CN110025308B

CN110025308B - Electrocardio feature extraction method, heart beat identification method and device

Info

Publication number: CN110025308B
Application number: CN201910283053.1A
Authority: CN
Inventors: 钱涛; 张立明; 谭春雨
Original assignee: University of Macau
Current assignee: University of Macau
Priority date: 2019-04-09
Filing date: 2019-04-09
Publication date: 2021-09-10
Anticipated expiration: 2039-04-09
Also published as: CN110025308A

Abstract

The embodiment of the application provides an electrocardio feature extraction method, a heart beat identification method and a heart beat identification device. The method comprises the following steps: acquiring an electrocardiosignal to be processed, and preprocessing the electrocardiosignal to be processed to obtain at least one piece of first heartbeat information corresponding to the electrocardiosignal to be processed; respectively carrying out self-adaptive Fourier decomposition on each piece of first heartbeat information to obtain a first time-frequency representation corresponding to each piece of first heartbeat information; and extracting corresponding first instantaneous frequency characteristics according to the first time frequency representation corresponding to each piece of first heartbeat information. The device is used for executing the method. The features extracted by the method of the self-adaptive Fourier decomposition in the embodiment of the application are all representative, the type of the first heartbeat information can be well reflected, and redundant useless features are avoided, so that the calculation workload is small.

Description

Electrocardio feature extraction method, heart beat identification method and device

Technical Field

The application relates to the technical field of electrocardiosignal processing, in particular to an electrocardio feature extraction method, a heartbeat identification method and a heartbeat identification device.

Background

An Electrocardiogram (ECG) is a signal that reflects a change in electric potential induced from the body surface by an electrocardiograph, and is a representation of the process of electrical activity of the heart. The electrocardiosignals have important reference value for basic heart functions and pathological researches thereof. The electrocardiosignals are identified and classified by utilizing the automatic diagnosis technology, so that the tedious work of monitoring the electrocardiosignals of patients for a long time by doctors can be reduced, and an important basis is provided for the household portable electrocardio monitoring equipment which is currently developed, so that users can timely find abnormal heart beats through self monitoring, and can timely see a doctor.

This process of feature extraction is important when identifying and classifying cardiac electrical signals. There are many conventional feature extraction methods, for example: fast fourier transform, Hermite decomposition parameters, wavelet transform, and the like, which can extract many features, but some of the extracted features are not useful and cause a problem of a large amount of computation.

Disclosure of Invention

In view of the above, an object of the embodiments of the present application is to provide an electrocardiographic feature extraction method, a heartbeat recognition method, and a heartbeat recognition device, which solve the above-mentioned technical problem that when feature extraction is performed, the amount of computation is large because many useless features are extracted.

In a first aspect, an embodiment of the present application provides an electrocardiographic feature extraction method, including: acquiring an electrocardiosignal to be processed, and preprocessing the electrocardiosignal to be processed to obtain at least one piece of first heartbeat information corresponding to the electrocardiosignal to be processed; respectively carrying out self-adaptive Fourier decomposition on each first heartbeat information to obtain a first time-frequency representation corresponding to each first heartbeat information; and extracting corresponding first instantaneous frequency characteristics according to the first time frequency representation corresponding to each piece of first heartbeat information.

The method comprises the steps of preprocessing electrocardiosignals to be processed to obtain at least one piece of first heartbeat information, and performing self-adaptive Fourier decomposition on each piece of first heartbeat information to obtain a corresponding first instantaneous frequency feature, wherein the first instantaneous frequency feature is a feature corresponding to the first heartbeat information to be extracted.

Further, the performing adaptive fourier decomposition on each piece of the first heartbeat information to obtain a first time-frequency representation corresponding to each piece of the first heartbeat information respectively includes: obtaining a first function corresponding to the first heartbeat information by using Hilbert transform; the first function is used for representing the corresponding relation between the real-value signal and the complex-value signal of the first heartbeat information; by using

Performing first decomposition on the first function by using the regeneration nuclear property of the nuclear and the maximum selection principle to obtain a first decomposition result; decomposing each decomposition result by using a maximum selection principle in sequence from the first decomposition result to obtain a next decomposition result, and decomposing the m-1 decomposition result to obtain a second function; wherein m is a positive integer greater than 1; the second function is used for representing a decomposition result of the complex-valued signal corresponding to the first heartbeat information; obtaining the first time-frequency representation from the second function. According to the embodiment of the application, firstly, the relation between the real-valued signal and the complex-valued signal of the first heartbeat information is constructed through Hilbert transformation, and then the real-valued signal and the complex-valued signal of the first heartbeat information are utilized

The kernel and the maximum selection principle decompose the first function for preset times to obtain a first time-frequency representation, representative features corresponding to the first heartbeat information can be extracted through the obtained first time-frequency representation, and the number of the extracted features is small and representative, so that the calculation workload in the feature extraction process can be reduced.

Further, the performing adaptive fourier decomposition on each piece of the first heartbeat information to obtain a first time-frequency representation corresponding to each piece of the first heartbeat information respectively includes: obtaining a first function s corresponding to the first heartbeat information according to Hilbert transform⁺S + iHs, where s is the real-valued signal of the first heartbeat information, s⁺The first heartbeat information is a corresponding complex value signal, H is Hilbert transform, and i is an imaginary number unit; by using

Regenerative nuclear properties of the nucleus

And carrying out n times of decomposition on the first function by using a maximum selection principle to obtain a second function

Wherein a is_nFor points determined on the unit disc using the maximum selection principle, and a_n＝argmax_a∈D{(1-|a|²)|s_n(a)|²D is the unit disc, B_nIs a rational orthogonal system, and

R_nis the residue after n decompositions, n is 1, 2.; obtaining said first heartbeat information from said second functionThe first time frequency is expressed as:

wherein, c_nIs the coefficient of the nth term, and N is the total number of decompositions. According to the embodiment of the application, firstly, the relation between the real-valued signal and the complex-valued signal of the first heartbeat information is constructed through Hilbert transformation, and then the real-valued signal and the complex-valued signal of the first heartbeat information are utilized

Further, the extracting, according to the first time-frequency representation corresponding to each piece of the first heartbeat information, the first temporal frequency feature corresponding to the first heartbeat information includes: the first instantaneous frequency characteristic obtained according to the first time-frequency representation of the first heartbeat information is as follows:

wherein the content of the first and second substances,

according to the embodiment of the application, the function after transformation is obtained by performing re-expansion transformation according to the first time-frequency representation, and representative features corresponding to the first heartbeat information are extracted from the function after transformation, so that the calculation workload can be reduced through an adaptive Fourier decomposition algorithm.

