LCF1 shows the unfavourable outcome with SE progression.

EFFECT: method provides more reliable assessment of the SE progression in encephalitis that is ensured by taking into account additional EEG data and calculating the linear classification functions.

2 ex

FIELD: medicine.

SUBSTANCE: invention refers to medicine, namely to neonatology and neurology. During a 300-second slow sleep recording, transitory patterns are recovered on the EEG: frontal sharp waves of an average duration of 0.13 seconds, spike acute waves of an average duration of 0.045 seconds, high-amplitude PTӨ-waves of an average duration of 0.1 seconds, STOP-wave patterns of an average duration of 0.1 seconds. Indices (K) are calculated for each pattern in percentage, as a relation of the number of patterns of 300 seconds multiplied by its average duration to the recording duration (300 seconds). If K of the most patterns determined is less than 1, a physiologic norm is diagnosed; K falling within the range of 1-2 shows a moderate neurophysiologic immaturity; and if K is more than 2, major disturbances of the electrobiologic brain activity are stated.

EFFECT: method enables assessing the electrobiologic brain maturity of newborns of different gestational age that is ensured by determining and considering the transitory age-related patterns in the EEG.

4 dwg, 3 ex

FIELD: medicine.

SUBSTANCE: invention refers to medicine, namely to psychiatrics. In addition to the clinical study, an electroencephalography coherence analysis is conducted within the range of 30-45 Hz before psychotropic agents are prescribed. That is followed by determining middle zonal indices of the left and right hemispheres calculated between the midtemporal and following cortical regions: frontal, central, parietal, occipital, anteriofrontal and posteriofrontal in the homolateral direction. A hemispheric asymmetry factor (HAF) is calculated by dividing the middle zonal index of the left hemisphere by the middle zonal index of the right hemisphere. If the HAF value is less than 1, paranoid schizophrenia is diagnosed; and the value more than 1 shows schizo-affective disorder.

EFFECT: technique enables providing the more reliable differential diagnostics that is ensured by determining the zonal indices of the right and left hemispheres to calculate the hemispheric asymmetry factor.

2 ex

FIELD: medicine.

SUBSTANCE: invention refers to medicine, namely to psychiatrics. A clinical examination is combined with recording an electroencephalogram (EEG); its spectral and coherent analyses are carried out. The following values are determined: T6-AA-lead power spectrum within the range of 3.5-5 Hz, F7-AA-lead power spectrum within the range of 2-3 Hz, T5-AA -lead power spectrum within the range of 23-24.5 Hz, hemispheric power asymmetry between F8-AA and F7-AA-leads within the range of 24.5-26 Hz, P4-C4-lead coherence within the range of 8-13 Hz, T4-F8 within the range of 23-24.5 Hz, T3-F7-lead power spectrum within the range of 26-27.5 Hz, T5-O1 within the range of 17-18.5 Hz, T3-F8 within the range of 20-21.5 Hz. Logarithms of the derived values are calculated. An integral diagnostic value is determined by mathematical formula taking into account the calculated logarithm and coefficient-corrective values. If the integral diagnostic value appears to be positive, a recurrent depressive disorder is diagnosed, while a bipolar affective disorder is shown by the negative integral diagnostic value.

EFFECT: method enables providing higher diagnostic reliability at the early stages of the disease that is ensured by the optimum selection of the values, and the mathematical calculation of the integral value.

2 ex

FIELD: medicine.

