EA035597B1 - Method of diagnosing border psychic disorders - Google Patents

Abstract

A method of diagnosing border psychic disorders relates to medicine, namely, to functional diagnostics. When implementing the method for diagnosing border psychic disorders, an electroencephalogram (EEG) is recorded, spectral power of the EEG signal and its local maxima are calculated, the informative sections of the EEG signal are additionally extracted, the differential amplitude-time characteristic is calculated, the signal is decomposed using Hilbert-Huang transform, and the energy value of energy density surface and local maxima are calculated that allow distinguishing individual amplitude-time components of the EEG signal, and if the amplitude value of the individual amplitude-time components of the EEG signal exceeds the normal value by 3.5 times and is in the range from 30 to 90 mV; the energy value of the energy density surface of the EEG signal exceeds the normal value by 2.5 times and is in the range from 1200 to 1900 V; the number of amplitude-time components of the EEG signal for individual leads is in the range from 10 to 12, then border psychic disorders are diagnosed. Efficient processing of EEG signals leads to an increase in reliable conclusions and, therefore, to an increase in the efficiency of diagnosis and treatment of border psychic disorders.

Description

The invention relates to medicine, namely to functional diagnostics.

The efficiency of detecting markers of electroencephalographic (EEG) signals for the diagnosis of borderline mental disorders is determined by the accuracy of measuring its amplitude and time, frequency and energy parameters, the reliability of detection and recognition of its individual elements.

The classic methods for diagnosing borderline mental disorders are psychometric methods based on the patient's self-report data on his personal health. Psychometric methods are currently an indispensable tool for determining changes in the patient's condition at various stages of the treatment of mental disorders.

The Hamilton scale is widely used to assess depression, but it does not have the ability to register daily fluctuations in depressive disorders. The Montgomery-Asberg scale, like the Hamilton scale, is also not able to take into account daily fluctuations. There are other screening tests used to diagnose mental disorders about the state of human health, such as the depression self-assessment scale and the depression questionnaire. Thanks to these methods, it is possible to identify signs characteristic of latent depressions.

The above methods often cannot be used for an objective diagnosis of borderline mental disorders, but can be used to assess the patient's emotional background, as well as to correct further therapy.

For the non-invasive diagnosis of borderline mental disorders, the human biological neuro-vegetative signaling system is widely used. The physical carrier of useful information of this biological system is the electroencephalographic (EEG) signal.

An EEG signal is a record of the time-varying biopotentials of the brain. The nature of the EEG is determined by the functional state of the central nervous system. The dependence of the EEG on the general state of the body in borderline mental disorders is widely used in biomedical practice.

Currently, in the field of psychiatry, the EEG signal is used to diagnose and treat seizures, fainting, neurological and mental disorders, traumatic brain injury, strokes, etc.

Changes in the EEG are of a different nature: from changes and disturbances in the amplitude and frequency characteristics of each of the rhythms, the replacement of one rhythm by another, and ending with a flat EEG picture. Most often, in patients with borderline mental disorders, generalization of the alpha rhythm with its predominance in the parieto-occipital regions is revealed. In addition, it is possible to increase alpha activity with hyperventilation. In patients with borderline mental disorders, outbreaks of slow waves in the anterior sections and paroxysmal activity are detected.

Depending on the type of borderline psychic, a certain set of markers is distinguished. In the hysterical type, the alpha activity in the posterior parts of the brain is synchronized, and the slow activity intensifies against the background of hyperventilation. The obsessive type is characterized by frequent changes in the alpha rhythm with periods of desynchronization, increased slow activity against the background of hyperventilation, and lack of response to eye opening and photostimulation. The phobic type is characterized by a pronounced decrease in the alpha activity index, in comparison with other rhythms, due to an increase in the severity of beta and theta activity of waves. The neurasthenic type is characterized by the presence of paroxysmal disorders against the background of hyperventilation.

Let us consider in more detail the known methods of processing EEG signals and detecting markers of borderline mental disorders.

A known method for the diagnosis of affective disorders in patients with coronary artery disease according to encephalographic studies (analogue) [1].

Analysis of the known analogue method [1] showed that its work consists in recording an EEG signal, visual analysis of the signal by a doctor, identifying informative leads characteristic of certain parts of the brain, calculating the spectral power of the signal, calculating local maxima, and making a diagnostic conclusion.

The block diagram of the known method is shown in Fig. 1.