Further, the preprocessing the to-be-processed electrocardiographic signal to obtain at least one piece of first heartbeat information corresponding to the to-be-processed electrocardiographic signal includes: performing R wave peak detection on the electrocardiosignals to be processed to obtain at least one R wave peak point; and acquiring first heartbeat information corresponding to each R wave peak point according to the at least one R wave peak point and the sampling frequency of the electrocardiosignals to be processed. When the electrocardiosignal to be processed is preprocessed, all R wave peak points included in the electrocardiosignal to be processed are obtained firstly, then first heartbeat information corresponding to each R wave peak point is obtained according to sampling frequency, and due to different sampling frequencies, the number of sampling points included in one complete first heartbeat information is different, so that the complete first heartbeat information can be guaranteed to be obtained by obtaining the first heartbeat information according to the sampling frequency.

In a second aspect, an embodiment of the present application provides a method for identifying a heartbeat, including: acquiring an electrocardiosignal to be identified; according to the electrocardio-feature extraction method of the first aspect, second heartbeat information corresponding to the electrocardiosignal to be identified and second instantaneous frequency features corresponding to the second heartbeat information are obtained; acquiring first preset traditional characteristics corresponding to each piece of second heartbeat information, wherein the first preset traditional characteristics comprise any one or combination of QRS waveform duration, forward RR interval, backward RR interval, average RR interval and amplitude of R wave peak point; identifying a feature vector corresponding to each second heartbeat information by using a pre-trained classifier to obtain a heartbeat category corresponding to each second heartbeat information; wherein the feature vector comprises the second instantaneous frequency feature and the first preset legacy feature; and the classifier performs feature extraction on a training set through the self-adaptive Fourier decomposition algorithm to obtain training features, and the training features are used for training.

According to the method, the second heartbeat information in the electrocardiosignals to be identified is subjected to feature extraction through the electrocardio feature extraction method, the category of each second heartbeat information is identified through the preset classifier, the heartbeat category corresponding to the second heartbeat information is obtained, and the accuracy of identification can be improved through representative feature identification.

Further, the method further comprises: acquiring a plurality of pieces of third heartbeat information, and labeling each piece of third heartbeat information to obtain a corresponding heartbeat label; performing feature extraction on each third heartbeat information according to the electrocardiogram feature extraction method of the first aspect to obtain a third instantaneous frequency feature corresponding to each third heartbeat information; acquiring a second preset traditional feature corresponding to each piece of third heartbeat information, wherein the second preset traditional feature comprises any one or combination of duration time of a QRS waveform, forward RR interval, backward RR interval, average RR interval and amplitude of an R wave peak point; and taking a feature vector formed by the third instantaneous frequency feature and the second preset traditional feature as input, and taking a heartbeat label corresponding to the third heartbeat information as output to perform model training to obtain the classifier. According to the embodiment of the application, the third heartbeat information is subjected to feature extraction through the self-adaptive Fourier decomposition algorithm, and representative features corresponding to the third heartbeat information are used for training, so that the performance of the classifier is improved.

In a third aspect, an embodiment of the present application provides an electrocardiographic feature extraction device, including: the preprocessing module is used for acquiring an electrocardiosignal to be processed, preprocessing the electrocardiosignal to be processed and acquiring at least one piece of first heartbeat information corresponding to the electrocardiosignal to be processed; the decomposition module is used for respectively carrying out self-adaptive Fourier decomposition on each piece of first heartbeat information to obtain a first time-frequency representation corresponding to each piece of first heartbeat information; and the extraction module is used for extracting corresponding first instantaneous frequency characteristics according to the first time frequency representation corresponding to each piece of first heartbeat information.

Further, the decomposition module is specifically configured to: obtaining a first function corresponding to the first heartbeat information by using Hilbert transform; the first function is used for representing the corresponding relation between the real-value signal and the complex-value signal of the first heartbeat information; by using

Performing first decomposition on the first function by using the regeneration nuclear property of the nuclear and the maximum selection principle to obtain a first decomposition result; using the first decomposition result

Regenerative nuclear properties and greatly selected sources of nucleiDecomposing the decomposition result of each time to obtain the next decomposition result, and decomposing the m-1 th decomposition result to obtain a second function; wherein m is a positive integer greater than 1; the second function is used for representing a decomposition result of the complex-valued signal corresponding to the first heartbeat information; and obtaining the first time-frequency representation according to the second function.

Further, the decomposition module is specifically configured to: obtaining a first function s corresponding to the first heartbeat information according to Hilbert transform⁺S + iHs, where s is the real-valued signal of the first heartbeat information, s⁺The first heartbeat information is a corresponding complex value signal, H is Hilbert transform, and i is an imaginary number unit; by using

Regenerative nuclear properties of the nucleus

And the maximum selection principle decomposes the second function for n times to obtain a second function

R_nis the residue after n decompositions, n is 1, 2.; obtaining a first time-frequency representation of the first heartbeat information according to the second function as:

wherein, c_nIs the coefficient of the nth term, and N is the total number of decompositions.

Further, the extraction module is specifically configured to: the first temporal frequency characteristic obtained according to the first temporal frequency representation is:

wherein phi is_n(t) is B_nThe phase function of (a); b is_nIs the rational orthogonal system, and

z＝e^itis a point on the unit circle.

Further, the preprocessing module is specifically configured to: performing R wave peak detection on the electrocardiosignals to be processed to obtain at least one R wave peak point; and acquiring first heartbeat information corresponding to each R wave peak point according to the at least one R wave peak point and the sampling frequency of the electrocardiosignals to be processed.

In a fourth aspect, an embodiment of the present application provides a heartbeat recognition device, including: the acquisition module is used for acquiring the electrocardiosignals to be identified; a first feature extraction module, configured to obtain second heartbeat information corresponding to the electrocardiographic signal to be identified and second instantaneous frequency features corresponding to the second heartbeat information according to the electrocardiographic feature extraction method in the first aspect; a second feature extraction module, configured to obtain a second preset conventional feature corresponding to each piece of third heartbeat information, where the second preset conventional feature includes any one or a combination of a QRS waveform duration, a forward RR interval, a backward RR interval, an average RR interval, and an amplitude of an R-wave peak point; the identification module is used for identifying the feature vector corresponding to each second heartbeat information by using a pre-trained classifier to obtain a heartbeat category corresponding to each second heartbeat information; wherein the feature vector comprises the second instantaneous frequency feature and the first preset legacy feature; and the classifier performs feature extraction on a training set through the self-adaptive Fourier decomposition algorithm to obtain training features, and the training features are used for training.

Further, the device further comprises a classifier building module, which is used for obtaining a plurality of pieces of third heartbeat information, labeling each piece of third heartbeat information, and obtaining a corresponding heartbeat label; performing feature extraction on each third heartbeat information according to the electrocardiogram feature extraction method of the first aspect to obtain a third instantaneous frequency feature corresponding to each third heartbeat information; acquiring a second preset traditional feature corresponding to each piece of third heartbeat information, wherein the second preset traditional feature comprises any one or combination of duration time of a QRS waveform, forward RR interval, backward RR interval, average RR interval and amplitude of an R wave peak point; and taking a feature vector formed by the third instantaneous frequency feature and the second preset traditional feature as input, and taking a heartbeat label corresponding to the third heartbeat information as output to perform model training to obtain the classifier.