SUBSTANCE: invention refers to medicine, labour safety, vocational selection of rescue workers. The invention can be used for vocational selection in the sectors of industry using personal protective equipment, as well as for the workers labour safety in the sectors of industry with harmful working conditions. The method involves vocational selection and duty control on the basis of electroencephalogram (EEG) values and cardiological findings. The examination is performed prior to and when using the personal protective equipment. The cardiological examination involves assessing the heart rate variability with using the amplitude-frequency spectrum Fourier analysis VLF at a vibration frequency within the range of 0.0033-0.04 Hz, LF - at a frequency of 0.05-0.15 Hz and HF - at a frequency of 0.16-0.80 Hz, and is five-staged: initial resting state, mental work load, recovery of mental work load, hyperventilation load, recovery of hyperventilation load. At the beginning, the heart rate variations and EEG are examined prior to using the personal protective equipment. If any of the five stages of the heart rate variation examination shows the pulse more than 90 beats per minute, as well as changes from the normal values of: approximating entropy - less than 180, LF - less than 6 point, an alpha wave amplitude - to 12 vibrations per second and the presence of the paroxysmal activity by EEG, the prevailing sympathetic nervous system is stated, or if any stage of the heart rate variation examination shows the pulse less than 60 beats per minute, as well as changes from the normal values of: blood pressure - more than 140/90 mmHg, VLF - more than 130 points, HF - more than 16 points, an alpha wave amplitude - less than 25 mcV, the prevailing parasympathetic nervous system is stated; a low level of adaptation to the personal protective equipment is predicted, and a rescue work is not recommended during the vocational selection; the examination is terminated. If the heart rate variation and EEG prior to using the personal protective equipment fall within the normal values, the heart rate variation when using the personal protective equipment is started with the patient examined when using the personal protective equipment and performing a cycle ergometer test, and recording the hyperadaptotic changes of the assessed values: VLF - more than 130 points in relation to the normal value when using the personal protective equipment and LF and HF vibrations; an incomplete or unfinished adaptation to the personal protective equipment, and the rescue worker is suspended from work for several hours; if VLF is more than 130 points recorded 10-15 min after activating the personal protective equipment, a good adaptation level to the personal protective equipment is predicted.

EFFECT: method enables assessing the vegetative nervous function and predicting the rescue workers' adaptation level to the personal protective equipment.

11 tbl, 5 ex

FIELD: medicine, neurology, psychopathology, neurosurgery, neurophysiology, experimental neurobiology.

SUBSTANCE: one should simultaneously register electroencephalogram (EEG) to detect the level of constant potential (LCP). At LCP negativization and increased EEG power one should detect depolarizational activation of neurons and enhanced metabolism. At LCP negativization and decreased EEG power - depolarized inhibition of neurons and metabolism suppression. At LCP positivation and increased EEG power - either repolarized or hyperpolarized activation of neurons and enhanced metabolism. At LCP positivation and decreased EEG power - hyperpolarized suppression of neurons and decreased metabolism of nervous tissue. The method enables to correctly detect therapeutic tactics due to simultaneous LCP and EEG registration that enables to differentiate transition from one functional and metabolic state into another.

EFFECT: higher accuracy of diagnostics.

5 dwg, 1 ex, 1 tbl

FIELD: medicine, neurology.

SUBSTANCE: one should establish neurological status, bioelectric cerebral activity, availability of perinatal and ORL pathology in patients, establish their gradations and numerical values followed by calculation of prognostic coefficients F_{1} and F_{2} by the following formulas: F_{1}=-31,42+1,49·a_{1}-2,44·a_{2}+0,2·а_{3}+1,63·a_{4}+0,62·а_{5}+3,75·a_{6}+1,8·а_{7}-3,23·a_{8}-0,8·а_{9}-1,32·а_{10}+3,26·а_{11}+8,92·a_{12}-2,0·a_{13}+3,88·а_{14}+1,79·a_{15}+0,83·a_{16}-2,78·a_{17}; F_{2}=-27,58+1,43·a_{1}+3,31·а_{2}+0,08·а_{3}+3,05·а_{4}-0,27·а_{5}+2,69·а_{6}+3,11·а_{7}-6,47·a_{8}-6,55·a_{9}+1,99·а_{10}+5,25·а_{11}+7,07·a_{12}-0,47·a_{13}+0,13·a_{14}+4,04·a_{15}-1,0·a_{16}-1,14·а_{17}, correspondingly, where a_{1} - patient's age, a_{2} - studying either at the hospital or polyclinic, a_{3} - duration of stationary treatment (in days), a_{4} - unconscious period, a_{5} - terms of hospitalization since the moment of light close craniocerebral trauma, a_{6} - smoking, a_{7} - alcohol misuse, a_{8} - arterial hypertension, a_{9} - amnesia, a_{10} - close craniocerebral trauma in anamnesis, a_{11} - psychoemotional tension, a_{12} - meteolability, a_{13} - cervical osteochondrosis, a_{14} - ORL pathology, a_{15} - availability of perinatal trauma in anamnesis with pronounced hypertension-hydrocephalic syndrome, a_{16} - availability of paroxysmal activity, a_{17} - availability and manifestation value of dysfunction of diencephalic structures. At F_{1} ≥ F_{2} on should predict the development of remote aftereffects in young people due to evaluating premorbid background of a patients at the moment of trauma.

