RU162110U1 - ECG STORAGE AND ANALYSIS - Google Patents

Abstract

An ECG storage and analysis device containing a cardioigenoscope 3, characterized in that the two-way input of the device is a two-way input of the authorization and access unit 1, the first two-way output of which is the first two-way input of the unit for setting operating modes, compression, storage and selection 2, whose two-way output is a two-way device output; the second two-sided output of the authorization and access unit 1 is the input of the cardioigenoscope 3, the two-sided output of which is the second two-way input of the unit for setting operating modes, compression control, storage and selection 2.

Description

The field of technology.

The utility model "ECG Storage and Analysis Device" (ECA ECG) refers to medical cardiological devices and is a device for analyzing and storing ECG in a compressed form without loss of information. ECA ECG allows you to repeatedly compress ECG information, separately analyze its uncorrelated components, and filter the ECG from interference. ECA ECG is implemented in the form of a hardware-software complex, access to which end users is via standard communication channels (including via the Internet).

The level of technology. Analogs and their shortcomings.

ECA ECG is based on processing a sequence of synchronized PQRST complexes [1] of the initial ECG (hereinafter, the synchronous ensemble — SA). The ACHA ECG uses an orthonormal basis (BSS) [8] of the eigenvectors (CB) of the covariance matrix (KM) [8], which makes it possible to best represent the PQRST complex on the finite analysis interval (KIA) [9].

The representation of time series (BP) on the KIA in the bases of the CB covariance matrices with emphasis on the properties of the eigenvectors is called “aigenoscopy”; structures carrying out aigenoscopy are enshrined in six patents of the Russian Federation for utility models - [2-7].

The proposed utility model uses the design embodied in the utility model [3] "Cardioigenoscope".

In known systems, including holter monitoring [11], a digitized ECG record is stored, for compression of which (after recording) standard means of data compression and archiving can be used, requiring archiving before using the data.

The proposed design allows you to store instead of a full ECG record only your own ECM ECG vectors and a small volume of numerical information, providing ECG recovery with a given accuracy and without loss of information.

The technical effect in the design proposed in the utility model (PM) is achieved primarily through the use of compression approaches based on the ECG compression model; therefore, this utility model is considered as the closest analogue. The design of a cardioigenoscope is shown in FIG. one.

Disclosure of a utility model.

Purpose and advantages: use the advantages of a cardioigenoscope [3] for adaptive (with specified accuracy) controlled compression of the ECG (without loss of information) for the purpose of subsequent storage, recovery, and analysis. The ACHA ECG provides a compression ratio (the ratio of the memory required to store the original ECG to the memory required for storage in a compressed form using the ACHA ECG) proportional to the number of PQRST complexes in the ECG; the use of standard data storage systems when implementing the ACHA ECG design allows storing at least 100 million compressed ECGs per terabyte of read-only memory.

The result is achieved due to the fact that for the additive (in the form of the sum of components) representation of the elements of the SA ECG, an orthonormal basis of the covariance matrix SV is used. Of all the possible options for additive representations, this option provides the smallest number of additive components to achieve the highest possible accuracy (for a given number of components). Each of the components is orthogonal (independent) [8] from all the others and is completely determined by the correlations of the elements of the PQRST complex. This is what allows you to compress ECG information as much as possible - practically without loss of information.

The structure of the ACHA ECG includes:

- authorization and access unit 1;

- unit for setting operating modes, compression control, storage and selection 2;

- block 3 - cardioigenoscope, the design of which is made in accordance with patent No. 128470 RU for a utility model.

Cardioigenoscope (Fig. 1) includes:

- block formation of the ensemble of cardiooscillations 4,

- block calculator matrix of mixed moments 5,

- block calculator of eigenvectors and eigenvalues 6,

- block recovery of cardiac signal and analysis of signs 7.

