RU128470U1 - CARDIOAIGENOSCOPE - Google Patents

Abstract

1. A cardioigenoscope containing a block of the calculator of the matrix of mixed moments 2, a block of the calculator of eigenvectors and eigenvalues 3, characterized in that at the input of the device there is a block for forming the ensemble of cardio-oscillations 1, the output of which is connected to the input of the block of the calculator of the matrix of mixed moments 2 and to the first input cardiac signal recovery and feature analysis 4; the output of the block of the calculator of the matrix of mixed moments 2 is connected to the input of the block of the calculator of eigenvectors and eigenvalues 3, the first and second outputs of which are connected to the second and third inputs of the block recovery cardiosignal and analysis of signs 4; the output channel of the cardiosignal recovery and sign analysis unit 4 is two-way and is connected with the environment of the cardioigenoscope. 2. A cardioigenoscope according to claim 1, characterized in that the cardio-oscillation ensemble formation unit 1 consists of: an ADC input unit - 5, a sample storage unit in the analysis interval 6, a centering unit 7, an inverter 8 unit, a threshold calculation unit 9, an asymmetry coefficient calculation unit 10 , a position determiner block 11, a calculator block of an average period of cardiooscillations 12, a block for forming a matrix of an ensemble of cardiooscillations 13, each of which, individually, is bilaterally connected to the output program control unit 14; the first output of the ADC - 5 unit is additionally connected to the input of the sample storage unit in the analysis interval 6.3. A cardiac eigenoscope according to claim 1, characterized in that the cardiac signal recovery and feature analysis unit 4 consists of an input memory unit 15, which is bilaterally connected

Description

The field of technology.

The Cardioigenoscope device is an analyzer of the independent components of cardiograms and refers to medical devices used in cardiology. A cardio-eigenoscope has additional functional qualities that allow for separate analysis of uncorrelated components of a cardiosignal, which can be useful in assessing the patient’s condition and can better filter the cardiosignal from interference. A cardioigenoscope can be implemented both in the form of a hardware-software complex, and in the form of a virtual device based on a standard computer.

The level of technology. Analogs and their shortcomings. Prototype.

It is known [1] that the process of cardio diagnostics is often fraught with errors. Approximately 38% of electrophysiologists (according to [2]) accept a cardiac signal artifact for ventricular tachycardia, and 55% of medical reports (according to [3]) of Holter daily monitoring contain methodological errors. Up to 43% of false alarms from bedside ECG monitors (according to [5]) are caused by incorrect interpretation of signal artifacts by the built-in software. According to [4], errors in electrocardiodiagnostics unacceptably often lead to the fact that the patient is prescribed the wrong treatment, up to the appointment of an operation to implant a pacemaker. It follows from the foregoing that the task of reducing the likelihood of an error in electrocardiodiagnostics is relevant.

The author [1] analyzed literature sources, catalogs of suppliers and manufacturers of medical equipment, during which model ranges of devices from 101 electrocardiographic equipment manufacturing companies were examined, including 17 domestic manufacturers (Altonika LLC, IzhMedical CJSC, LLC Neurosoft and others). It has been established that leading foreign manufacturers currently offer additional specialized software for their devices, which allows assessing the diagnostic significance of the recorded ECG signal, for example, the General Electric Marquette Hookup Advisor technology. However, most manufacturers offer this functionality only in the older and most expensive models of their lineups. For example, the Philips Page Writer Trim II cardiograph allows you to evaluate the diagnostic significance of the signal at each lead in real time. But due to the high cost of these devices are not available for widespread and widespread use. At the same time, domestic appliances practically do not have the proper capabilities. So, for example, the Ankar-131 electrocardioanalyzer of the Medicom-MTD firm allows you to monitor the quality of electrode installation and to indicate the lead abstraction. However, it does not provide an opportunity to assess the diagnostic significance and the presence of artifacts in the signal. The author [1] has not yet identified publications in the domestic literature on the development of metrics for the diagnostic significance of the electrocardiographic signal. At the same time, there is an increased interest in this problem from the side of the foreign scientific community. This is evidenced by the increase in the number of publications on this topic and the competition held in 2011 on the basis of the international resource PhysioNet.com aimed at developing methods to improve the quality of electrocardiographic studies, in particular telemedicine systems based on mobile phones (smartphones) [6]. Therefore, it is extremely important to improve the known or develop new algorithms for assessing the diagnostic significance of an electrocardiographic signal, in order to reduce the computational cost and expand the ability to recognize more artifacts.

The foregoing makes it relevant to use in cardioscopy new principles of signal processing, including those based on the use of an aigenoscope (Eigenvector Analyzer and signal components [7]).

The principle underlying the work of the Cardio-Eigenoscope is the same as the principle of the Eigenoscope, which is considered as a prototype (Fig. 1). Thus, a Cardio-Eigenoscope can be considered as a specialized Aigenoscope, adapted for the analysis of cardiograms, which due to the claimed design features allows you to get the claimed technical result.

Disclosure of a utility model.

Cardiac eigenoscope is designed to solve the problem of early diagnosis of changes in cardiac signal; it uses a joint analysis of uncorrelated components and eigenvectors of the cardiac signal. The direct use of the device [7] (prototype) does not allow to obtain the expressiveness of the uncorrelated components provided by the cardioigenoscope. By expressiveness, we mean the ability to approximate the cardiac signal as accurately as possible with a minimum number of uncorrelated components. The proposed device is a tool that for the first time provides a technical result, consisting in obtaining maximum expressiveness of the first 2-4 uncorrelated components of the cardiac signal.

Let us dwell on the general features of an aigenoscope [7] and the proposed Cardioaigenoscope, the designs of which in statics (structural diagrams) are shown in Figs. 1 and 2, respectively. Table 1 shows the features of the prototype with the selection of those that coincide with the essential features of the claimed utility model, in table 2, on the contrary, the features that distinguish the proposed utility model from the prototype are highlighted.