In a fifth aspect, an embodiment of the present application provides an electronic device, including: a processor, a memory, and a bus, wherein,

the processor and the memory are communicated with each other through the bus;

the memory stores program instructions executable by the processor, the processor invoking the program instructions to be capable of performing the method steps of the first or second aspect.

In a sixth aspect, an embodiment of the present application provides a non-transitory computer-readable storage medium, including:

the non-transitory computer readable storage medium stores computer instructions that cause the computer to perform the method steps of the first or second aspect.

Additional features and advantages of the present application will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by the practice of the embodiments of the present application. The objectives and other advantages of the application may be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.

Fig. 1 is a schematic flow chart of an electrocardiographic feature extraction method according to an embodiment of the present application;

fig. 2 is a schematic flow chart of an adaptive fourier decomposition process according to an embodiment of the present disclosure;

fig. 3 is a schematic diagram of another adaptive fourier decomposition process provided in an embodiment of the present application;

fig. 4 is a schematic flowchart of a classifier construction method according to an embodiment of the present application;

fig. 5 is a schematic flow chart of a heartbeat identification method according to an embodiment of the present application;

fig. 6 is a schematic flow chart of another heartbeat identification method provided in the embodiment of the present application;

fig. 7 is a schematic structural diagram of an electrocardiographic feature extraction device according to an embodiment of the present application;

fig. 8 is a schematic structural diagram of a heartbeat recognition device according to an embodiment of the present application;

fig. 9 is a block diagram of an electronic device in an embodiment of the present application;

fig. 10 is a block diagram of another electronic device in this embodiment.

Detailed Description

The technical solution in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.

It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures. Meanwhile, in the description of the present application, the terms "first", "second", and the like are used only for distinguishing the description, and are not to be construed as indicating or implying relative importance.

Before the application, the traditional method for extracting the electrocardio characteristics has the problems of overlarge operation amount and high calculation cost. Moreover, the feature vector dimension is too high, so that the operation amount of the classifier is increased. Therefore, the electrocardiosignal is preprocessed to obtain the included heartbeat information, and then the representative time-frequency features are extracted from the heartbeat information by the self-adaptive Fourier decomposition algorithm, so that the quantity of extracted useless features is reduced, and the calculated quantity is small. After the time-frequency characteristics of the heart beat are extracted, the time-frequency characteristics are identified through the classifier, the operation amount of the classifier is low, and the heart beat category corresponding to the heart beat can be accurately obtained.

The method for extracting the cardiac electrical characteristics, and the heartbeat recognition method will be described below by the following embodiments.

Fig. 1 is a schematic flow chart of an electrocardiographic feature extraction method provided in an embodiment of the present application, and as shown in fig. 1, the method includes:

step 101: acquiring an electrocardiosignal to be processed, and preprocessing the electrocardiosignal to be processed to obtain at least one piece of first heartbeat information corresponding to the electrocardiosignal to be processed.

In a specific implementation process, when feature extraction needs to be performed on an electrocardiographic signal to be processed, the electrocardiographic signal to be processed can be input into a feature extraction device, and the feature extraction device performs preprocessing on the electrocardiographic signal to be processed after acquiring the electrocardiographic signal to be processed. The purpose of the preprocessing is to divide the electrocardiosignals to be processed to obtain complete heartbeat signals included in the electrocardiosignals to be processed, and for convenience of distinguishing, in the embodiment of the application, heartbeat information obtained after preprocessing the electrocardiosignals to be processed is called first heartbeat information. It should be noted that the number of first heartbeat information obtained after preprocessing depends on the number of complete heartbeat information included in the cardiac signal to be processed. For example, only one complete first heartbeat information is included in the electrocardiographic signal to be processed, and the number of the acquired first heartbeat information is 1. Furthermore, when the first heartbeat information is acquired, all the heartbeat information in the electrocardiographic signal to be processed may be taken as the first heartbeat information, or one heartbeat may be taken every predetermined number of heartbeats, and the acquired heartbeat information may be taken as the first heartbeat information.

It is understood that the normal cardiac potential conduction process to produce an electrocardiogram waveform may include five parts: p-wave, P-R interval, QRS complex, ST segment, and T-wave. Sometimes, a wavelet with the same direction as the T wave may exist after the T wave is ended, and the wavelet is called as a U wave. A complete first heartbeat message includes the above-mentioned parts. Thus, the first heartbeat information is ECG signal waveform information corresponding to a complete heartbeat.

Step 102: and respectively carrying out self-adaptive Fourier decomposition on each first heartbeat information to obtain a first time-frequency representation corresponding to each first heartbeat information.

In a specific implementation process, for the condition that one piece of first heartbeat information is acquired in step 101, performing adaptive fourier decomposition on the first heartbeat information to acquire a corresponding first time-frequency representation; for the case that a plurality of pieces of first heartbeat information are acquired in step 101, performing adaptive fourier decomposition on each piece of first heartbeat information to acquire a first time-frequency representation corresponding to each piece of first heartbeat information. It is to be understood that the first time-frequency representation refers to representing the first heartbeat information as a joint function of time and frequency.

Step 103: and extracting corresponding first instantaneous frequency characteristics according to the first time frequency representation corresponding to each piece of first heartbeat information.

In a specific implementation process, after the first time-frequency representation corresponding to each piece of first heartbeat information is obtained, the feature extraction device may extract the corresponding first instantaneous-frequency feature according to the first time-frequency representation. Therefore, the first instantaneous-frequency characteristic corresponding to each piece of first heartbeat information can be obtained. It can be understood that the first instantaneous-frequency feature is a result obtained by the feature extraction device performing feature extraction on the first heartbeat information through an adaptive fourier decomposition algorithm.

On the basis of the foregoing embodiment, the preprocessing the to-be-processed electrocardiographic signal to obtain at least one piece of first heartbeat information corresponding to the to-be-processed electrocardiographic signal includes: performing R wave peak detection on the electrocardiosignals to be processed to obtain at least one R wave peak point; and acquiring first heartbeat information corresponding to each R wave peak point according to the at least one R wave peak point and the sampling frequency of the electrocardiosignals to be processed.

In a specific implementation process, when the feature extraction device preprocesses the electrocardiosignals to be processed, the feature extraction device firstly detects the R wave peak of the electrocardiosignals to be processed to obtain the R wave peak point in the electrocardiosignals to be processed. Because the QRS complex in the electrocardiographic signal is usually the most obvious waveform and has a high-rise sharp R peak, it is relatively easy to acquire the R wave peak in the electrocardiographic signal. After the R wave peak point is obtained, first heartbeat information is obtained according to the sampling frequency of the electrocardiosignals to be processed, because the number of sampling points contained in one complete heartbeat is related to the sampling frequency.