The disadvantage of this method for the diagnosis of affective disorders in patients with coronary artery disease according to encephalographic research is the low efficiency of setting the correct diagnostic conclusion only by visual and classical spectral analysis.

Indeed, the known analogue method [1] uses the approach of detecting the maximum signal power in a certain lobe of the brain, which does not allow taking into account all the local features and changes in the EEG signal in a given frequency range. In addition, visual signal analysis is usually subjective and requires high qualifications and experience of the doctor. The general condition of the doctor can be influenced by fatigue, external factors and subjective errors. Another disadvantage of the known analogue method is the probability of accepting the maximum signal power marker for a high level of physical and physiological interference only according to the results of the classical spectral analysis of the EEG signal in the amplitude-time domain.

- 1 035597

The closest to the proposed invention is a method for early electroencephalographic diagnosis of Parkinson's disease (prototype) [2], which allows the registration of an EEG signal, signal decomposition using wavelet transform, the allocation of an informative frequency range of the signal, characteristic of changes in the brain during the period of the disease, the calculation of local maxima , statement of a diagnostic conclusion.

The block diagram of the known method is shown in Fig. 2.

The disadvantage of the known method [2] is the low efficiency of setting the correct diagnostic conclusion, due to the impossibility of adapting the basic wavelet to the local features of the EEG signal.

Indeed, in the known method for early electroencephalographic diagnosis of Parkinson's disease, a fixed base Morlet wavelet is used to decompose the EEG signal, and therefore it is impossible to take into account all local features (maxima and minima) of a particular investigated EEG signal and its accompanying interferences.

The proposed invention is aimed at improving the accuracy of making a diagnostic conclusion about the presence of borderline mental disorder by detecting new markers of recorded EEG signals.

This is achieved by the fact that in the method of detecting new markers of borderline mental disorders on electroencephalographic signals, which consists in recording the EEG signal, isolating informative signal sections, calculating the spectral power of the signal, calculating the difference amplitude-time characteristic, decomposing the signal using the Hilbert-Huang transform, calculating the values of the energy of the surface of the energy density, the calculation of local maxima, the statement of the diagnostic conclusion.

The block diagram of the proposed method is shown in Fig. 3.

The main distinguishing feature of the proposed method for detecting new markers of borderline mental disorders on electroencephalographic signals from the known prototype [2] is the use of the adaptive method of Hilbert-Hang transformation for the decomposition of the studied EEG signal into frequency components.

Let us consider the implementation features of the proposed method for detecting new markers of borderline mental disorders on electroencephalographic signals. The first stage (registration of the EEG signal) is similar to the stage of the known method [2], which consists in digital registration of the signals under study. As part of this stage on the basis of GBUZ OPB im. K.R. Evgrafova (Penza) for the development and research of the proposed method formed a verified base of EEG signals of the subjects of conditionally healthy and with symptoms of borderline mental disorders.

The database of EEG signals included 100 people in the experimental group and 100 people in the control group. The average age of the subjects was 39 years old, 42 of whom are males and 58 females.

According to clinical signs, the composition of each group of subjects with borderline mental disorders is represented mainly by four diagnostic categories:

F48.0 - neurasthenia - 10 subjects;

F45.3 - somatoform dysfunction of the autonomic nervous system - 20 subjects;

F43.2 - disorder of adaptive reactions - 40 subjects;

F41.2 - mixed anxiety and depressive disorder - 30 subjects.

An 8-channel electroencephalograph Neurospetr-1 is used to record EEG signals.

To conduct the research, a sample was formed of 100 relatively healthy subjects and 100 subjects with borderline mental disorders. The total number of recorded signals of the subjects was 1600 EEG signals (200 subjects with 8 lead signals).

The next stage in the selection of informative signal sections consists in the selection of individual signal regions in all (eight) leads in the amplitude-time domain, characteristic of the time periods of signal registration before / after the patient is exposed to an external stimulus. The selection of informative signal sections is carried out according to the following formulas:

6,) »About where t _m is the EEG signal time count at the moment of the onset of a mental disorder; t _n - time count of the signal at the end of the EEG signal recording during the period of mental disorder, but not more than 30 s.

T2 = (t _k - t _m \ (2) where t _k is the signal time count equal to the duration (t _m -tn).

This stage makes it possible to identify the range of the studied EEG signal during the period of exposure to external stimuli, thus, to form a data sample for the effective detection of new markers of the studied EEG signals during the period of borderline mental disorders. The result of this stage is the formation of 3200 EEG signals.