Table 1 Npp Sign of analogue (prototype) The presence of a sign in the claimed utility model one Scaling block Absent 2 Mixed moment matrix calculator Is present 3 Block for calculating eigenvectors and eigenvalues Is present four Block for calculating scalar products and an analyzer of signs Absent 5 Communication “Device input - the first input of the scalar product calculation unit and the feature analyzer” Absent 6 Communication "Device output unit for calculating the matrix of mixed moments- Unit for calculating the eigenvectors and eigenvalues Is present 7 Communication “The output of the block for computing the matrix of mixed moments - the second input of the block for calculating scalar products and the analyzer of signs” Absent 8 Communications "the first and second outputs of the block for calculating eigenvectors and eigenvalues - the third and fourth inputs of the block for calculating scalar products and the analyzer of signs" Absent 9 Two-way communication “scalar product and feature analyzer calculation unit - environment” Absent

table 2 Npp Utility Model Feature The presence of the characteristic in the prototype one Block for the formation of an ensemble of cardiac oscillations Absent 2 Mixed moment matrix calculator Is present 3 Block for calculating eigenvectors and eigenvalues Is present four Cardiac signal recovery and feature analysis unit Absent 5 Communication “Input of calculating the matrix of mixed moments - the first input of the recovery block of the cardiosignal and analysis of signs” Absent 6 Communication "Output block computing the matrix of mixed moments - block computing the eigenvectors and eigenvalues" Is present 7 Communication “The output of the block for computing the matrix of mixed moments - the second input of the block for calculating scalar products and the analyzer of signs” Absent 8 Communication "the first and second outputs of the block for calculating eigenvectors and eigenvalues - the second and third inputs of the block for restoring the cardiosignal and analysis of signs" Absent 9 Two-way communication “cardiac signal recovery and feature analysis unit - environment” Absent

If in an aigenoscope to form a matrix of mixed moments (including a covariance matrix), the so-called trajectory matrix, then in the proposed device the matrix of mixed moments (in the particular case, the covariance matrix) is calculated on the basis of the matrix of the ensemble of cardio-oscillations, which is formed from the segment of the cardiac signal corresponding to several tens of heartbeats. Each element of the ensemble (matrix row) is a segment of a signal of a certain length, which contains a signal from a single heart beat. This feature of the Cardioeigeposcope, along with others that distinguish it from the prototype (underlined in table 2), are essential differences that allow to obtain a technical result, which consists in a significant reduction (in comparison with the prototype - an aigenoscope) of the number of energetically significant uncorrelated components of the cardiac signal, which is essential simplifies their analysis and use in diagnosis.

This technical result is obtained due to the totality of the features of the claimed model and is not achievable in the prototype (this will be discussed separately below); this result is manifested objectively in the manufacture and use of a cardioigenoscope always, regardless of how the blocks that make up the cardioigenoscope are implemented.

We show that the number of energetically significant uncorrelated components of the cardiosignal (when the QRS complex is pronounced) is directly (causally) related to the method of formation of the ensemble of cardiocycles. Fig. 18 shows a comparison of the contribution of uncorrelated components to the energy of a cardiosignal, which are obtained using a cardioigenoscope (curve 1) and a prototype (curves 2-4) for the case of processing a cardiosignal from a group with myocardial infarction (signal database [10]). As can be seen from the graphs, a cardioigenoscope provides the claimed technical result - the maximum contribution to the energy of the cardiac signal of the first two to four uncorrelated components in this case is higher than that of the prototype. So in the case of a cardioigenoscope (curve 1 of Fig. 18), the first 2 uncorrelated components have an approximately 83% contribution to the energy of the cardiac signal, while in the case of the prototype this value ranges from 46-55% (curves 2-4 of Fig. 18). The reason why such a technical result is achieved by the cardioigenoscope is that the ensemble of cadio oscillations is formed by block 1. Similar cause-effect relationships are observed for other nosological groups of cardiac signals. Thus, the technical result is achieved precisely due to the block forming the ensemble of cardiac oscillations and its relationships with other blocks.

Once again, we note that such an effect is manifested objectively regardless of how the blocks included in the device are implemented and to which nosological group the analyzed cardiac signal belongs.

Thus, the reason that impedes the obtaining of a technical result in the prototype is that in the prototype, in the block for calculating the matrix of mixed moments, the trajectory matrix is used, which is formed on an arbitrary analysis interval that is not connected in any way with the period of cardiac oscillations, and in the utility model, block 1 for forming the ensemble of cardiac oscillations forms an ensemble , in which the R-peak has a fixed position, which helps to obtain a technical result. Therefore, the set of features that provides a technical result in all cases to which the requested amount of legal protection applies is the presence of a unit for forming an ensemble of cardiac oscillations associated with a unit for calculating the matrix of mixed moments.

The inventive device is an unchanged (idle) set of blocks, the purpose of which (what they should do while working, but do not do it in a static state, while being ready to do it) is given in the descriptions of figures 1,2 and 3. Blocks 2 and 3 - the same as in the prototype. Block 1 in a static state consists of blocks shown in Figure 2, and block 4 in a static state consists of blocks shown in Fig. 4. The interconnections (arrows) shown in these figures between the blocks are also present in a static state and are input-output interface devices ([8, p. 120]).

The device contains two-way communication with weapons. The word "environment" in the technical language is generally accepted. For example, it is included in the dictionary [8, p. 195] as a translation of the English word environment. The environment of a system usually refers to many elements and their essential properties that are not parts of the system, but a change in any of them can cause a change in the state of the system.

Typical but obviously not the only environments of a cardioigenoscope can include:

- a device for collecting information and displaying information - in the event that several cardioigenoscopes are connected to the same workplace (for example, a person on duty in the intensive care unit)

- a device for remote control of a cardioigenoscope (for example, allowing you to change the modes of its operation). For communication with the environment can be used a variety of

well-known devices (which can be divided according to level

descriptions):

1. Connecting cords (path cord) - [8, p. 367]

2. Data buses (data bus) - [8, p. 365]

3. Interface (interface device) - [8, p.120]]

Touch screen (touch-sensitive screan) - [8, p. 371]

Пример кардиосигнала, в котором наблюдается доминирование пиков, приведен на фиг.5.а. Такой сигнал наблюдается далеко не всегда. На фиг.5.6 приведен кардиосигнал, соответствующий фибрилляции (пациент находится в коме). Как видно из рисунка, пики в этом случае не доминируют.A cardioeigenoscope processes successive segments of a sampled and digitized cardiosignal having a duration of N samples, into which, as already noted, several dozen cardiac oscillations fit. The signal on each of these segments is set by a sequence of discrete samples

An example of a cardiac signal in which peak dominance is observed is shown in FIG. Such a signal is not always observed. Figure 5.6 shows a cardiac signal corresponding to fibrillation (the patient is in a coma). As can be seen from the figure, peaks in this case do not dominate.