After the position of the R wave peak point is determined, a certain number of sampling points are respectively intercepted forwards and backwards by taking the R wave peak point as a reference point, so that complete first heartbeat information is obtained. For example: the first 99 sampling points before the peak point of the R wave and the last 200 sampling points can be selected, and a complete first heartbeat message has 300 sampling points, including the P wave, the QRS complex, the T wave, and the like.

When the electrocardiosignal to be processed is preprocessed, all R wave peak points included in the electrocardiosignal to be processed are obtained firstly, then first heartbeat information corresponding to each R wave peak point is obtained according to sampling frequency, and due to different sampling frequencies, the number of sampling points included in one complete first heartbeat information is different, so that the complete first heartbeat information can be guaranteed to be obtained by obtaining the first heartbeat information according to the sampling frequency.

On the basis of the foregoing embodiment, fig. 2 is a schematic flow diagram of adaptive fourier decomposition provided in an embodiment of the present application, and as shown in fig. 2, the performing adaptive fourier decomposition on each piece of first heartbeat information to obtain a first time-frequency representation corresponding to each piece of first heartbeat information includes the following steps:

step 201: obtaining a first function corresponding to the first heartbeat information by using Hilbert transform; the first function is used for representing the corresponding relation between the real-value signal and the complex-value signal of the first heartbeat information.

In a specific implementation procedure, let the real-valued signal of the first heartbeat information be s, and the complex-valued signal of the first heartbeat information be s⁺Wherein s is⁺Is an analytical function of Hardy space of the real-valued signal s. Fig. 3 is a schematic diagram of another adaptive fourier decomposition process according to an embodiment of the present application, as shown in fig. 3, in which a solid line bar represents a real-valued signal, and a dashed line bar represents a complex-valued signal corresponding to the real-valued signal. According to the waveform diagram of the electrocardiosignal EGC to be processed, a Hardy space projection diagram can be obtained, and a projection diagram of the real-valued signal in a complex plane can be obtained. Obtaining a first function s using a Hilbert transform⁺S + iHs, and further, by s⁺S + iHs gets s 2Res⁺-c₀Wherein H is Hilbert transform and i is an imaginary unit. c. C₀Is the 0 th Fourier coefficient, Re represents s⁺The real part of (a). The corresponding relation between the real-valued signal and the complex-valued signal of the first heartbeat information can be obtained through the first function.

Step 202: by using

Performing first decomposition on the first function by using the regeneration nuclear property of the nuclear and the maximum selection principle to obtain a first decomposition result; decomposing each decomposition result by using a maximum selection principle in sequence from the first decomposition result to obtain a next decomposition result, and decomposing the m-1 decomposition result to obtain a second function; wherein m is a positive integer greater than 1; whereinThe second function is used for representing a decomposition result of the complex-valued signal corresponding to the first heartbeat information; and obtaining the first time-frequency representation according to the second function.

In a specific implementation, let s₁＝s⁺According to

The first function is transformed by the regenerated kernel properties of the kernel. In which the nuclear nature is regenerated

The core is

z is e_aRepresents a point on the unit disk D; a is any point on the D, and a is the point on the D,

is the complex conjugate of a. The first function obtained by the transformation is

To make s₁Parameters can be selected in a self-adaptive mode to obtain a faster convergence speed, the transformed first function can be decomposed for the first time through a great selection principle to obtain a first decomposition result, then the first decomposition result is decomposed again to obtain a second decomposition result, decomposition is carried out in sequence, namely, the decomposition result is decomposed again each time to obtain a next decomposition result, and finally, the m-1 decomposition results are decomposed to obtain a second function. Wherein the value of m is a positive integer greater than 1.

The specific decomposition process is as follows: a can be found on the unit disk D₁Point such that formula a₁＝argmax_a∈D{(1-|a|²)|s₁(a)|²Is true, so that the first function can be decomposed to obtain the first decomposition result

Wherein the content of the first and second substances,

as residue of the first decomposition, B₁Is a function when n is 1 in a rational orthogonal system, or is a Takenaka-Malmqis system (TM system for short), and the TM system is a TM system

Wherein the content of the first and second substances,

a_nis a point in the unit disc, and n is 1, 2.

The first decomposition result is decomposed for the second time by using the maximum selection principle again to obtain a second decomposition result

a₂As a point in the unit disc, B₂For a function when n is 2 in a rational orthogonal system,

as a residue of the second decomposition. And then, carrying out third decomposition on the second decomposition result, obtaining an nth decomposition result after carrying out n-time decomposition, wherein the nth decomposition result is a second function

，a_nN is 1,2, a point determined on the unit disc using the maximum selection principle, and a_n＝argmax_a∈D{(1-|a|²)|s_n(a)|²}。

Step 203: and obtaining the first time-frequency representation according to the second function.

In a specific implementation process, the residual R in the second function is used_n(z) obtaining a first time-frequency representation corresponding to the first heartbeat information after the removing

As can also be seen in FIG. 3, each will be

As in each iteration

And (4) taking the iteration number as 10 as an example, and obtaining corresponding time-frequency representation after 10 iterations.

Step 204: and extracting corresponding first instantaneous frequency characteristics according to the first time frequency representation corresponding to each piece of first heartbeat information.

In a specific implementation, a calculation expression of the first instantaneous frequency characteristic may be obtained according to the first time-frequency representation, that is, the first instantaneous frequency characteristic is obtained

It is understood that a here_nIs shown in the form of a polar coordinate,

it should be noted that phi_n' (t) is the first extracted temporal frequency feature.

Before the category of each second heartbeat information included in the electrocardiosignal to be recognized needs to be recognized, a classifier needs to be constructed in advance, so that the feature vector corresponding to the second heartbeat information is input into the classifier, the classifier can recognize the feature vector, and the category corresponding to the second heartbeat information is output. Therefore, fig. 4 is a schematic flowchart of a classifier construction method provided in an embodiment of the present application, and as shown in fig. 4, the method includes:

step 401: and acquiring a plurality of pieces of third heartbeat information, labeling each piece of third heartbeat information, and acquiring a corresponding heartbeat label.

In a specific implementation process, when a classifier is obtained through training, a large number of data samples need to be obtained first, and the data samples adopted in the embodiment of the application can be an MIT-BIH electrocardiogram abnormity database which is widely used internationally. The MIT-BIH database is recorded data for analyzing arrhythmia given by laboratories of the american academy of labor and science and the Beth Israel Hospital (Beth Israel Hospital), is a standard library that is globally recognized at present, and is widely applied in the field of electrocardiographic analysis. A total of forty-eight records are available in the MIT-BIH database, with clinical ecg signals from twenty-five males and twenty-two females ranging from thirty to ninety years of age, with a 30 minute record time for each patient. The first twenty-three records of all records were randomly drawn from the four thousand records, and the last twenty-five records were selected as relatively rare but clinically significant abnormal manifestations. More than two related field experts are marked in the information of the heartbeat type recorded in the database to obtain the heartbeat label corresponding to each third heartbeat information, so that a part of heartbeat information in the MIT-BIH database can be used as the third heartbeat information. It is understood that the third heartbeat information and the corresponding heartbeat label form a training set.