EFFECT: higher reliability of prediction.

2 ex, 1 tbl

FIELD: medicine; medical engineering.

SUBSTANCE: method involves doing multi-channel recording of electroencephalogram and carrying out functional tests. Recording and storing rheoencephalograms is carried out additionally with multi-channel recording of electroencephalogram synchronously and in real time mode in carotid and vertebral arteries. Electroencephalograms and rheoencephalograms are visualized in single window with single time axis. Functional brain state is evaluated from synchronous changes of electroencephalograms, rheoencephalograms and electrocardiograms in response to functional test. The device has electrode unit 1 for recording bioelectric brain activity signals, electrode unit 2 for recording electric cardiac activity signals, current and potential electrode unit 3 for recording rheosignals, leads commutator 4, current rheosignal oscillator 5, synchronous rheosignal detector 6, multi-channel bioelectric brain activity signals amplifier 7, electrophysiological signal amplifier 8, demultiplexer 9, multi-channel rheosignal amplifier 10, multi-channel analog-to-digital converter 11, micro-computer 12 having galvanically isolated input/output port and personal computer 13 of standard configuration.

EFFECT: enhanced effectiveness of differential diagnosis-making.

11 cl, 6 dwg

FIELD: medicine; medical engineering.

SUBSTANCE: method involves recording multichannel electroencephalogram, electrocardiogram record and carrying out functional test and computer analysis of electrophysiological signals synchronously with multichannel record of electroencephalogram and electrocardiogram in real time mode. Superslow brain activity is recorded, carotid and spinal artery pools rheoelectroencephalogram is recorded and photopletysmogram of fingers and/or toes is built and subelectrode resistance of electrodes for recording bioelectrical cerebral activity is measured. Physiological values of bioelectrical cerebral activity are calculated and visualized in integrated cardiac cycle time scale as absolute and relative values of alpha-activity, pathological slow wave activity in delta and theta wave bandwidth. Cerebral metabolism activity dynamics level values are calculated and visualized at constant potential level. Heart beat rate is determined from electrocardiogram, pulsating blood-filling of cerebral blood vessels are determined from rheological indices data. Peripheral blood vessel resistance level, peripheral blood vessel tonus are determined as peripheral photoplethysmogram pulsation amplitude, large blood vessel tonus is determined from pulse wave propagation time data beginning from Q-tooth signal of electrocardiogram to the beginning of systolic wave of peripheral photoplethysmogram. Postcapillary venular blood vessels tonus is determined from constant photoplethysmogram component. Functional brain state is determined from dynamic changes of physiological values before during and after the functional test. Device for evaluating functional brain state has in series connected multichannel analog-to-digital converter, microcomputer having galvanically isolated input/output ports and PC of standard configuration and electrode unit for reading bioelectric cerebral activity signals connected to multichannel bioelectric cerebral activity signals amplifier. Current and potential electrode unit for recording rheosignals, multichannel rheosignals amplifier, current rheosignals generator and synchronous rheosignals detector are available. The device additionally has two-frequency high precision current generator, master input of which is connected to microcomputer. The first output group is connected to working electrodes and the second one is connected to reference electrodes of electrode unit for reading bioelectrical cerebral activity