- 2 035597

The third stage Calculation of the spectral power of the signal is similar to the stage of the known analogue method [1] and consists in obtaining the spectral characteristics of the selected signal sections by means of the classical Fourier transform.

However, in the works of the known analogue method [1], this stage is intermediate for calculating the local maxima of the EEG signal in all leads.

FIG. 4 shows the corresponding results of constructing the spectral power of the EEG signal. In the proposed method, this stage allows us to conclude that the amplitude of the EEG signal in the 1st and 2nd leads in subjects with borderline mental disorders in the frequency range (0-10) Hz is 1.5-2 times higher in comparison with that the same indicator in conditionally healthy subjects.

The next, fourth stage of the proposed method for detecting new markers of borderline mental disorders on electroencephalographic signals Calculation of the difference amplitude-temporal characteristics of paired EEG leads is that for each pair of leads (symmetric leads of the cerebral cortex) of the selected areas T1 and T2, their difference is calculated and compared the received amplitude value with a threshold value.

FIG. 5 shows the sections of paired derivations of the EEG signal. It was revealed that the amplitude of the difference in leads in the period of borderline mental disorders is 1.5-2 times higher in comparison with the same indicator in conditionally healthy subjects.

The next stage Signal decomposition using the Hilbert-Huang transform includes the sequential execution of actions [3]:

1. Determination of local extrema (maxima and minima) of frequency components of EEG signals / j (ti):

the value of the i-th sample / j (ti) is a local maximum if the condition / j (t _i . ₁ ) </ j (ti)> / j (ti ₊₁ ) is fulfilled;

the value of the i-th sample / j (ti) is a local minimum if the condition / j (t _i . ₁ )> / j (ti) </ j (ti +!) is satisfied.

2. Determination of the upper ej (ti) and lower gj (ti) envelopes of the frequency components of the EEG signal using cubic spline interpolation [4] according to the found local extrema / j (ti):

^e A) = ^a A ³ + b „1 ( ² + cBt, + d _B (3), gj th) = α, - ζ ³ + b {t, ² + C; tt + d ₍ (4), where a _in , b _in , c _in , du are the coefficients for each value of the i-th sample of the upper envelope of the frequency components of the ECS; a _n , b _n , s _n , d _n are the coefficients for each value of the i-th sample of the upper envelopes of the frequency components of the EEG signal.

3. Calculation of the average value of the envelopes of the frequency components of the EEG signal in accordance with the expression:

where hj (ti) is the average value of the envelopes of the frequency components of the EEG signal;

e _j (t _i ) and g _j (t _i ) are the upper and lower envelopes of the initial frequency components of the EEG signals, respectively.

4. Calculation of the remainder of the frequency components of the EEG signal by the formula:

γ (Ο = / ₇ 0,) - Λ «·) (6), where Sj (ti) is the remainder of the frequency components of the EEG signals.

5. Calculation of the stop criterion value. As a criterion for stopping the decomposition, the value of the normalized quadratic difference is used, defined as:

6. Checking the stop condition. At this stage, the value of the remainder of the frequency components of the EEG signal is compared with the value of the normalized quadratic difference. If a:

SD> Sj (ti), then go to action 1;

SD <s _j (t _i ) and h _j (t _i )> s _j (t _i ), then proceed to the next action.

7. Conclusion of modes of frequency components of the EEG signal. At this stage, the modes mk (ti) and the remainder Sj (ti) of the frequency components of the EEG signals are output.

Thus, the result of this stage is to obtain the frequency components of the T1 and T2 sections of the EEG signal, which are adaptive to the characteristics of the signal under study.

The analysis of the results of the Hilbert-Huang transformation of the studied EEG signals revealed the dependence of a fixed number of frequency components of conventionally healthy subjects and subjects with borderline mental disorders. The number of modes for the first and second leads in the T1 section is in the range from 10 to 12, in the T2 section - from 9 to 10 modes (Fig. 6).

Next step Calculation of the energy density value of the surface.

The Hilbert-Huang transform, in addition to the decomposition into modes, includes Hilbert's spectral analysis. When constructing a Gilbert spectrum, an instantaneous frequency is calculated for each frequency component, then it is plotted on a 3D spectrogram. In [5], a 3D spectrogram constructed using the Hilbert-Huang transform is called the energy density surface (EP).

The Hilbert conjugate signal M _k (ti) is represented by an indefinite integral:

where mk (ti) - modes of the EEG signal; where ti - discrete time readings; k is the mode number; τ - time shift proportional to the phase of the signal.