After being labeled by experts in the related field, all heartbeats are classified into five major categories according to the classification standard of electrocardiogram (ecg) established by the Association for the Advancement of Medical Instrumentation (AAMI). N, S, V, F, Q, the concrete contents of the five classes are shown in table 1, and according to practical considerations, the class Q is not generally classified due to the rare number and rarity thereof, and in practical operation, signals which cannot be classified into N, S, V, F are classified into the class Q.

TABLE 1

It should be noted that the remaining heart beat information in the MIT-BIH database can also be used as a test set, and therefore, the number of training sets corresponding to each heart beat category included in the test set can be obtained, as shown in Table 2.

TABLE 2

Class AAMI	N	S	V	F
					Total number of heart beats	89682	2770	6836	801
Training set of heart beat counts	45641	941	3753	423
					Testing heart beat count	44041	1829	3083	378

Step 402: according to the electrocardio feature extraction method provided by each embodiment, feature extraction is performed on each third heartbeat information, and a third instantaneous frequency feature corresponding to each third heartbeat information is obtained.

In a specific implementation process, the feature extraction is performed on each piece of third heartbeat information by using the electrocardiogram feature extraction method provided by each embodiment, so as to obtain a third instantaneous frequency feature corresponding to each piece of third heartbeat information. The specific method for extracting the features of the third heartbeat information is consistent with the above embodiments, and is not described here again.

It can be understood that feature extraction can be performed on each heartbeat information in the test set by the above-mentioned electrocardiogram feature extraction method.

Step 403: and acquiring a second preset traditional characteristic corresponding to each piece of third heartbeat information, wherein the second preset traditional characteristic comprises any one or combination of duration time of a QRS waveform, forward RR interval, backward RR interval, average RR interval and amplitude of a peak point of an R wave.

In a specific implementation process, for each piece of third heartbeat information, besides the features extracted by the above-mentioned electrocardiographic feature extraction method, some conventional features can better characterize the heartbeat type, and therefore, a second preset conventional feature corresponding to each piece of third heartbeat information can be obtained, where the second preset conventional feature includes any one or a combination of a QRS waveform duration, a forward RR interval, a backward RR interval, an average RR interval, and an amplitude of an R wave peak point.

It is understood that QRS waveform duration is the QRS waveform end-QRS waveform start; the forward RR interval is the coordinate of the R wave peak point of the current heart beat-the coordinate of the R wave peak point of the previous heart beat; the backward RR interval is the coordinate of the R wave peak point of the next heart beat-the coordinate of the R wave peak point of the current heart beat; the average RR interval is the average value of the RR intervals of the previous ten heartbeats of the current heart beat; the amplitude of the R-wave peak point is equal to the amplitude of the R-wave peak point position of each heartbeat.

Step 404: and taking a feature vector formed by the third instantaneous frequency feature and the second preset traditional feature as input, and taking a heartbeat label corresponding to the third heartbeat information as output to perform model training to obtain the classifier.

In a specific implementation process, after a third instantaneous frequency feature and a second preset traditional feature corresponding to each piece of third heartbeat information are obtained, a feature vector is formed according to the third instantaneous frequency feature and the second preset traditional feature, the feature vector is used as input, a heartbeat label corresponding to the third heartbeat information is used as output to perform model training, and a classifier is obtained. The model training can be performed by adopting a Support Vector Machine (SVM) classifier, and can also be performed by adopting a decision tree classifier or a random forest classifier and the like.

The embodiment of the present application does not limit the execution steps of the model training.

It should be noted that, as can be seen from table 2, the number distribution of each heartbeat category sample is unbalanced, which is unavoidable, and the classification decision of each category of classifiers is greatly affected, in order to solve the problem, the present application may employ a penalty parameter asymmetric support vector machine, that is, a weighted support vector machine, to set a high value for an interested wrong penalty parameter, so as to alleviate the above problem. During training, can adopt the Libsvm toolbox, it can be fine be applicable to many categorised problems, it alleviates the influence that each kind of data unbalance brought through setting for different weighted values for different classes. The optimal parameters to be determined for training the classifier include: kernel function parameter γ, penalty for loss function parameter C, weighting parameter w of different classes_iAnd i represents the ith category.

After training is completed and a classifier is obtained, all parameters in the classifier are adjusted by adopting a ten-fold cross-validation method, and finally obtained internal parameters of the classifier are shown in a table 3:

TABLE 3

Content of parameters	C	γ	w₁	w₂	w₃	w₄
							Value of parameter	3	0.0008	0.42	36	2.5	1.79

Inputting the feature vectors of the test set into a trained classifier for classification, and outputting a classification result, where table 4 is a classification accuracy provided by the embodiment of the present application:

TABLE 4

The method and the device for detecting the abnormal heart beat verify through the test set, the total accuracy can reach more than 85%, and the accuracy of each abnormal category reaches more than 80%, so that the classifier obtained through training in the embodiment of the application can improve the detection accuracy of the abnormal heart beat, and particularly can classify the heart beat in the electrocardiosignal of the single lead.

Fig. 5 is a schematic flow chart of a heartbeat identification method provided in an embodiment of the present application, and as shown in fig. 5, the method includes:

step 501: acquiring an electrocardiosignal to be identified; the electrocardiosignal to be identified can be obtained by detecting the electrocardiosignal of a certain person through an electrocardio detector, and can also be obtained through other modes.

Step 502: according to the electrocardio-feature extraction method provided by each embodiment, second heartbeat information corresponding to the electrocardiosignal to be identified and second instantaneous frequency features corresponding to the second heartbeat information are obtained; it should be noted that the method for extracting the features of the second heartbeat information is consistent with the above embodiments, and is not described herein again.

Step 503: and acquiring each piece of second heartbeat information corresponding to a first preset traditional characteristic, wherein the first preset traditional characteristic comprises any one or combination of duration of a QRS waveform, a forward RR interval, a backward RR interval, an average RR interval and an amplitude of an R wave peak point.

In a specific implementation process, in order to further improve the accuracy of classifying the second heartbeat information, a first preset conventional feature corresponding to each piece of second heartbeat information may be further obtained, and it may be understood that the first preset conventional feature is a common feature that can reflect a category of the second heartbeat information, and specifically may include any one of or a combination of the following:

QRS waveform duration is equal to QRS waveform end-QRS waveform start;

the forward RR interval is the coordinate of the R wave peak point of the current heart beat-the coordinate of the R wave peak point of the previous heart beat;

the backward RR interval is the coordinate of the R wave peak point of the next heart beat-the coordinate of the R wave peak point of the current heart beat;

the average RR interval is the average value of the RR intervals of the previous ten heartbeats of the current heart beat;

the amplitude of the R-wave peak point is equal to the amplitude of the R-wave peak point position of each heartbeat.