формируется матрица синхронного ансамбля W_N×R=[S_i,j]_N×R с размером N×R, где R - число элементов СА (число PQRST-комплексов в обрабатываемом сегменте ЭКГ). Все эти операции выполняются в блоке формирования ансамбля кардиоосцилляций 4.A cardioigenoscope [3] processes successive segments of a discretized and digitized ECG, into which dozens of PQRST complexes with a given duration of N discrete fit. The signal of the PQRST complex on the j-th KIA is set by a sequence of discrete samples written in the column matrix X _j = [S _{1, j} ; S _{2, j} ; ...; S _{N, j} ] so that the R-peaks of the ECG are tied to a certain a predetermined discrete time instant on the KIA. The size of the KIA N is chosen to be known to be a long duration of the PQRST complex. Of the columns X _j ,

the matrix of the synchronous ensemble W _{N × R} = [S _{i, j} ] _{N × R} with the size N × R is formed, where R is the number of CA elements (the number of PQRST complexes in the processed ECG segment). All these operations are performed in the block for the formation of the ensemble of cardiooscillations 4.

Based on the matrix of the ensemble W _{N × R} , the CM CA determined by the relation

where (...) ′ - means transposition of the matrix.

For KM (1) are found SV and eigenvalues (SZ) satisfying the relation [8]

where ψ _i is an eigenvector (column matrix),

λ _i - eigenvalue (number).

. СВ КМ образуют ОНБ [8], следовательно,

, если i≠j и

.The number of SW and SZ coincides with the size of the CM and is equal to N - [8]. CM is symmetric and non-negative definite [8], that is

. SW CM form ONB [8], therefore,

if i ≠ j and

.

All R PQRST complexes written as columns in the matrix W _{N × R}

can be represented as

Where

Ψ _{N × N} = [ψ ₁ , ..., ψ _N ] is a square matrix into which all CBs are written as columns in order,

- матрица коэффицентов разложения кэлементов СА на КИА, j-ый столбец которой

- matrix of coefficients of decomposition of CA elements into KIA, j-th column of which

Relations (3) and (4) give an accurate representation of PQRST complexes.

When compressing it is enough to use the number of CB Z <N; usually Z = 4 is enough. In this case, relations (3) and (4) are simplified and have the form:

Where

- аппроксимация W_N×R,

- approximation of W _{N × R} ,

α _{Z × R} is the matrix of approximation coefficients of CA using Z <N eigenvectors.

It is convenient to present information on the SZ obtained by solving problem (2) in the form of a spectrum of eigenvalues (GCC), which is understood [3, 9] as a descending order of SZ. It is known [3] that the average energy of a signal observed in the interval N is determined by the relation

средней энергии составляющей, соответствующей СВ ψ_i. Из этого (с учетом ортонормированности базиса СВ) следует, чтоand each eigenvalue λ _i is equal to the contribution to

the average energy of the component corresponding to CB ψ _i . From this (taking into account the orthonormality of the ST basis), it follows that

- среднее по СА,Where

- average for CA,

- среднее по СА значение α_i,

- average SA value α _i ,

- выборочная дисперсия α_i по СА.

- sample variance α _i for CA.

We will call the normalized spectrum of eigenvalues (NCCS), the sequence

- нормированное собственное значение (НСЗ),Where

- normalized eigenvalue (NHA),

N is the dimension of KM (the size of the KIA).

в

.The physical meaning of the NHA is as follows. Each NHA value is the proportion of the uncorrelated component corresponding to

at

.

Since each normalized eigenvalue (NHA) expresses information about the fraction of energy that is explained by the corresponding SV, we will call the value of the NSV corresponding to SV the expressiveness of SV [3]. Further we will indicate expressiveness in%. The total value of the expressivities of all eigenvectors is 100%. The higher the expressiveness of the SV, the more information it carries.

The orthonormal basis given by the SW of the covariance matrix has the largest possible expressiveness [9], that is, representing a PQRST complex as a sum of eigenvectors requires a minimum number of additive components for a given fixed error.

We also note that KM (2) can be represented through its NE and NW as follows [9]

Where

- квадратные неотрицательно определенные матрицы размера N×N, полностью определяемые соответствующими СВ.

- square non-negative definite matrices of size N × N, completely determined by the corresponding CB.