, а затем формируется ансамбль кардиоосцилляций. Если доминирования пиков нет, то сразу определяется средний период, а затем формируется ансамбль. Каждый элемент ансамбля представляет собой матрицу-строку, описывающую отдельный сегмент кардиосигнала на интервале, равном (как правило) среднему периоду. В случае доминирования пиков, в центре строки расположен пик кардиоосцилляций, а общая длина матрицы-строки равна дискретному значению среднего периода

. Пример такого ансамбля приведен на фиг.б.а - для случая домнирования пиков, а на фиг.б.б - для случая отсутствия доминирования.Each of the sequences of length N is centered and reduced (if necessary) to the “normal” form (with positive peaks if the latter dominate). If the peaks dominate, then their position and the average period of their succession are determined

, and then an ensemble of cardiooscillations is formed. If there is no dominance of peaks, then the average period is immediately determined, and then an ensemble is formed. Each element of the ensemble is a row matrix that describes a separate segment of the cardiosignal in an interval equal to (as a rule) the average period. In the case of peaks dominance, the peak of cardio-oscillations is located in the center of the row, and the total length of the row matrix is the discrete value of the average period

. An example of such an ensemble is shown in Fig. Ba for the case of peak domination, and in Fig. B for the case of lack of dominance.

с размером

, где К - число элементов ансамбля, равное числу целых кардиоосцилляций размера

, которые укладываются в интервал анализа N.The ensemble is a rectangular matrix.

with size

where K is the number of ensemble elements equal to the number of integer cardiooscillations of size

which fit into the analysis interval N.

определяется ковариационная матрица кардиоосцилляций, задаваемая соотношениемBased on matrix

the covariance matrix of cardiooscillations is determined by the relation

Where (...) ^' - means transposition of the matrix.

For the covariance matrix (1), eigenvectors and eigenvalues are found that satisfy the relation

≤700 с использованием стандартных вычислительных средств составляет 1…5 секунд, что позволяет оценивать собственные векторы и собственные значения ковариационной матрицы. Средняя энергия сигнала, наблюдаемого на интервале

, определяется соотношениемThe dimension of the covariance matrix for the case of discretization of a cardiosignal with a period of 10 ^-3 seconds is usually not more than 700. The calculation time of the covariance matrix based on relation (1) for K≤50 and calculation of the eigenvectors and eigenvalues of the matrices at

≤700 using standard computing tools is 1 ... 5 seconds, which allows us to estimate the eigenvectors and eigenvalues of the covariance matrix. The average energy of the signal observed in the interval

is determined by the relation

This means that the sequence

which will be called the normalized spectrum of eigenvalues, determines the relative fraction in the average energy (3) belonging to the corresponding eigenvector. Further, as in [7], we will order the normalized spectrum of eigenvalues in descending order. Since each normalized eigenvalue expresses information about the fraction of energy that the corresponding eigenvector “explains”, we will call the value of the normalized eigenvalue corresponding to the eigenvector the expressiveness of the eigenvector. Further we will indicate expressiveness in%. The total value of the expressivities of all eigenvectors is 100%. The higher the expressiveness of the eigenvector, the more information it carries.

имеет количество ненулевых собственных значений равное К. На экранной форме, представленной на фиг.7.а (верхнее правое окно) приведен типичный нормированный спектр (4), построенный для случая К=35 и

=680. Как видно из графика, представленного в окне, только 35 первых нормированных собственных значений имеют уровень выше 10^-5.It can be shown that the normalized spectrum of eigenvalues for the covariance matrix (1) constructed using

has the number of nonzero eigenvalues equal to K. The typical normalized spectrum (4) constructed for the case K = 35 and

= 680. As can be seen from the graph presented in the window, only the first 35 normalized eigenvalues have a level above 10 ^-5 .

But even such a number of eigenvalues is redundant for analysis. In the upper left window of the form shown in Fig.7.a, the first 10 normalized eigenvalues are shown, of which only 4 have expressivity in excess of 1%, and only 7 have expressivity that exceeds 0.1%. Therefore, for diagnosis, it is enough to visualize the first 10 normalized eigenvalues. In the lower windows of the screen form shown in Fig.7.a, graphs of cumulative expressiveness are shown. Each point of this graph represents the sum of the number of first eigenvalues given on the abscissa axis. So from the lower left graph it is clear that the first 4 normalized eigenvalues provide cumulative expressiveness (explained average energy) at the level of 99%. This value is typical for a healthy heart.

In the case of fibrillation, the situation changes: the expressiveness of the first eigenvector is significantly reduced, and the expressiveness of the first and second eigenvectors becomes comparable. This is illustrated by the screen form shown in Fig.7.b. In the absence of dominance, the first 10 eigenvalues may not be enough to analyze expressiveness. Therefore, in this case, it is recommended to use all normalized eigenvalues, the value of which exceeds 10 ^-5 .

The screen form shown in Fig. 8.b shows a typical view of the first four eigenvectors for the case where there is no peak dominance.

The expressiveness of the eigenvectors can vary with time. On the screen form shown in Fig. 9, the graph in the upper right window shows how the expressiveness of the first four eigenvectors has changed over 11 analysis intervals; in the upper left window shows the initial cardiac signal in the last analysis interval, and in the lower windows the normalized spectrum of the first 10 eigenvalues for the last analysis interval (left) and the cumulative normalized spectrum for the same analysis interval (right).

An analysis of the normalized spectra of eigenvalues shows that it makes sense to analyze only the first eigenvectors that have the most expressiveness. The screen form shown in Fig. 8.a shows the first four eigenvectors for a particular analysis interval — for the case of peak dominance. As can be seen from the figures, indeed, the first eigenvector has the maximum expressiveness (over 88%). It makes sense to analyze the remaining three vectors in special cases. So, for example, when the level of the cardiosignal is weakened below the level of harmonic interference, the first (and possibly the second) eigenvector will represent the form of interference, and the following will express the cardio signal.

The eigenvectors with numbers 2-4 are less expressive than the first one and, as a rule, carry information about the interference acting on the cardio signal and its fluctuations. The following approach can be used to identify areas of cardiooscillation instability. For eigenvector modules with numbers 2-4, the geometric mean is determined (an example is shown in the left window of the screen form shown in Fig. 10). In those time intervals at which all (with numbers 2-4) eigenvectors have large values, the geometric mean also has large values. For a geometric mean, a quantile can be calculated

of a given order (the figure shows an example of the percentile X _0.97 ) and the boundaries of the time intervals at which the geometric mean exceeds the quantile are determined. These boundaries define the zone of instability of cardiac oscillations and can be visualized on a graph of the first eigenvector, as shown in the right window of the screen form shown in Fig. 10. Indication of instability zones facilitates the diagnostic procedure.