Step 504: identifying a feature vector corresponding to each second heartbeat information by using a pre-trained classifier to obtain a heartbeat category corresponding to each second heartbeat information; wherein the feature vector comprises the second instantaneous frequency feature and the first preset legacy feature; and the classifier performs feature extraction on a training set through the self-adaptive Fourier decomposition algorithm to obtain training features, and the training features are used for training.

In a specific implementation process, after the second heartbeat information is subjected to feature extraction to obtain a second instantaneous frequency feature corresponding to each second heartbeat information, a corresponding feature vector is obtained according to the second instantaneous frequency feature and a first preset traditional feature, the feature vector is used as an input of a classifier, and the classifier identifies the second instantaneous frequency feature to obtain a heartbeat category corresponding to the second heartbeat information. The classifier is obtained by training in the above embodiment, and the specific training process is referred to the above embodiment, which is not described herein again.

Fig. 6 is a schematic flow chart of another heart beat identification method provided in the embodiment of the present application, and as shown in fig. 6, the whole flow includes preprocessing, feature extraction, and classification;

firstly, acquiring an electrocardiosignal (ECG signal) to be processed, wherein the ECG signal can be acquired through an MIT-BIH database, the MIT-BIH database comprises a plurality of heart beat information and a heart beat label corresponding to each heart beat, and R wave peak point detection is carried out on the ECG signal to obtain the R wave peak point included in the ECG signal. According to the sampling frequency corresponding to the ECG signal, the R wave peak point is used as a reference, and a preset number of sampling points are respectively obtained forwards and backwards, so that the ECG signal is divided into single heart beats. A part of all the single heartbeats obtained is used as a training set, and the rest is used as a testing set. It will be appreciated that the training set is used to train the acquisition classifier and the classifier is used to classify the test set. And, the single heartbeat included in the training set is taken as the third heartbeat information, and the single heartbeat included in the testing set is taken as the second heartbeat information.

And respectively extracting the features of the third heartbeat information and the second heartbeat information, and during extraction, obtaining a third instantaneous frequency feature corresponding to the third heartbeat information and a second instantaneous frequency feature corresponding to the second heartbeat information by utilizing self-adaptive Fourier decomposition. In addition, a first preset traditional feature corresponding to the second heartbeat information and a second preset traditional feature corresponding to the third heartbeat information can be extracted. Wherein the second preset conventional feature and the first preset conventional feature comprise duration of QRS waveform, forward RR interval, backward RR interval, average RR interval, and amplitude of R-wave peak point. The third instantaneous frequency characteristic and the second preset traditional characteristic corresponding to the third heartbeat information can form a training set characteristic vector, and the second instantaneous frequency characteristic and the first preset traditional characteristic corresponding to the second heartbeat information can form a test set characteristic vector.

And training the Support Vector Machine (SVM) by using the training set feature vector as an input and using the heartbeat label corresponding to the third heartbeat information as an output to obtain a classifier, and then testing the test set by using the classifier obtained by training to obtain the heartbeat category corresponding to each second heartbeat information in the test set.

Fig. 7 is a schematic structural diagram of an electrocardiographic feature extraction device according to an embodiment of the present application, and as shown in fig. 7, the device includes: a preprocessing module 701, a decomposition module 702, and an extraction module 703, wherein:

the preprocessing module 701 is configured to acquire an electrocardiographic signal to be processed, and preprocess the electrocardiographic signal to be processed to acquire at least one piece of first heartbeat information corresponding to the electrocardiographic signal to be processed; the decomposition module 702 is configured to perform adaptive fourier decomposition on each piece of the first heartbeat information, so as to obtain a first time-frequency representation corresponding to each piece of the first heartbeat information; the extracting module 703 is configured to extract a corresponding first temporal frequency feature according to the first time frequency representation corresponding to each piece of the first heartbeat information.

On the basis of the above embodiment, the preprocessing module is specifically configured to: performing R wave peak detection on the electrocardiosignals to be processed to obtain at least one R wave peak point; and acquiring first heartbeat information corresponding to each R wave peak point according to the at least one R wave peak point and the sampling frequency of the electrocardiosignals to be processed.

On the basis of the above embodiment, the decomposition module is specifically configured to: obtaining a first function corresponding to the first heartbeat information by using Hilbert transform; the first function is used for representing the corresponding relation between the real-value signal and the complex-value signal of the first heartbeat information; by using

Decomposing the decomposition result of each time by the regeneration nuclear property of the nuclear and the maximum selection principle to obtain the next decomposition result, and decomposing the m-1 decomposition result to obtain a second function; wherein m is a positive integer greater than 1; the second function is used for representing a decomposition result of the complex-valued signal corresponding to the first heartbeat information; and obtaining the first time-frequency representation according to the second function.

In the above-mentioned implementationOn the basis of the example, the decomposition module is specifically configured to: obtaining a first function s corresponding to the first heartbeat information according to Hilbert transform⁺S + iHs, where s is the real-valued signal of the first heartbeat information, s⁺The first heartbeat information is a corresponding complex value signal, H is Hilbert transform, and i is an imaginary number unit; by using

Regenerative nuclear properties of the nucleus

On the basis of the above embodiment, the extraction module is specifically configured to: obtaining the first instantaneous frequency characteristic according to the first time-frequency representation as follows:

z＝e^itis a point on the unit circle.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working process of the apparatus described above may refer to the corresponding process in the foregoing method, and will not be described in too much detail herein.

In summary, in the embodiment of the application, the electrocardiosignals to be processed are preprocessed to obtain at least one piece of first heartbeat information, and then each piece of first heartbeat information is subjected to adaptive fourier decomposition to obtain a corresponding first instantaneous-frequency feature, wherein the first instantaneous-frequency feature is a feature corresponding to the first heartbeat information to be extracted.

Fig. 8 is a schematic structural diagram of a heartbeat recognition device according to an embodiment of the present application, and as shown in fig. 8, the device includes: an obtaining module 801, a first feature extraction module 802, a second feature extraction module 803, and an identifying module 804, wherein:

the acquisition module 801 is used for acquiring an electrocardiosignal to be identified; the first feature extraction module 802 is configured to obtain second heartbeat information corresponding to the electrocardiographic signal to be identified and second instantaneous frequency features corresponding to the second heartbeat information according to the electrocardiographic feature extraction method; the second feature extraction module 803 is configured to obtain a second preset conventional feature corresponding to each piece of third heartbeat information, where the second preset conventional feature includes any one or a combination of a QRS waveform duration, a forward RR interval, a backward RR interval, an average RR interval, and an amplitude of an R-wave peak point; the identification module 804 is configured to identify a feature vector corresponding to each second heartbeat information by using a pre-trained classifier, and obtain a heartbeat category corresponding to each second heartbeat information; wherein the feature vector comprises the second instantaneous frequency feature and the first preset legacy feature; and the classifier performs feature extraction on a training set through the self-adaptive Fourier decomposition algorithm to obtain training features, and the training features are used for training.