Then, for each mode of the EEG signal, an analytical (complex) signal is determined:

where j = C1 is the imaginary unit.

After performing the Hilbert transform, each mode of the EEG signal is represented as a surface in the amplitude-frequency-time coordinate system and is expressed as follows:

P

4 = 1

To represent the energy density of the frequency components of the EEG signal and build a surface in the energy-frequency-time coordinate system, the instantaneous amplitude is given a quadratic form:

(10), (Π), n) = J; V ω · β ^{,, 4} “ ⁾ '* 4 = 1

The result of performing actions (8-11) is the surface of the energy density (PEP) in the energy-frequency-time coordinate system built according to the modes of the EEG signal.

The next stage The calculation of local maxima is performed for the frequency components and the surface of the energy density of the EEG signal.

The essence of the stage is to calculate the threshold value of the amplitude of the EEG modes for the period of onset of borderline mental health.

The results of the studies showed that in the second and third modes of the EEG signal at the time of the onset of borderline mental disorders, the signal amplitude changes, which is 2-3 times higher than the signal amplitude of a conditionally healthy subject (Fig. 7).

FIG. 8 shows the PEP of the EEG signal during the period of borderline mental disorder. The energy of the PEP EEG is 1.5-2 times higher than the same value in comparison with the PEP of the conditional healthy section of the EEG signal.

The final stage of the proposed method for detecting new markers of borderline psychidiagnostic mental comb disorders on electroencephalographic signals is to formulate a conclusion, which consists in making a decision on the presence of borderline structures in subjects according to the data of EEG signal markers.

To assess the effectiveness of determining borderline mental disorders using new markers of EEG signals and accepting them as decisive rules, the value of errors of the first and second kind is used. From the research results it follows that the error of the 1st kind is 2%, and the error of the 2nd kind is 5%.

The above description of the method for detecting new markers of borderline mental disorders on electroencephalographic signals shows that the proposed method eliminates the shortcomings of the known method for the early electroencephalographic diagnosis of Parkinson's disease [2], namely, to carry out adaptive decomposition of the studied EEG signal into frequency components, to detect new markers of EEG signals on modes and surfaces of energy density, and therefore with high accuracy to make a decision about the presence of borderline mental disorders in a person.

Effective processing of EEG signals leads to an increase in reliable conclusions and, consequently, to an increase in the efficiency of diagnosis and treatment of borderline mental disorders.

Sources of information:

1. RF patent Method for the diagnosis of affective disorders in patients with coronary artery disease according to encephalographic studies. Minakov E.V., Kudashova E.A., Voronina E.A., No. 2419383 dated 05/27/2011, bulletin No. 15.

- 4 035597

2. RF patent Method for early electroencephalographic diagnosis of Parkinson's disease.

Obukhov Yu.V., Korolev M.S., Gabova A.V., Kuznetsova G.D., Ugryumov M.V., No. 2484766 dated 20.06.2013, bul.

No. 17.

3. Huang N.E., Shen Z., Long S.R. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis. Proc. R. Soc. Lond. A., Vol. 454, 1998, pp. 903-995.

4. Huang N.E., Shen S.S. The Hilbert-Huang transform and its applications - World Scientific Publication, 2005, pp. 323.

5. Tychkov A.Yu., O.N. Bodin, P.P. Churakov Automated system for heart contour delineation in photofluorographic images. Biomedical Engineering. - Springer New York. - USA, 2011. Vol. 29. P. 144-151.

Claims (1)

CLAIM A method for the diagnosis of borderline mental disorders, including the registration of electroencephalography (EEG), the calculation of the spectral power of the EEG signal, its local maxima, characterized in that, additionally, the selection of informative sections of the EEG signal, the calculation of the difference amplitude-time characteristics, the decomposition of the signal using the Gilbert-Huang transformation , calculation of the value of the energy of the surface of the energy density and the calculation of local maxima, which make it possible to isolate individual amplitude-time components of the EEG signal, and if the value of the amplitude of the individual amplitude-time components of the EEG signal exceeds the normal value by 3.5 times and is in the range from 30 to 90 mV ; the value of the surface energy of the energy density of the EEG signal exceeds the normal value by 2.5 times and is in the range from 1200 to 1900 μV ² ; the number of amplitude-time components of the EEG signal for individual leads is in the range from 10 to 12, then borderline mental disorders are diagnosed.