WO2022110524A1

WO2022110524A1 - Method and device for electrocardiogram heartbeat data clustering, electronic device, and medium

Info

Publication number: WO2022110524A1
Application number: PCT/CN2021/072841
Authority: WO
Inventors: 刘盛捷
Original assignee: 深圳邦健生物医疗设备股份有限公司
Priority date: 2020-11-30
Filing date: 2021-01-20
Publication date: 2022-06-02
Also published as: CN112528783B; CN112528783A

Abstract

Disclosed are a method and device for electrocardiogram heartbeat data clustering, an electronic device, and a medium. The method comprises: comparing heartbeat data to be processed with multiple data templates; if a matching target data template is present, using information of said heartbeat data to update information of the target data template; if none is present, creating, on the basis of said heartbeat data, a new data template corresponding to a new template identifier; determining the new data template as the target data template matching said heartbeat data; outputting the template matching result of said heartbeat data, the template matching result comprising the template identifier of the target data template; insofar as the volume of said heartbeat data reaches a preset volume threshold, executing an organizing operation with respect to the data templates, comprising a template merging operation, a template sorting operation, and/or a template deleting operation, to acquire a template organizing result; and acquiring a clustering result of said heartbeat data on the basis of the template matching result and of the template organizing result.

Description

Electrocardiogram heartbeat data clustering method, apparatus, electronic device and medium

technical field

The present invention relates to the technical field of electrocardiogram data processing, in particular to a method, device, electronic device and medium for clustering electrocardiogram heartbeat data.

Background technique

In order to analyze the acquired ECG data more quickly and clearly, more and more computer-aided diagnostic ECG software has been developed. The heartbeat clustering (or template matching) technology can divide the ECG signal into several templates according to the different morphological characteristics of the QRS wave, and each template stores the heartbeat index with the similar QRS wave shape. The work is concentrated on several templates, which greatly saves the doctor's inspection time and improves the diagnosis efficiency.

At present, the ECG heartbeat data clustering can use the feature parameter method. Specifically, after using signal transformation methods such as fast Fourier transform, wavelet transform, and Hilbert transform to convert the ECG signal from the time domain into a transform domain with features that are easier to distinguish, the classical clustering method is used to analyze the ECG signal. Automatic template matching. However, using the signal transformation method will significantly increase the amount of calculation, and many classical clustering methods need to set the number of templates to be matched in advance. When this method is used to perform template matching on cases with low signal-to-noise ratio and multiple sources of premature ventricular arrhythmia, different heartbeats will also be mixed in the same template, the ECG heartbeat data clustering is not accurate enough, and the processing efficiency is not high.

SUMMARY OF THE INVENTION

The present application provides an electrocardiogram heartbeat data clustering method, apparatus, electronic device and medium.

In a first aspect, a method for clustering ECG heartbeat data is provided, including:

Obtaining heartbeat data to be processed, and comparing the heartbeat data to be processed with multiple data templates;

If there is a target data template matching the heartbeat data to be processed, update the information of the target data template using the information of the heartbeat data to be processed;

If there is no target data template matching the heartbeat data to be processed, a new data template is created according to the heartbeat data to be processed, and the new data template corresponds to the new template identifier; determining the target data template that matches the heartbeat data to be processed;

outputting a template matching result of the heartbeat data to be processed, where the template matching result includes a template identifier of the target data template;

Whenever the number of the processed heartbeat data to be processed reaches a preset number threshold, a sorting operation is performed on the data template, and the sorting operation includes a template merging operation, a template sorting operation and/or a template deleting operation, Get the template finishing result;

According to the template matching result and the template sorting result, the clustering result of the heartbeat data to be processed is obtained.

In a second aspect, an ECG heartbeat data clustering device is provided, including a template matching module, a template sorting module and a clustering module, wherein:

The template matching module is used for:

The template sorting module is configured to perform sorting operations on the data templates whenever the number of processed heartbeat data to be processed reaches a preset number threshold, and the sorting operations include template merging operations, template sorting operations operation and/or template deletion to obtain template sorting results;

The clustering module is configured to obtain the clustering result of the heartbeat data to be processed according to the template matching result and the template sorting result.