On the basis of the above embodiment, the apparatus further includes a classifier building module, configured to obtain a plurality of pieces of third heartbeat information, and label each piece of the third heartbeat information to obtain a corresponding heartbeat tag; performing feature extraction on each third heartbeat information according to the electrocardiogram feature extraction method of the first aspect to obtain a third instantaneous frequency feature corresponding to each third heartbeat information; acquiring a second preset traditional feature corresponding to each piece of third heartbeat information, wherein the second preset traditional feature comprises any one or combination of duration time of a QRS waveform, forward RR interval, backward RR interval, average RR interval and amplitude of an R wave peak point; and taking a feature vector formed by the third instantaneous frequency feature and the second preset traditional feature as input, and taking a heartbeat label corresponding to the third heartbeat information as output to perform model training to obtain the classifier.

In summary, in the embodiment of the application, the second heartbeat information in the electrocardiosignal to be identified is subjected to feature extraction by the above electrocardio feature extraction method, the category of each second heartbeat information is identified by using the preset classifier, the heartbeat category corresponding to the second heartbeat information is obtained, and the accuracy of identification can be improved by identifying the category through representative features.

Fig. 9 is a schematic structural diagram of an entity of an electronic device provided in an embodiment of the present application, and as shown in fig. 9, the electronic device includes: a processor (processor)901, a memory (memory)902, and a bus 903; wherein the content of the first and second substances,

the processor 901 and the memory 902 complete communication with each other through the bus 903;

the processor 901 is configured to call program instructions in the memory 902 to perform the methods provided by the above-mentioned method embodiments, for example, including: acquiring an electrocardiosignal to be processed, and preprocessing the electrocardiosignal to be processed to obtain at least one piece of first heartbeat information corresponding to the electrocardiosignal to be processed; respectively carrying out self-adaptive Fourier decomposition on each first heartbeat information to obtain a first time-frequency representation corresponding to each first heartbeat information; and extracting corresponding first instantaneous frequency characteristics according to the first time frequency representation corresponding to each piece of first heartbeat information.

The present embodiment discloses a computer program product comprising a computer program stored on a non-transitory computer readable storage medium, the computer program comprising program instructions which, when executed by a computer, enable the computer to perform the method provided by the above-mentioned method embodiments, for example, comprising: acquiring an electrocardiosignal to be processed, and preprocessing the electrocardiosignal to be processed to obtain at least one piece of first heartbeat information corresponding to the electrocardiosignal to be processed; respectively carrying out self-adaptive Fourier decomposition on each first heartbeat information to obtain a first time-frequency representation corresponding to each first heartbeat information; and extracting corresponding first instantaneous frequency characteristics according to the first time frequency representation corresponding to each piece of first heartbeat information.

The present embodiments provide a non-transitory computer-readable storage medium storing computer instructions that cause the computer to perform the methods provided by the above method embodiments, for example, including: acquiring an electrocardiosignal to be processed, and preprocessing the electrocardiosignal to be processed to obtain at least one piece of first heartbeat information corresponding to the electrocardiosignal to be processed; respectively carrying out self-adaptive Fourier decomposition on each first heartbeat information to obtain a first time-frequency representation corresponding to each first heartbeat information; and extracting corresponding first instantaneous frequency characteristics according to the first time frequency representation corresponding to each piece of first heartbeat information.

Fig. 10 is a schematic structural diagram of an entity of an electronic device provided in an embodiment of the present application, and as shown in fig. 10, the electronic device includes: a processor (processor)1001, a memory (memory)1002, and a bus 1003; wherein the content of the first and second substances,

the processor 1001 and the memory 1002 complete communication with each other through the bus 1003;

the processor 1001 is configured to call the program instructions in the memory 1002 to execute the methods provided by the above-mentioned method embodiments, for example, including: acquiring an electrocardiosignal to be identified; according to the electrocardio-feature extraction method in each embodiment, second heartbeat information corresponding to the electrocardiosignal to be identified and second instantaneous frequency features corresponding to the second heartbeat information are obtained; acquiring first preset traditional characteristics corresponding to each piece of second heartbeat information, wherein the first preset traditional characteristics comprise any one or combination of QRS waveform duration, forward RR interval, backward RR interval, average RR interval and amplitude of R wave peak point; identifying a feature vector corresponding to each second heartbeat information by using a pre-trained classifier to obtain a heartbeat category corresponding to each second heartbeat information; wherein the feature vector comprises the second instantaneous frequency feature and the first preset legacy feature; and the classifier performs feature extraction on a training set through the self-adaptive Fourier decomposition algorithm to obtain training features, and the training features are used for training.

The present embodiment discloses a computer program product comprising a computer program stored on a non-transitory computer readable storage medium, the computer program comprising program instructions which, when executed by a computer, enable the computer to perform the method provided by the above-mentioned method embodiments, for example, comprising: acquiring an electrocardiosignal to be identified; according to the electrocardio-feature extraction method of each embodiment, second heartbeat information corresponding to the electrocardiosignal to be identified and second instantaneous frequency features corresponding to the second heartbeat information are obtained; acquiring first preset traditional characteristics corresponding to each piece of second heartbeat information, wherein the first preset traditional characteristics comprise any one or combination of QRS waveform duration, forward RR interval, backward RR interval, average RR interval and amplitude of R wave peak point; identifying a feature vector corresponding to each second heartbeat information by using a pre-trained classifier to obtain a heartbeat category corresponding to each second heartbeat information; wherein the feature vector comprises the second instantaneous frequency feature and the first preset legacy feature; and the classifier performs feature extraction on a training set through the self-adaptive Fourier decomposition algorithm to obtain training features, and the training features are used for training.

The present embodiments provide a non-transitory computer-readable storage medium storing computer instructions that cause the computer to perform the methods provided by the above method embodiments, for example, including: acquiring an electrocardiosignal to be identified; according to the electrocardio-feature extraction method in each embodiment, second heartbeat information corresponding to the electrocardiosignal to be identified and second instantaneous frequency features corresponding to the second heartbeat information are obtained; acquiring first preset traditional characteristics corresponding to each piece of second heartbeat information, wherein the first preset traditional characteristics comprise any one or combination of QRS waveform duration, forward RR interval, backward RR interval, average RR interval and amplitude of R wave peak point; identifying a feature vector corresponding to each second heartbeat information by using a pre-trained classifier to obtain a heartbeat category corresponding to each second heartbeat information; wherein the feature vector comprises the second instantaneous frequency feature and the first preset legacy feature; and the classifier performs feature extraction on a training set through the self-adaptive Fourier decomposition algorithm to obtain training features, and the training features are used for training.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other ways. The apparatus embodiments described above are merely illustrative, and for example, the flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

In addition, functional modules in the embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

The above description is only a preferred embodiment of the present application and is not intended to limit the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application. It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application.