signals. Lead switch is available with its first input group being connected to potential electrodes of current and potential electrodes unit for recording rheosignals. The second group of inputs is connected to outputs of current rheosignals oscillator. The first group of outputs is connected to current electrodes of current and potential electrodes unit for recording rheosignals. The second group of outputs is connected to inputs of synchronous detector of rheosignals. Demultiplexer input is connected to output of synchronous detector of rheosignals and its outputs are connected to multichannel rheosignals amplifier inputs. Outputs of multichannel bioelectrical cerebral activity signals amplifier, multichannel rheosignals amplifier and electrophysiological signal amplifier are connected to corresponding inputs of multichannel analog-to-digital converter. Microcomputer outputs are connected to control input of lead switch, control input of multichannel demultiplexer, control input of multichannel analog-to-digital converter and synchronization inputs of current rheosignals oscillator and synchronous detector of rheosignals. To measure subelectrode resistance, a signal from narrow bandwidth current generator of frequency f_{1} exceeding the upper frequency f_{up} of signals under recording is supplied. A signal from narrow bandwidth current generator of frequency f_{2}≠ f_{1}>f_{up} is supplied to reference electrode. Voltages are selected and measured at output of each amplifier with frequencies of f_{1}, f_{2} - U_{f1} and U_{f2} using narrow bandwidth filtering. Subelectrode resistance of each working electrode is determined from formula Z_{j}=U_{jf1 }:(J_{f1}xK_{j}), where Z_{j} is the subelectrode resistance of j-th electrode, U_{jf1} is the voltage at output from j-th amplifier with frequency of f_{1}, K_{j} is the amplification coefficient of the j-th amplifier. Subelectrode resistance of reference electrode is determined from formula Z_{A}=U_{jf2 }:(J_{f2}xK_{j}), where Z_{A} is the subelectrode resistance of reference electrode, U_{jf2} is the voltage at output from j-th amplifier with frequency of f_{2}, J_{f2} is the voltage of narrow bandwidth current oscillator with frequency of f_{2}.

EFFECT: wide range of functional applications.

15 cl, 10 dwg

FIELD: medicine, psychiatry.

SUBSTANCE: one should conduct EEG-testing to detect total value of the indices of spectral power or percentage spectral power of delta- and teta-rhythms due to spectrometric technique in frontal, parietal, central and temporal areas both before and during emotional-negative loading when visual emotionally negative stimuli are presented followed by their imaginary reproduction. In case of higher indices to visual stimuli being above 15% against the background one should diagnose epilepsy. The method enables to increase the number of diagnostic means, increase accuracy and objectivity in predicting epilepsy with polymorphic paroxysms at dissociation of clinical and EEG-values.

EFFECT: higher efficiency of diagnostics.

1 ex, 1 tbl

FIELD: medicine, neurophysiology.

SUBSTANCE: one should carry out EEG survey to detect spectrometrically the index of full range if alpha-rhythm both before and after therapy. Moreover, power index of full range of alpha-rhythm and the index of 9-10 Hz-strip's spectral power should be detected in occipital cerebral areas. One should calculate the value of the ratio of the index of 9-10 Hz-strip's spectral power to the index of full range of alpha-rhythm and at the increase of this value by 20% against the background it is possible to evaluate positive result of therapy. The method increases the number of diagnostic means applied in evaluating therapeutic efficiency in the field of neurophysiology.

EFFECT: higher efficiency of evaluation.

1 ex

FIELD: medicine, neurology.