In a third aspect, an electronic device is provided, comprising a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor causes the processor to perform the first aspect and any of the above. Steps for a possible implementation.

In a fourth aspect, a computer storage medium is provided, the computer storage medium stores one or more instructions, the one or more instructions are adapted to be loaded and executed by a processor as described in the first aspect and any one thereof Steps of possible implementations.

In the present application, multiple heartbeats with similar shapes can be aggregated into one template through template matching, and the templates can be integrated in time, so that real-time clustering can be realized without complex signal transformation, and the data processing efficiency is higher.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application or the background technology, the accompanying drawings required in the embodiments or the background technology of the present application will be described below.

1 is a schematic flowchart of a method for clustering electrocardiogram heartbeat data according to an embodiment of the present application;

FIG. 2 is a schematic flowchart of a signal quality evaluation process provided by an embodiment of the present application;

3 is a schematic flowchart of a template matching process provided by an embodiment of the present application;

4 is a schematic diagram of a template matching result after heartbeat offset search provided by an embodiment of the present application;

5 is a schematic diagram of a template matching result after template merging when there is a template offset provided by an embodiment of the present application;

6 is a schematic diagram of a template matching result after template deletion when the number of templates is too large, according to an embodiment of the present application;

FIG. 7 is a schematic structural diagram of an apparatus for clustering electrocardiogram and heartbeat data according to an embodiment of the present application;

FIG. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed ways

In order to make those skilled in the art better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only It is a part of the embodiments of the present application, but not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

The terms "first", "second" and the like in the description and claims of the present application and the above drawings are used to distinguish different objects, rather than to describe a specific order. Furthermore, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally also includes unlisted steps or units, or optionally also includes For other steps or units inherent to these processes, methods, products or devices.

Reference herein to an "embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the present application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor a separate or alternative embodiment that is mutually exclusive of other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

The ECG signals involved in the embodiments of the present application can be obtained by converting the electrical signals received by the ECG collection box through the electrode pads connected to various parts of the human body through signal conversion. A typical ECG signal consists of P wave, QRS complex and T wave, which reflects the state of human heart activity. By observing the shape (heartbeat waveform) and rhythm (heart rate) of the ECG signal, doctors can learn many aspects of a patient's heart disease.

The Holter signal involved in the embodiments of the present application is a long-term ECG signal, generally 24 hours or more. Because the onset time and triggering conditions of many heart diseases are relatively secret, the general static ECG cannot find its existence within a dozen seconds, which requires the acquisition of long-term Holter signals to continuously record the patient's cardiac electrical signals. However, the recording time span of the Holter signal is large, and the Holter signal with a duration of 24 hours has about 100,000 heartbeats, which makes the diagnosis method of the doctor's visual inspection of the patient's condition inefficient. Signal processing techniques are developed accordingly.

The heartbeat clustering (or template matching) technology can divide the ECG signal into several templates according to the different morphological characteristics of the QRS wave, and each template stores the heartbeat index with the similar QRS wave shape. The work is concentrated on several templates, which greatly saves the doctor's inspection time and improves the diagnosis efficiency.

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

Please refer to FIG. 1. FIG. 1 is a schematic flowchart of a method for clustering electrocardiogram heartbeat data according to an embodiment of the present application. The method may include:

101. Obtain heartbeat data to be processed, and compare the heartbeat data to be processed with multiple data templates.

The execution subject of the embodiment of the present application may be an electrocardiogram and heartbeat data clustering device, which may be an electronic device. Other portable devices such as mobile phones, laptops, or tablet computers with sensitive surfaces (eg, touch screen displays and/or touch pads). It should also be understood that, in some embodiments, the above-described devices are not portable communication devices, but rather desktop computers with touch-sensitive surfaces (eg, touch screen displays and/or touch pads). In an optional implementation manner, the above-mentioned method for clustering electrocardiogram and heartbeat data may be supported and run by a server in response to an operation instruction of a user.

The above-mentioned heartbeat data to be processed may be heartbeat signal data clipped from the collected electrocardiogram data.