Finally, the resulting eigenvectors can be used to restore cardiac oscillation. In this case, the relation

Where

- ансамбль, состоящий из К кардиоосцилляций.восстановленных с помощью L собственных векторов,

- an ensemble consisting of K cardiooscillations restored with the help of L eigenvectors,

- исходный ансамбль кардиоосцилляции, используемый для построения собственных векторов,

- the initial ensemble of cardiac oscillations used to construct eigenvectors,

- матрица собственных векторов, используемых при восстановлении (собственные векторы записаны в эту матрицу как столбцы).

- matrix of eigenvectors used in reconstruction (eigenvectors are written in this matrix as columns).

11 and 12 show an example of recovery using relation (5) for the case L = 4 and L = 2, respectively.

In a cardioigenoscope with the structure described above, various visualization options can be performed. The most economical is the interface with two screen forms, the use of which for the case of peak dominance and lack of dominance is shown in Figs. 13 and 14.

As shown in fig.14.b, if necessary, the restoration of the components of the cardiac signal can be performed repeatedly. On the screen form shown in Fig. two - repeated recovery. As a result, the high-frequency component of the cardiosignal, which has a frequency of about 4.3 Hz, is detected, while the component recovered from the first two eigenvectors has a frequency of about 2.4 Hz.

The formation of an ensemble of cardiooscillations for the case of dominance and the absence of pronounced peaks occurs, as already noted, on an interval equal to the average period of cardiooscillations (corresponding to the pulse rate). However, the algorithm for measuring the average period is different for these two cases. The device should automatically distinguish the type of cardiac signal, which is analyzed on the analysis interval.

To study the distinction between cardiac signals with dominance and absence of peaks, four criteria were analyzed, which are shown in Table 1.

On Fig shows estimates of the probability density for the criteria of distinction indicated in table 1. As can be seen from the figures, the highest quality of discrimination

выборочная дисперсия,

- выборка кардиосигнала на интервале анализа.the criterion is different using the asymmetry coefficient γ ₁ calculated by the standard formula

sample variance

- sampling of the cardiac signal in the analysis interval.

As the analysis of Fig. 15a shows, the best threshold value for the dominance criterion using the asymmetry coefficient is 2. Therefore, it is proposed to use the following decision rule in the device:

if γ ₁ ≥1, then there is a dominance of peaks,

if γ ₁ <2, then there is no dominance of peaks.

Let us dwell on how the ensemble of cardiooscillations is formed in the device, which is used later to calculate the covariance matrix and its eigenvectors and eigenvalues.

Table 3 Criteria and decision rules for revealing the dominance of cardiac signal peaks Npp Criterion Decision rule one The asymmetry coefficient γ ₁ if γ ₁ ≥1, then there is a dominance of peaks, if γ ₁ <2, then there is no dominance of peaks. 2

,

- выборка кардиосигнала на интервале анализа.The ratio of the deviations of the extremes of the cardiac signal from the sample average

,

- sampling of the cardiac signal in the analysis interval. if

then there is a dominance of peaks if

then the dominance of the peaks is absent where

number selectable within

3 The ratio of the deviations of the extremes of the cardiac signal from the sample median

,

- sampling of the cardiac signal in the analysis interval. if

then there is a dominance of peaks if

then the dominance of the peaks is absent where

number selectable within

. four The ratio of quantile deviations q _0.95 and q _0.05 from the median q _0.5

if

then there is a dominance of peaks if

then the dominance of the peaks is absent where

number selectable within

The structural diagram of the formation of the ensemble of cardiooscillations is presented in figure 3. All the blocks that make up the block forming the ensemble of cardiac oscillations 1 have two-way communication with the program control unit 14; the actions performed by the blocks are described in table 2. Since with the same composition of the blocks included in block 1 in different modes they interact with each other in different ways under the control of block 14, we will separately consider the formation of an ensemble of cardio-oscillations in the modes of presence and absence of peak dominance.

Peak dominance mode. After the asymmetry coefficient was calculated for the cardio signal segment in block 10 and it turned out that its value exceeds threshold value 2, the ensemble formation scheme for the peak domination mode comes into effect. Block 9 calculation of the threshold in this mode determines the current

the analysis interval N for the analyzed segment of the cardiosignal quantile q _0.95 or q _0.97 - The magnitude of the order of the quantile is chosen so that the quantile is located close to the linear portion of the distribution function of the cardiac signal. On Figa presents a typical distribution function for the mode of dominance of the peaks. A comparison of Figures 16.a and 16.b shows that in the absence of dominance mode, the upper right portion of the distribution function has a curved (close to quadratic) character.

удовлетворяют логическому условию

; в этом случае номер отсчета i записывается в первую последовательность. Вторая последовательность соответствует ситуации когда два смежных отсчета

удовлетворяют логическому условию

; в этом случае номер отсчета i записывается во вторую последовательность. Таким образом, в эти последовательности записываются номера элементов выборки кардиосигнала на анализируемом интервале, которые соответствуют переходам кардиосигнала через пороговое значение «снизу-вверх» и «сверху-вниз», соответственно. Нижний рисунок фиг.17 иллюстрирует сказанное. Естественно, может наблюдаться ситуация, когда длины этих последовательностей не совпадают. Последовательности (целиком или частями) передаются из блока 11 в блок 14; после того как эти последовательности сформированы полностью в этом же блоке определяется положение максимумов (пиков) кардиосигнала. Положение максимумов может быть определено как по переходам «снизу-вверх» и «сверху-вниз», так и по средним значениям, находящимся между этими переходами. Естественно, это возможно только в том случае, если длины последовательностей переходов «снизу-вверх» и «сверху-вниз» совпадают.Определенные одним из перечисленных способов положения передаются в блок 14 где запоминаются для дальнейшего использования при формировании ансамбля кардиоосцилляций, и передаются в блок 12. Итак, в блок 12 поступает последовательность номеров отсчетов исходного сегмента кардиосигнала, которым соответствуют максимумам пиков. В блоке 12 вычисляется длительность (в дискретах) интервалов между пиками, после чего она усредняется, и от полученного значения берется целая часть. Эта величина и представляет собой средний период следования кардиоосцилляций (целое число); она передается в блок 14.The threshold value from the output of block 9 is transmitted to block 14, which transmits this value together with the samples of the cardiosignal in the analysis interval (all together or sequentially counted) to the position determination unit 11, which generates two sequences of sample numbers. The first sequence corresponds to the situation when two adjacent samples