Relation (8) is interesting in that it allows the relationship between the BP samples on the KIA (these relationships are described by the CM) to be decomposed into the sum of the independent λ _i · U _i relationships, which are completely determined by the corresponding NE and SZ.

.In most cases, four CBs and four decomposition coefficients are sufficient to present without loss of information on the KIA of any PQRST complex; If it is necessary to restore the CM, SZ will also be added. To illustrate this fact in FIG. 3 presents data from the description of the PM [3]. An analysis of the NHA shown in the lower left window of FIG. 3a shows that

.

Utility Model Design.

The design of the utility model is shown in FIG. 2.

Authorization and access unit 1 is a server that ensures the implementation of the law on personal data and authorized access to the ECA ECG. Information from the second output of this unit is fed to the input of the cardioigenoscope 3, the design of which meets the patent for PM No. 128470 RU.

In the cardioigenoscope 3, the position of the R peaks is determined, the matrix of the ensemble W _{N × R is formed} , CB, C3 (in accordance with relations (5) and (6)) and NCCC are calculated, Z decomposition coefficients for each element of the ensemble are calculated. For compression and subsequent recovery of SA, CM and ECG (containing R PQRST complexes) are used:

- CBs grouped into a matrix Y _{N × Z} (N × Z - numeric values),

- row matrix SZ [λ ₁ , ..., λ _Z ] (Z - numeric values),

- matrix of approximation coefficients SA α _{Z × R} (Z × R - numerical values),

- a matrix row of the positions of the R-peaks in the ECG [τ ₁ , ..., τ _R ] (R - numerical values),

- one number Z.

These data are fed through the two-way output of the cardioigenoscope 3 to the first two-way input of the unit for setting operating modes, compression control, storage and sampling 2. At the same two-sided input of the unit for setting operating modes, compression control, storage and sampling 2, the cardiogenoscope is controlled, in particular, R, Z values and the position of the R peak at the KIA.

The number of CB - Z, which is necessary to ensure the given accuracy of the ECG representation in the block of the unit for setting the operating modes, compression control, storage and sampling 2 and determining the compression mode, is selected so that the sum of the first Z normalized eigenvalues exceeds the specified ECG restoration accuracy (see explanations formula (7)). That is, the value of Z is determined directly from the graph of cumulative expressiveness (see the lower right window in Fig. 4; as can be seen from the graph, for this particular session it is enough to select Z = 3 for accuracy)

The unit for setting operating modes, compression, storage, and selection control 2 is connected at its second two-way input to the first two-way output of the authorization and access unit 1. for setting operating modes, compression control, storage and selection 2), they get access to the unit for setting operating and control modes compression, storage and selection 2 through the first two-way output of the authorization and access unit 1, connected to the second two-way input of the unit for setting operating modes, compression, storage and selection 2.

As means of access to the authorization and access unit 1, users use the means and communication channels provided for by the design of this unit.

Evidence of technical results. Causal relationships.

As a result of compression in block 3 of the ECG segment of length _ECG = R · N, a certain number of numerical values arise, which is determined by the ratio

Where

Z · N is the number of numerical cells required to store Z CB,

Z is the number of numerical cells required to store Z SZ,

Z · R is the number of number cells needed to store the matrix of coefficients α _{Z × R} ,

R is the number of numerical cells required to store the positions of the R-peaks,

1 - a numerical cell for storing the value of Z.

Attitude

The ECG compression coefficient is determined by the inverse ratio (10) and is equal to

Given that usually Z≤4, the compression ratio usually exceeds the number of stored PQRST complexes, divided by four.

At N = 100, R = 100, and Z = 4

N _EAR = Z · N + Z + Z · R + R + 1 = 4 · 100 + 4 + 4 · 100 + 100 + 1 = 905 numbers. Their storage, for example in the Scilab system - [10], requires 10 Kbytes. Thus, using standard software, the proposed design allows storing at least 100 million compressed ECGs per terabyte of read-only memory of the storage, sampling and control unit 2.