In an optional embodiment, before the above step 101, the method further includes:

Obtain electrocardiogram data, divide the above electrocardiogram data, and obtain heart beat segments;

The above-mentioned acquisition of the heartbeat data to be processed includes:

The to-be-processed heartbeat data with a preset signal length is acquired from the heartbeat segment.

The electrocardiogram data in this embodiment of the present application may be an electrocardiogram signal collected by an electrocardiogram acquisition device within a period of time. Generally, the position of the heartbeat can be determined according to the characteristics of the electrocardiogram, and then the signal can be divided into multiple equal-length segments reflecting the characteristics of the heartbeat. .

Optionally, the above-mentioned electrocardiogram data is divided to obtain a plurality of heart beat segments, including:

The ECG data is divided according to the R wave position information obtained by detection, and a plurality of heartbeat segments of equal length centered on the R wave position are obtained.

A complete ECG signal is usually composed of P wave, QRS complex and T wave, in which the QRS complex is the main part of the ECG, and the R wave is dominant in the QRS complex, so the marked position of the heartbeat can be marked with R. Wave prevail. Equal length ECG segments centered on the R-wave position can be acquired for template matching.

Optionally, before obtaining the electrocardiogram data, the method further includes:

Acquire the collected multi-lead ECG data; construct dual-lead ECG data according to the main analysis lead number and the sub-analysis lead number in the above multi-lead ECG data;

Band-pass filtering is performed on the above-mentioned two-lead electrocardiogram data to obtain the above-mentioned electrocardiogram data.

In this embodiment of the present application, multi-lead electrocardiogram data collected by Holter may be used for processing. Holter is an ECG signal recording system, which can provide a powerful information aid for doctors to comprehensively analyze the patient's heart condition by adopting portable ECG signal acquisition and continuous measurement of the patient's cardiac activity for up to 24 hours.

Specifically, the Holter data can be read first, and the data includes the 24-hour multi-lead Holter data segment Signals∈R ^N*M collected by the Holter device, where N represents the number of signal leads, and M represents the signal length; it can also include related parameters, such as:

The R wave position (RPos, in sampling point or time) obtained by the detection of the QRS wave;

RR interval (representing the distance between the current heartbeat and the adjacent previous heartbeat, in sampling points or time units);

QRS wave width (width, in sampling points or time);

Holter device sampling rate fs;

Primary analysis lead number and secondary analysis lead number.

Wherein, the above-mentioned main analysis lead number and sub-analysis lead number are the two dominant lead data numbers determined by analysis in the above-mentioned multi-lead Holter data. Further, signal preprocessing can be performed: construct dual-lead Holter data according to the above-mentioned main analysis lead number and the above-mentioned secondary analysis lead number, and then filter out low-frequency baseline drift interference and high-frequency noise through the above-mentioned band-pass filtering.

For the aforementioned ECG division, the specific can be as follows: according to the R wave position RPos, the Holter data is divided into ECG data fragments X∈R ^N*2*W with length W corresponding to the R wave position in the segment center, where N represents the divided The number of ECG data, 2 means double lead, W means single lead data length.

The above-mentioned heartbeat data to be processed may include a signal of a preset signal length cut out from a two-lead Holter data segment, which may be used as a recent signal RecentSignals∈R ^2*L , where L represents a preset signal length (recent signal length).

In one embodiment, the above step 101 may include:

respectively acquiring multiple relative errors between the waveform of the heartbeat data to be processed and the representative waveforms of the multiple data templates;

The minimum error among the plurality of relative errors is obtained, and if the minimum error is smaller than a preset error threshold, it is determined that the target data template corresponding to the minimum error matches the heartbeat data to be processed.

In this embodiment of the present application, each data template may correspond to a representative waveform, which may be a waveform mean value of template members or determined by other means, which is not limited in this embodiment of the present application. Specifically, the heartbeat data to be processed can be classified by calculating the relative error to measure the similarity between the waveform of the heartbeat data to be processed and the representative waveform of each data template. Optionally, at the same time, the heartbeat offset search range can also be selected and set according to requirements, so as to reduce the data processing amount of template search and matching, and improve the efficiency.