satisfy the logical condition

; in this case, the reference number i is written in the first sequence. The second sequence corresponds to the situation when two adjacent samples

satisfy the logical condition

; in this case, the reference number i is written in the second sequence. Thus, in these sequences, the numbers of the elements of the cardiac signal sample on the analyzed interval are recorded, which correspond to the transitions of the cardiosignal through the threshold value “bottom-up” and “top-down”, respectively. The bottom figure of FIG. 17 illustrates the foregoing. Naturally, a situation may be observed when the lengths of these sequences do not coincide. Sequences (in whole or in part) are transmitted from block 11 to block 14; after these sequences are formed completely in the same block, the position of the maxima (peaks) of the cardiosignal is determined. The position of the maxima can be determined both from the “bottom-up” and “top-down” transitions, and by the average values located between these transitions. Naturally, this is possible only if the lengths of the sequences of transitions “bottom-up” and “top-down” coincide. The positions determined by one of the listed methods are transferred to block 14 where they are stored for later use in the formation of the ensemble of cardio-oscillations and transferred to the block 12. So, in block 12 receives a sequence of numbers of samples of the initial segment of the cardiac signal, which correspond to the maximum peaks. In block 12, the duration (in discrete) of the intervals between the peaks is calculated, after which it is averaged, and the integer part is taken from the obtained value. This value represents the average period following cardiac oscillations (integer); it is transmitted to block 14.

Next, the average follow-up period of cardio-oscillations, the sequence of peaks and the sequence of samples of the cardiac signal in the analysis interval are transferred to block 13, in which an ensemble of cardio-oscillations is formed. The formation of the ensemble is as follows. The sequence of peaks is sorted. If the peak is separated from the left or right edge of the analysis interval by an amount exceeding the whole half of the average period of the cardio-oscillations, then in the matrix of the ensemble of cardio-oscillations, as a row matrix corresponding to the ensemble element, the segment of the cardiosignal that is separated from the peak (left and right) by a value equal to the integer part of half the average period of cardiooscillations. The foregoing is illustrated in Fig.6.a.

In the absence of dominance of cardiooscillations, the formation of the ensemble is carried out similarly. The difference is that the threshold calculation unit 9 determines the median q _0.50 , and not the quantile q _0.95-0.97 - An example of a thus formed ensemble of cardio-oscillations is shown in Fig.6.b.

The choice of the average period as the analysis interval (both in the case of dominance and in the absence thereof) has the advantage (as shown by specially performed studies) that in this case the maximum expressiveness of eigenvectors is observed. The foregoing is more relevant to the case of peak dominance. The situation is not excluded when the interval not equal to the average period is selected as the interval over which the ensemble of cardiooscillations is formed. To do this, the device provides for a fixed period selection mode (both with reference to the peak, when it is placed in the middle of the ensemble element, and without reference).

Table 2 describes in sufficient detail the operation of each of the blocks included in the cardioigenoscope.

Table 4 N block Name and description of the unit one

Cardiooscillation Ensemble Formation Unit: Forms a Matrix

2 Mixed moment matrix calculator block: calculates the matrix

3 Block of the calculator of eigenvectors and eigenvalues: determines φ _i , and λ _i satisfying the relation

four Block analysis of signs: provides information for visual qualitative and quantitative analysis of eigenvectors and the spectrum of eigenvalues, restores the cardiac signal. 5 ADC: converts an analog signal supplied to its input into a digital one (it may be absent if the Cardioigenoscope is a prefix to a cardiograph with a digital output). 6 Sample storage unit for the analysis interval: provides storage of samples

arriving at the input of block 6 and their transfer to other blocks through the program control block 14. 7 Centering unit: calculates on the analysis interval N the average value for a sequence of samples received at its input

according to the formula

and calculates the centered sequence according to the formula

8 Invert block: provides sequence multiplication

, by (-1), if necessary for the formation of the normal form (peaks up) of the cardiac signal.

Table 4 (continued) N block Name and description of the unit 9 Threshold calculation unit: determines a quantile of the order of 0.95 or 0.97 for the analyzed segment of the cardiosignal (the case of peak dominance) or the median (case of the absence of dominance). 10 Block for calculating the asymmetry coefficient; Calculates the asymmetry coefficient γ _i for the analyzed segment of the cardiosignal. The obtained value is transmitted to the program control unit 14, in which it is compared with a threshold value of 2. If the threshold is exceeded by an asymmetry coefficient, a decision is made about the presence of peak dominance; otherwise, about his absence. eleven

, то в первую последовательность записывается i. Если

, то во вторую последовательность также записывается число i. При доминировании пиков z=0.95-0.97, при отсутствии доминирования z=0.5. x_i - отсчеты.Positioning unit: Generates two sequences of sample numbers in the analyzed segment of the cardiosignal. If

, then i is written in the first sequence. If

, then the number i is also written in the second sequence. With the dominance of peaks, z = 0.95–0.97; in the absence of dominance, z = 0.5. x _i are the samples. 12 Block for calculating the average period of cardio-oscillations: determines the intervals t _i ,. between neighboring peaks of cardioscillations and then on their basis calculates the average period according to the formula

,. where M is the number of intervals between the peaks of cardiac oscillations in the analysis interval N, ceil (.) is the integer part of the number in brackets.

Table 4 (continued) N block Name and description of the unit 13

и последовательности моментов дискретного времени на интервале анализа, соответствующих пикам кардиоосцилляций t_i, а также последовательности отсчетов приведенного к «нормальной» форме (с положительными пиками)кардиосигнала

формирует ансамбль кардиосцилляций. Каждый элемент ансамбля представляет собой матрицу-строку, получаемую из

как сегмент

, в котором номера соответственно равны:

. Где i-те номера отсчетов максимумов (пиков) кардиосигналов, для которых A_i. и B_i, не выходят за границы интервала анализа

- коэффициент пропорциональности, который как правило равен 0.5. Таким образом сформированная матрица

поступает на выход блока 1.Cardiooscillation Ensemble Matrix Formation Unit: Based on the Average Period of Cardiooscillation Following

and the sequence of discrete time instants in the analysis interval corresponding to the peaks of cardio-oscillations t _i , as well as the sequence of samples reduced to the “normal” form (with positive peaks) of the cardiac signal

forms an ensemble of cardioscillations. Each element of the ensemble is a row matrix obtained from

as a segment

, in which the numbers are respectively equal:

. Where i are the numbers of samples of the maximums (peaks) of the cardiosignals for which A _i . and B _i , do not go beyond the analysis interval

- coefficient of proportionality, which is usually equal to 0.5. Thus formed matrix

arrives at the output of block 1. fourteen Program control block: Provides time-consistent operation of blocks 5-13. Interacts with feature analysis unit 4. fifteen Memory block: provides storage of data used in visualization, program and manual control. 16 Program control block: ensures the coordinated operation of the Cardioigenoscope blocks and the software implementation of the algorithms.