SUBSTANCE: method involves carrying out the standard vascular and nootropic therapy. Diazepam is administrated under EEG control with the infusion rate that is calculated by the following formula: y = 0.0015x - 0.025 wherein y is the rate of diazepam administration, mg/h; x is an average EEG amplitude, mcV. Method provides enhancing the effectiveness of treatment of patients. Invention can be used for treatment of patients in critical severe period of ischemic insult.

EFFECT: enhanced effectiveness of treatment.

2 tbl, 1 dwg, 1 ex

FIELD: medicine.

SUBSTANCE: method involves selecting signals showing patient consciousness level and following evoked auditory potentials as responses to repeating acoustic stimuli, applying autoregression model with exogenous input signal and calculating AAI index showing anesthesia depth next to it.

EFFECT: quick tracing of unconscious to conscious state and vice versa; high accuracy of measurements.

9 cl, 3 dwg

FIELD: medicine; experimental and medicinal physiology.

SUBSTANCE: device can be used for controlling changes in functional condition of central nervous system. Device has receiving electrodes, unit for reading electroencephalograms out, analog-to-digital converter and inductor. Low noise amplifier, narrow band filter linear array which can be program-tuned, sample and store unit, online memory, microcontroller provided with controlled permanent storage, liquid-crystal indicator provided with external control unit are introduced into device additionally. Receiving electrodes are fastened to top part of patient's head. Outputs of electrodes are connected with narrow band filters linear array through electroencephalograph. Output of linear array is connected with input of input unit which has output connected with input of analog-to-digital converter. First bus of analog-to-digital converter is connected with online storage. Recording/reading bus of microcontroller is connected with control input of input unit and its starting bus is connected with address input of online storage. Third control bus is connected with narrow band filters linear array. Second control bus is connected with liquid-crystal indicator. Output bus is connected with inductor. External control (keyboard) of first control bus is connected with microcontroller. Output of online storage is connected with data input of microcontroller through 12-digit second data bus. Efficiency of influence is improved due to getting specific directed influence being based onto general technological transparency of processing of human brain's signals and strictly specific influence based on the condition of better stimulation.

EFFECT: increased efficiency.

3 cl, 1 dwg, 1 tbl

FIELD: medicine, neurology, professional pathology.

SUBSTANCE: one should carry out either biochemical blood testing and electroencephalography or SMIL test, or ultrasound dopplerography of the main cranial arteries, rheoencephalography (REG) to detect the volume of cerebral circulation and hypercapnic loading and their digital values. Then it is necessary to calculate diagnostic coefficients F by the following formulas: Fb/e=6.3-0.16·a1+0.12·a2-1·a3+0.2·a4, or F_{SMIL}=9.6+0.16·a5-0.11·a6-0.14·a7+0.07·a8, or Fhem=48.6-0.04·a9+0.15·a10+13.7·a11-0.02·a12+24.7·a13, where Fb/e -diagnostic coefficient for biochemical blood testings and EEG; F_{SMIL} - diagnostic coefficient for SMIL test; Fhem - diagnostic coefficient for hemodynamic testing; 6.3; 9.6 and 48.6 - constants; a1 - the level of vitamin C in blood; a2 - δ-index by EEG; a3 - atherogenicity index; a4 - the level of α-proteides in blood; a5 - scale 3 value by SMIL; a6 - scale K value by SMIL; a7 - scale 5 value by SMIL; a8 - scale 7 value by SMIL; a9 - the level of volumetric cerebral circulation; a10 - the value of linear circulatory rate along total carotid artery, a11 - the value of resistive index along total carotid artery; a12 - the value for the tonicity of cerebral vessels at carrying out hypercapnic sampling by REG; a13 - the value for the intensity of cerebral circulation in frontal-mastoid deviation by REG. At F value being above the constant one should diagnose toxic encephalopathy, at F value being below the constant - discirculatory encephalopathy due to applying informative values.

EFFECT: higher accuracy of diagnostics.

6 ex, 1 tbl