Specifically, the above-mentioned preset error threshold RE _Thre1 may be stored in the device, the templates in the template library represent the waveform TW∈R ^T*2*W , where T is the number of templates, and the heartbeat offset search range ShiftLtm1=[-Smax1 , Smax1], Smax1 represents the maximum excursion search range of the heartbeat. Match the heartbeat waveform with the template in the template library, and calculate the relative error RE(s,t) between the heartbeat and different templates (represented by t) under different offsets s∈ ShiftLtm1 . The error calculation formula can be as follows:

Among them, abs(·) means taking the absolute value, and ppv(·) means taking the peak value , that is, the maximum value-minimum value.

The obtained relative errors are searched for the minimum error RE _min , and the optimal offset Shift _opt corresponding to the minimum error RE _min and the matching template identifier MatchId (ie, the template identifier of the template matching the data waveform to be processed).

An offset search mechanism is introduced in the embodiment of the present application, which can avoid the problem of excessive templates caused by heartbeat offset and template offset, and reduce template information storage overhead.

In one embodiment, before the above-mentioned comparison of the above-mentioned heartbeat data to be processed with a plurality of data templates in the template library, the above-mentioned method further includes:

Calculate the signal-to-noise ratio of the above-mentioned heartbeat data to be processed;

According to the signal-to-noise ratio of the above-mentioned heartbeat data to be processed, the signal quality parameter of the above-mentioned heartbeat data to be processed is determined;

The above-mentioned establishment of a new data template according to the above-mentioned heartbeat data to be processed includes:

When the signal quality parameter of the above-mentioned heartbeat data to be processed is the first quality threshold, a new data template is established according to the above-mentioned heartbeat data to be processed.

Combined with the R wave position and the two-lead Holter data segment, the signal quality of the detected heartbeat data to be processed can be evaluated. Please refer to a schematic diagram of a signal quality evaluation process shown in FIG. 2 . As shown in FIG. 2 , the signal quality evaluation process includes three steps of recording recent signals, recording recent heartbeat information, and evaluating signal quality.

In this embodiment of the present application, a signal-to-noise ratio (SIGNAL-NOISE RATIO, SNR) may be used to measure the signal quality. Specifically, as mentioned above, the recent signal RecentSignals ∈ R2*L can be cropped from the two-lead Holter data segment, where L represents the length of the recent signal; and the recent heart beat information RecentBeatInfo can also be updated, including adding the newly detected R wave position , update the position of the historical heartbeat in the recent signal, and remove the historical heartbeat that exceeds the length L of the recent signal, etc.

Using the recent signal and recent heartbeat information to evaluate the signal quality SigQuality, it can be measured by the signal-to-noise ratio expressed by the following formula:

Among them, Vs represents the standard deviation of the QRS waveform amplitude of the recent heartbeat, and Vn represents the standard deviation of the signal amplitude in the recent signal excluding the area of the QRS waveform of the recent heartbeat. The above-mentioned signal quality parameters can be obtained by setting different rules according to requirements and calculating the signal-to-noise ratio, which is not limited in this embodiment of the present application. In one embodiment, the signal-to-noise ratio threshold SNR _thre can be preset. If SNR>SNR _thre , the output signal quality parameter SigQuality=1 indicates good signal quality; otherwise, output SigQuality=0 indicates poor signal quality.

The above-mentioned first quality threshold may be preset, and the signal quality is detected, and template matching is performed only when the signal quality meets the requirements, otherwise, template matching may not be performed. For example, 1, if SigQuality=1, the signal quality is good, a new template is created based on the newly detected heartbeat information, and an identifier of the new template is output.

Optionally, if RE _min < RE _Thre1 , the template matching is successful, update the matching template information with the newly detected heartbeat information, and output the matching template identifier; if RE _min ≥ RE _Thre1 . If the template matching fails, the signal quality can be further judged.

102. If there is a target data template matching the above-mentioned heartbeat data to be processed, update the information of the above-mentioned target data template using the above-mentioned information of the above-mentioned heartbeat data to be processed.

Specifically, if a target data template matching the above-mentioned heartbeat data to be processed is matched, the information of the target data template can be updated with the information of the heartbeat data to be processed, such as the newly detected R wave position, etc. The heartbeat segment of the heartbeat data to be processed is classified into a matching template, becomes a template member of the template, and can output the template identifier of the target data template.