Table 4 (continued) N block Name and description of the unit 17 Manual control unit: provides a manual change of parameters and operating modes of a cardioigenoscope. eighteen Block of visualization of eigenvectors and eigenvalues: provides (at the request of the user) visualization within a variable set of graphical interfaces. 19 Block of calculation of statistics of eigenvectors: provides the calculation of statistical characteristics necessary for the functioning of the block of visualization of eigenvectors and eigenvalues. twenty Cardiooscillation recovery unit: provides restoration of cardiooscillations in accordance with the ratio (5).

Here is a description of the device shown in figure 2, in action, supplementing the information given in table 2.

The block for the formation of the ensemble of cardiac oscillations 1 acts as follows:

1. The block converts segments of a cardiosignal of a certain duration (examples are shown in figures 5.a and 5.b - for the case when the segment is represented by 10,000 samples, which at a sampling frequency of 1 kHz corresponds to a duration of a segment of a cardiosignal of 10 seconds, into ensembles of cardiac oscillations. Each ensemble corresponds to a rectangular matrix, each row of which contains information about individual cardiac oscillations.The row length is defined as the average value of the sequence of R peaks (if they dominate) or the average period of the intersection eny cardiosignals midline if R peaks dominiruyut.Vizualno ensemble of these two cases is illustrated in figure 6.b. ba and has its own line In these figures, each line.

2. If R peaks dominate (as shown in figure 5.a), then the position of each of the R peaks is determined, as shown in the lower figure of figure 17. The number of R peaks in the analyzed signal segment is counted, and the number of samples between the first and last R peaks. The number of intervals between R peaks is calculated (it is equal to the number of peaks minus one) and the number of samples between the first and last R peaks in the interval (by subtracting from the discrete number corresponding to the last R peak, the discrete number corresponding to the first R peak). Then, by dividing thus a certain number of samples by the number of peaks, the average period between R peaks is determined. This value is halved and rounded (using any rounding rule). From each R peak, the integer number of samples obtained in the above manner is laid off to the left and to the right; if such a postponement operation succeeds (it may turn out to be unsuccessful for the first and last R peaks), then we get the boundaries of the corresponding row of the matrix described in paragraph 1. 3. If R peaks do not dominate, then the operations of forming the rows of the matrix

occur similarly to those described in clause 2 with the difference that instead of the position of the maxima of the R peaks, discrete moments of the signal transition through its median value are determined (either only from the bottom up, or only from top to bottom - it does not matter which option will be selected when implementing the block)

The above description of the action of block 1 corresponds to the first paragraph of the formula. This block may have a specific implementation presented in the formula by paragraph 2. The description of this specific option is given below.

An analog-to-digital converter 5, controlled by the program control unit 14 (which sets the conversion parameters, which include the sampling interval) transfers the digitized samples to the sample storage unit on the analysis interval, which has two-way communication with the program control unit 14. The program control unit sets analysis intervals (which, as mentioned above, can include from several to several dozen periods of heartbeats), and also receives a sample from block 6 digitized samples on the analysis interval and transfers it to the centering unit 7; the centered sample is transmitted through the program control block to the inverter block, where, if necessary, the sample is inverted, after which the sample is sent through the program control block to block 6 for further storage and the asymmetry coefficient calculation unit 10. Block 10 calculates the asymmetry coefficient for the entire sample and transfers its value to the program control unit. Next, the program control unit 14 requests a sample that was previously processed by blocks 7 and 8 and is stored in block 6 and transfers it to the threshold 9 calculation unit. If the value of the asymmetry coefficient determined by block 10 is higher than a certain predetermined threshold value (a value of 2 is recommended) , then the program control unit 14 transmits a control signal to the threshold calculation unit, which initiates a threshold calculation in this block, the value of which is taken to be quantiles of a sufficiently high order (0.9 is recommended - 0.97). If the value of the asymmetry coefficient determined by block 10 is lower than the predetermined threshold value, then the median value (quantiles of the order of 0.5) is taken as the threshold. The value of the thus calculated threshold is transmitted to the program control unit 14, which transmits this value together with the sample that was previously processed by blocks 7 and 8 and stored in block 6, to the position determination unit 11. The position determination unit 11 analyzes adjacent pairs of samples received in it samples and if the first sample in the pair is less, and the second sample is greater than the threshold value received in the block, the value of the second sample is stored. After analyzing the entire sample, the sequence of samples that block 12 remembers is transmitted to the program control unit 14, which transmits them to the average period calculation unit 12. The average period calculation unit 12 arranges the obtained sequence in ascending order, calculates the differences of adjacent elements and calculates their arithmetic average , which is then rounded to an integer value (any rounding method is used). The average value obtained in this way is transmitted to the program control unit 14, which transmits it together with the sample, was previously processed by blocks 7 and 8 and stored in block 6, and together with the sequence of discrete numbers received from the position determination unit, into the block for generating the cardio-oscillation ensemble 13. At the same time, a signal is transmitted to block 13 from block 14 that the asymmetry coefficient has exceeded the set threshold. If a signal is received that indicates that the coefficient of asymmetry exceeds the threshold (which indicates the dominance of R peaks), the formation unit of the matrix of the ensemble of cardiac oscillations divides the average period coming from block 14 in half, followed by rounding (using any rounding method), then from each the values of the sequence (formed by the position determination unit 11) are subtracted half the average period, and also half the average period is added to each value - as defined yayutsya oscillation boundaries for the case of R dominating peaks. If the boundaries do not go beyond the analysis interval, then these boundaries are considered admissible and are used further to form the ensemble. For each of the permissible boundaries, a sequence of sample samples is determined (which is stored in block 6 after centering and, if necessary, inverting); each of these sequences is considered a line of the ensemble of cardiac oscillations (the order of the lines in the matrix does not matter). If a signal corresponding to an asymmetry coefficient that is less than the threshold (which corresponds to the absence of dominance of R peaks) is received from the program control lock 14, then the sample segments starting from the count following the lower boundary of the next interval (starting from the first) to its upper boundary (all intervals are considered valid). The ensemble matrix obtained in this way is transmitted through the program control block to the input of the mixed moment matrix calculation block 2.