The main technical result is a significant compression of information. This result is explained by the inclusion in the composition of the ECG storage and analysis device of a cardioigenoscope, which provides compression and recovery due to the use of the RS BSS covariance matrix SA.

Description of the drawings.

FIG. 1. The design of a cardioigenoscope according to Patent No. 128470 RU for a utility model: 4 - a unit for generating an ensemble of cardio-oscillations, 5 - a unit for calculating a matrix of mixed moments, 6 - a unit for calculating eigenvectors and eigenvalues, 7 - a unit for restoring a cardiosignal and analyzing symptoms.

FIG. 2. ECG storage and analysis device:

1 - authorization and access unit;

2 - unit for setting operation modes, compression control, storage and selection;

3 - cardioigenoscope.

FIG. 3. (corresponds to FIG. 13 of the description [3]). A typical graphical interface of a cardioigenoscope for displaying an ECG.

The first screen form is FIG. 3.a (corresponds to FIG. 13a of the description [3]):

upper left window is the first eigenvector;

lower left window - expressiveness of eigenvectors;

The upper right and lower windows are not of interest for consideration in the framework of the proposed PM.

The second screen form is FIG. 3.b (corresponds to FIG. 13b of the description [3]):

upper window — the cardiac signal in the analysis interval reconstructed from the first four eigenvectors (Z = 4);

the lower window is the cardiac signal in the analysis interval reconstructed from the first two eigenvectors (Z = 2).

FIG. 4. (corresponds to FIG. 9 of the description [3]). Screen form for expressive analysis.

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 - the accumulated sum of the first 10 eigenvalues in%

(cumulative expressiveness) for the current analysis interval.

The implementation of the invention.

ECA ECG is implemented as a server complex, in which blocks 1 and 2 are implemented in any standard form, and a cardioigenoscope is implemented in hardware and software in accordance with utility model patent No. 116242 RU Cardioagenoscope.

Information sources.

1. Strutynsky A.V. Electrocardiogram: analysis and interpretation / A.V. Strutinsky. - 15th ed. - M.: MEDpress-inform, 2013 .-- 244 p.: Ill.

2. Isakevich V.V., Isakevich D.V., Grunskaya L.V. Eigenvector analyzer and signal components. Utility Model No. 116242 RU.

3. Isakevich V.V., Isakevich D.V., Batin A.S. Cardioigenoscope. Utility Model No. 128470 RU.

4. Isakevich V.V., Isakevich D.V., Batin A.S. Reflection Detector. Utility Model No. 128724 RU.

5. Isakevich V.V., Isakevich D.V. Signaling device of significant differences. Utility model No. 133642 RU.

6. Isakevich V.V., Isakevich D.V., Grunskaya L.V., Firstov P.P. Signaling device for changes in the main components. Utility model No. 141416 RU.

7. Isakevich V.V., Isakevich D.V., Grunskaya L.V., Rubay D.V., Leshchev I.A. Detector of weak reflections in impulse noise. Utility model No. 141582 RU.

8. Korn G., Korn T. Handbook of mathematics (for scientists and engineers). M .: Nauka, 831 p.

9. Isakevich D.V., Isakevich V.V. Cardioeigenoscope - a new useful model for processing ECG - M. Publishing House Pero, 2014. - 138 p. ISBN 978-5-00086-280-3

10. Scilab CeCILL. http://scilab.org

11. Ryabykina G.V., Sobolev A.V. Holter and bifunctional monetization of ECG and blood pressure. M.: Publishing House Medpraktika-M, 2010, 320 pp.

Claims (1)

An ECG storage and analysis device containing a cardioigenoscope 3, characterized in that the two-way input of the device is a two-way input of the authorization and access unit 1, the first two-way output of which is the first two-way input of the unit for setting operating modes, compression, storage and selection 2, whose two-way output is a two-way device output; the second two-sided output of the authorization and access unit 1 is the input of the cardioigenoscope 3, the two-sided output of which is the second two-way input of the unit for setting operating modes, compression control, storage and selection 2.

Images