It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.

Claims

1. An electrocardiogram feature extraction method is characterized by comprising the following steps:

acquiring an electrocardiosignal to be processed, and preprocessing the electrocardiosignal to be processed to obtain at least one piece of first heartbeat information corresponding to the electrocardiosignal to be processed;

respectively carrying out self-adaptive Fourier decomposition on each first heartbeat information to obtain a first time-frequency representation corresponding to each first heartbeat information;

extracting corresponding first instantaneous frequency characteristics according to the first time frequency representation corresponding to each piece of first heartbeat information;

the performing adaptive fourier decomposition on each piece of the first heartbeat information to obtain a first time-frequency representation corresponding to each piece of the first heartbeat information respectively includes:

obtaining a first function corresponding to the first heartbeat information by using Hilbert transform; the first function is used for representing the corresponding relation between the real-value signal and the complex-value signal of the first heartbeat information;

by using

Decomposing the decomposition result of each time by the regeneration nuclear property of the nuclear and the maximum selection principle to obtain the next decomposition result, and decomposing the m-1 decomposition result to obtain a second function; wherein m is a positive integer greater than 1; the second function is used for representing a decomposition result of the complex-valued signal corresponding to the first heartbeat information;

and obtaining the first time-frequency representation according to the second function.

2. The method according to claim 1, wherein the performing an adaptive fourier decomposition on each of the first heartbeat information to obtain a first time-frequency representation corresponding to each of the first heartbeat information respectively comprises:

obtaining a first function s corresponding to the first heartbeat information according to Hilbert transform⁺S + iHs, where s is the real-valued signal of the first heartbeat information, s⁺The first heartbeat information is a corresponding complex value signal, H is Hilbert transform, and i is an imaginary number unit;

by using

Regenerative nuclear properties of the nucleus

Wherein z is e_aRepresents a point on the unit disk D; a is_nFor points determined on the unit disc using the maximum selection principle, and a_n＝argmax_a∈D{(1-|a|²)|s_n(a)|²D is the unit disc, a is any point on D, B_nIs a rational orthogonal system, and

R_nis the residue after n decompositions, n is 1, 2.;

obtaining a first time-frequency representation of the first heartbeat information according to the second function as:

3. The method according to claim 2, wherein extracting the first temporal frequency feature of the corresponding first heartbeat information according to the time-frequency representation corresponding to each first heartbeat information comprises:

the first instantaneous frequency characteristic obtained according to the first time-frequency representation of the first heartbeat information is as follows:

wherein phi is_n(t) is B_nThe phase function of (a);

z＝e^itis a point on the unit circle.

4. The method according to any one of claims 1 to 3, wherein said preprocessing said to-be-processed cardiac signal to obtain at least one first heartbeat information corresponding to said to-be-processed cardiac signal comprises:

performing R wave peak detection on the electrocardiosignals to be processed to obtain at least one R wave peak point;

and acquiring first heartbeat information corresponding to each R wave peak point according to the at least one R wave peak point and the sampling frequency of the electrocardiosignals to be processed.

5. A heart beat identification method is characterized by comprising the following steps:

acquiring an electrocardiosignal to be identified;

the method for extracting electrocardiographic features according to any one of claims 1 to 4, wherein second heartbeat information corresponding to the electrocardiographic signal to be identified and second instantaneous frequency features corresponding to each piece of the second heartbeat information are obtained;

acquiring first preset traditional characteristics corresponding to each piece of second heartbeat information, wherein the first preset traditional characteristics comprise any one or combination of QRS waveform duration, forward RR interval, backward RR interval, average RR interval and amplitude of R wave peak point;

identifying a feature vector corresponding to each second heartbeat information by using a pre-trained classifier to obtain a heartbeat category corresponding to each second heartbeat information; wherein the feature vector comprises the second instantaneous frequency feature and the first preset legacy feature; and the classifier performs feature extraction on a training set through the self-adaptive Fourier decomposition algorithm to obtain training features, and the training features are used for training.

6. The method of claim 5, wherein the classifier is obtained by:

acquiring a plurality of pieces of third heartbeat information, and labeling each piece of third heartbeat information to obtain a corresponding heartbeat label;

the method for extracting electrocardiographic features according to any one of claims 1 to 4, wherein each of the third heartbeat information is subjected to feature extraction to obtain a third instantaneous frequency feature corresponding to each of the third heartbeat information;

acquiring a second preset traditional feature corresponding to each piece of third heartbeat information, wherein the second preset traditional feature comprises any one or combination of duration time of a QRS waveform, forward RR interval, backward RR interval, average RR interval and amplitude of an R wave peak point;

and taking a feature vector formed by the third instantaneous frequency feature and the second preset traditional feature as input, and taking a heartbeat label corresponding to the third heartbeat information as output to perform model training to obtain the classifier.

7. An electrocardiographic feature extraction device characterized by comprising:

the preprocessing module is used for acquiring an electrocardiosignal to be processed, preprocessing the electrocardiosignal to be processed and acquiring at least one piece of first heartbeat information corresponding to the electrocardiosignal to be processed;

the decomposition module is used for respectively carrying out self-adaptive Fourier decomposition on each piece of first heartbeat information to obtain a first time-frequency representation corresponding to each piece of first heartbeat information;

the extraction module is used for extracting corresponding first instantaneous frequency characteristics according to the first time-frequency representation corresponding to each piece of first heartbeat information;

the decomposition module is specifically configured to:

by using

The regenerative nuclear nature of the nucleus and the maximum selection principle divide the first function for the first timeSolving to obtain a first decomposition result; using the first decomposition result

8. A heart beat recognition device, comprising:

the acquisition module is used for acquiring the electrocardiosignals to be identified;

a first feature extraction module, configured to obtain second heartbeat information corresponding to the electrocardiographic signal to be identified and second instantaneous frequency features corresponding to the second heartbeat information according to the electrocardiographic feature extraction method according to any one of claims 1 to 4;

a second feature extraction module, configured to obtain a first preset conventional feature corresponding to each piece of second heartbeat information, where the first preset conventional feature includes any one or a combination of a QRS waveform duration, a forward RR interval, a backward RR interval, an average RR interval, and an amplitude of an R-wave peak point;

the identification module is used for identifying the feature vector corresponding to each second heartbeat information by using a pre-trained classifier to obtain a heartbeat category corresponding to each second heartbeat information; wherein the feature vector comprises the second instantaneous frequency feature and the first preset legacy feature; and the classifier performs feature extraction on a training set through the self-adaptive Fourier decomposition algorithm to obtain training features, and the training features are used for training.

9. An electronic device, comprising: a processor, a memory, and a bus, wherein,

the processor and the memory are communicated with each other through the bus;

the memory stores program instructions executable by the processor, the processor invoking the program instructions to perform the method of any of claims 1-4 or the method of any of claims 5-6.