All actions carried out by blocks 7-13 (in cooperation with block 6) under the control of block 14 should be performed in less than the selected analysis interval. During the execution of the above actions, block 6 under the control of block 14 continues to receive information from block 5 - in this way a sample of discrete digital samples for the next analysis interval is formed. After receiving the next sample, the cycle of the above processing is repeated.

Block 2 for calculating the matrix of mixed moments in the simplest case calculates the matrix of the second mixed off-center moments (covariance matrix) according to the relation given in Table 2. The block for calculating the eigenvectors and eigenvalues calculates the eigenvectors and eigenvalues in any known manner. Blocks 2 and 3 must perform actions for a time not exceeding the analysis interval. If the total execution time of operations (actions) in blocks 1,2 and 3 does not exceed the analysis interval (which is quite feasible with the current state of the art), then information in block 4 arrives with a delay of one analysis interval, otherwise this delay is two intervals analysis (which is quite acceptable for most uses of cardio equipment).

The cardiosignal recovery and sign analysis unit 4 implements the following actions:

1. Removal of any uncorrelated components from cardiac oscillations (elimination of uncorrelated components)

2. Visualization of the initial ensemble of cardio-oscillations, eigenvectors, the spectrum of eigenvalues, the ensemble of cardio-oscillations with the removed (eliminated) uncorrelated components and their statistics

3. Program control and manual control of the elimination of components, visualization, storage of results, as well as with the environment (for the environment of an aigenoscope and the connection of an aigenoscope with the environment, see the response to expert comment 7) of a cardioigenoscope, which receives visual and non-visual information from it and controls the parameters Eigenoscope through the program control unit.

Description of the drawings.

Figure 1. Structural diagram of the Eigenvector Analyzer and signal components (Aigenoscope [1]) according to claim 1 of the utility model 116242RU: 21 is a scaling unit, 2 is a calculator block of the matrix of mixed moments, 3 is a calculator block of eigenvectors and eigenvalues, 22 is a calculator block scalar products and Cardioigenoscope signs.

Figure 2. Cardioeigenoscope: 1 - block for generating an ensemble of cardiooscillations, 2 - block for calculating a matrix of mixed moments, 3 - block for calculating eigenvectors and eigenvalues, 4 - block for restoring a cardiosignal and analysis of symptoms.

Figure 3. Block for generating an ensemble of cardiac oscillations: 5 - ADC, 6 - block for storing samples on the analysis interval, 7 - centering block, 8 - inverter block, 9 - threshold calculation block, 10 - asymmetry coefficient calculation block, 11 - position determinant block, 12 - block computing the average follow-up period of cardio-oscillations, 13 — block for the formation of the matrix of the ensemble of cardio-oscillations, 14 — program control block.

Figure 4. Cardiac signal recovery and feature analysis unit: 15 - memory unit, 16 - program control unit, 17 - manual control unit, 18 - visualization unit, 19 - statistics calculation unit, 20 - cardio-oscillation recovery unit.

Figure 5. Examples of a cardiosignal segment for peak domination modes - 5.a and lack of domination - 5.b.

6. Examples of the ensemble of cardiooscillations for the regimes of peak dominance - B.a. and lack of dominance - 6.b.

7. Screen form for visualizing the normalized spectrum of eigenvalues (expressiveness of eigenvectors in%) for peak domination modes - 7.a and lack of domination - 7.6:

upper left window - the first 10 normalized eigenvalues in% (expressiveness),

upper right window - all normalized eigenvalues exceeding 10 ^-5 ,

bottom left window - the accumulated sum of the first 10 normalized eigenvalues in% (cumulative expressiveness),

the lower right window is the accumulated sum of all normalized eigenvalues in%, (cumulative expressiveness) that exceed 10 ^-5 .

Fig. 8. The screen form for visualizing the first four eigenvectors for the dominance mode is 8.a and the mode of absence of dominance is 8.b. The windows above the graphs indicate the number of the eigenvector and its expressiveness in%.

Fig.9. Screen form for analyzing the time series of expressions.

In the upper left window - a cardiac signal for the current analysis interval,

in the upper right window - the sequence of expressiveness values of the first four eigenvectors (in%) for the sequence of analysis intervals,

in the lower left window - the first 10 eigenvalues in% (expressiveness) for the current analysis interval,

in the lower right window is the accumulated sum of the first 10 eigenvalues in% (cumulative expressiveness) for the current analysis interval.

Figure 10. Screen form for analyzing variations of the first eigenvector. In the right window, the first eigenvector indicating the zones of instability (shown by vertical sights). To determine the zones of instability, a graph of the geometric mean value of the modules of 2-5 eigenvectors is used (shown in the left window). According to the graph of the geometric mean, the quantile value of the specified order is calculated (the case of a quantile of the level of 97% is shown on the left), then the regions where the geometric mean value exceeds the quantile value are calculated, which are indicated on the right window as instability zones.

11. Screen form for visualizing reconstructed cardiac oscillations. The last 9 cardioscillations are shown (on a specific analysis interval, consisting of 13 oscillations) reconstructed using the first 4 eigenvectors.

Fig. 12. Screen form for visualizing reconstructed cardiac oscillations. The last 9 cardioscillations are shown (on a specific analysis interval, consisting of 13 oscillations) reconstructed using the first 2 eigenvectors.

Fig.13. Typical GUI of a cardioigenoscope for displaying a cardiac signal with dominant peaks. The first screen form - Fig.13.a:

upper left window is the first eigenvector;

lower left window - expressiveness of eigenvectors;

upper right window - time series of expressiveness of the first eigenvector

(upper graph) and the total expressiveness of all other eigenvectors (lower graph);

the lower right window is the time series of the asymmetry coefficient. The second screen form - Fig.13.6:

upper window — the cardiac signal in the analysis interval reconstructed from the first four eigenvectors;

the bottom window is the cardiosignal in the analysis interval reconstructed from the first two eigenvectors.

Fig.14. A typical graphical interface of a cardioigenoscope for displaying a cardiosignal in a mode of absence of dominance. The first screen form - Fig.a:

upper left window - the first and second eigenvectors;

lower left window - expressiveness;

upper right window - time series of expressiveness of the first eigenvector

(upper graph) and the total expressiveness of all other proper

vectors (bottom graph);

the lower right window is the time series of the asymmetry coefficient. The second screen form - fig.b:

upper window — the cardiac signal in the analysis interval reconstructed from the first two eigenvectors;

bottom window — components corresponding to the first four eigenvectors are eliminated from the cardiosignal, then a basis of eigenvectors is constructed for the signal obtained in this way, and using the first two of the newly constructed eigenvectors, the high-frequency component is detected.

Fig.15. Estimates of the probability densities of the values of the criteria of dominance from table 1:

15.a is a criterion using an asymmetry coefficient; 15.b - criterion K _m ;

15.c - K _med criterion; 15.d - criterion K _q .

Fig.16. The distribution function of the value of the cardiosignal over an interval of 10,000 samples for two cases: 16.a - peaks of cardio-oscillations dominate; 16.b - peaks of cardiooscillations do not dominate; on the right graphs, with magnification, the regions of large values of the distribution function are shown. The circle indicates the point of the distribution function corresponding to the quantile q _0.95 .

Fig.17. The segment of the analyzed cardiac signal (upper window) and the moments of transition of the quantile values q _0.95 from bottom to top and top to bottom (bottom window).

Fig. 18. The energy contribution of the uncorrelated components of the cardiosignal when using block 1 in a cardioigenoscope (curve 1) and in the prototype (curves 2-4):

1 - The covariance matrix is formed from an ensemble of cardiac oscillations when the R peak has a fixed position with the duration of the analysis interval (cardioigenoscope), which coincides with the average period of R-peaks (the first 2 components have approximately 83% contribution to energy)

2 - The covariance matrix is formed (as in the prototype) by the trajectory of the matrix, which in turn is obtained from successive intervals having a duration that coincides with the average period of R-peaks (the first 2 components have approximately 55% contribution to energy)

3 - The covariance matrix is formed as in claim 2, but the analysis interval is increased by approximately 10% (the first 2 components have approximately 46% contribution to energy)

4 - The covariance matrix is formed as in claim 2, but the analysis interval is reduced by approximately 10% (the first 2 components have approximately 46% contribution to energy) The abscissa axis shows the number of the uncorrelated component (the components are numbered in accordance with the decrease in their energy contribution) . The ordinate is the relative energy contribution (the fraction of the energy of the cardiac signal corresponding to the uncorrelated component).

Industrial applicability of utility model.

A cardio-eigenoscope can be produced in the form of a hardware-software device or in the form of a virtual device that complements a conventional cardiograph and communicates with it through its standard interfaces.

Cardio-eigenoscopy expands the capabilities of cardiac diagnostics and can be used as an additional device, which is part of the stationary and mobile kits, including portable, medical equipment.

Information sources

1. Kozyura A.V. Assessment of the diagnostic significance of the electrocardiographic signal. Proceedings of the X international scientific conference "Physics and Radioelectronics in Medicine and Ecology", Vladimir, 2012, v.1, pp. 152-155.

2. Garcha-NiebIa, J, Technical Mistakes during the Acquisition of the Electrocardiogram / P Llontop-Garcña, JI Valle-Racero, G Serra-Autonell, V.N. Batchvarov, A.B.De Luna // Ann. Noninvasive Electrocardiology. - 2009 .-- No 14 (4). - P.389-403.

3. Shubik, Yu.V. The quality of medical opinions according to the daily monitoring of ECG / Yu.V. Shubik, I.V. Aparina, M.M. Medvedev, A.P. Feldman // Bulletin of Arrhythmology. - CJSC "Institute of Cardiology Technology" ("INCART"). - 2007.- No 49.- P.25-34.

4. Knight BP, Pelosi F, Michaud GF, et al. Clinical consequences of electrocardiographic artifact mimicking ventricular tachycardia // N Eng J Med. - 1999 .-- No. 341. -P.1270-1274.

5. Abokuhall, A., Nielsen, L., Saeed, M., Mark, R.G., Clifford, G. D., Reducing false alarm rates for critical arrhythmias using arterial blood pressure waveform // Journal of Biomedical Informatics. - 2008. - Vol.41, Issue 3. - P.442-451.

6. Ikaro, S., Moody, G. B., Celi, L. Improving the Quality of ECGs Collected Using Mobile Phones: The PhysioNet / Computing in Cardiology Challenge 2011 // Computers in Cardiology. - 2011 .-- No38. - P.273-276.

7. Eigenvector analyzer and signal components. Utility Model 116242RU

8. Modern Russian-English Dictionary of Radio Electronics / comp. M.S. Tiasto, S.N. Tulivertov. - Rostov n / a: Phoenix, 2009 .-- 377 p.

9. A Brief Dictionary of Foreign Words / compiled by S.M. Lokshin. M .: Owl Encyclopedia, 1966 .-- 384 p.

10. The PTB Diagnostic ECG Database - The PTB Diagnostic ECG Database -http: //www.physionet.org/physiobank/database/ptbdb/

Claims (3)

1. A cardioigenoscope containing a block of the calculator of the matrix of mixed moments 2, a block of the calculator of eigenvectors and eigenvalues 3, characterized in that at the input of the device there is a block for forming the ensemble of cardio-oscillations 1, the output of which is connected to the input of the block of the calculator of the matrix of mixed moments 2 and to the first input cardiac signal recovery and feature analysis 4; the output of the block of the calculator of the matrix of mixed moments 2 is connected to the input of the block of the calculator of eigenvectors and eigenvalues 3, the first and second outputs of which are connected to the second and third inputs of the block recovery cardiosignal and analysis of signs 4; the output channel of the cardiosignal recovery and sign analysis unit 4 is two-way and is associated with the environment of the cardioigenoscope. 2. The cardioigenoscope according to claim 1, characterized in that the block for generating the ensemble of cardiooscillations 1 consists of: an ADC input unit - 5, a sample storage unit in the analysis interval 6, a centering unit 7, an inverter 8, a threshold calculation unit 9, a coefficient calculation unit asymmetries 10, position determiner block 11, calculator for the average period of cardio-oscillations 12, block for forming the matrix of the ensemble of cardio-oscillations 13, each of which, individually, is bilaterally connected to the output program control unit 14; wherein the first output of the ADC - 5 unit is additionally connected to the input of the sample storage unit in analysis interval 6. 3. A cardioigenoscope according to claim 1, characterized in that the cardiosignal recovery and feature analysis unit 4 consists of an input memory unit 15, which is bi-directionally connected to the output program control block 16, to which, in turn, are bi-directional: manual control unit 17, a visualization unit 18, a statistics calculation unit 19, a cardio-oscillation recovery unit 20, the output of the program control unit 16 is two-sided and connected with the environment of the cardioigenoscope.

Images