103. If there is no target data template matching the above-mentioned heartbeat data to be processed, a new data template is established according to the above-mentioned heartbeat data to be processed, and the above-mentioned newly-added data template corresponds to the newly-added template identifier; the above-mentioned newly-added data template is determined as: The above-mentioned target data template matching the above-mentioned heartbeat data to be processed.

Specifically, in this embodiment of the present application, it can be determined whether the newly detected heartbeat data to be processed is the first heartbeat type, that is, whether there is a target data template that matches the above-mentioned heartbeat data to be processed, and if there is no match with the pending heartbeat data The target data template that matches the heartbeat data can be used to create a new data template based on the heartbeat data to be processed. The new data template can use the QRS waveform of the heartbeat data to be processed as the representative waveform of the new data template, and save The RR interval and width values of the heartbeat data to be processed and output the newly added template identifier.

104. Output the template matching result of the heartbeat data to be processed, where the template matching result includes the template identifier of the target data template.

In order to describe the template matching process more clearly, reference can be made to a schematic diagram of a template matching process shown in FIG. 3 . As shown in FIG. 3 , the heartbeat data to be processed is used as the newly detected heartbeat, and the specific process may include:

Determine whether the newly detected heartbeat is the first heartbeat, and if so, create a new template with the newly detected heartbeat information, output the number of the new template, and end the matching process; otherwise, continue to the next step;

Heartbeat match. If the matching is successful, update the matching template information with the newly detected heartbeat information, and output the matching template number; if the matching fails. Further judge the signal quality;

If the signal quality meets the requirements, a new template is created with the newly detected heartbeat information, and the number of the new template is output.

If the signal quality does not meet the requirements, the template matching fails, and the invalid template number represented by a negative number can be output.

The specific method for the above-mentioned heartbeat matching and signal quality judgment can refer to the relevant description in the embodiment shown in FIG. 1 , and details are not repeated here.

105. Whenever the quantity of the above-mentioned heartbeat data to be processed reaches a preset quantity threshold, perform a sorting operation on the above-mentioned data template, and the above-mentioned sorting operation includes a template merging operation, a template sorting operation, and/or a template deletion operation, and obtains: Template collation results.

In this embodiment of the present application, the data template may be sorted out while the clustering is performed. For example, the above preset number threshold is 100, that is, whenever 100 pieces of data to be processed are newly detected, the template information in the existing template library needs to be sorted out.

In an optional implementation manner, in the case that the above-mentioned sorting operation includes a template merging operation, the above-mentioned performing the sorting operation on the above-mentioned data template includes:

Calculate the relative error between the processing template and the reference template, compare the relative error with the preset matching threshold, the reference template is a template in the unorganized template, and the processing template is the template other than the reference template in the unorganized template ;

Obtaining a processing template with the relative error less than the preset matching threshold, combining the processing template with the relative error less than the preset matching threshold with the corresponding reference template, and recording the obtained template as an organized template;

Repeat the above steps until the number of unsorted templates above is zero.

Specifically, the function of template merging is mainly to merge similar templates in the template library to reduce the number of templates. The specific operations can be as follows:

(1) The preset matching threshold RE _Thre2 , the representative waveform TW∈R ^T*2*W of the template in the template library, where T is the number of templates, the template offset search range ShiftLtm2=[-Smax2, Smax2], Smax2 represents the template Maximum offset search range, unsorted template ID list UnMatchIdx={0,1,2,...,T-1}.

(2) The reference template identifier ReferId is set as the first parameter in the unsorted template identifier list, and the processing template identifier TargetIdx to be matched is the remaining parameters in the unsorted template identifier list.

(3) The relative error RE(s,t) of the reference template and each processing template can be calculated by using the same error calculation formula above, s∈ShiftLtm2, t∈TargetIdx; search for the processing template satisfying RE(s,t)<RE _Thre2 Identifies the list TargetIdx1. These processing templates are similar to the reference template, and the information of the reference template will be updated with the information of these processing templates and the corresponding offsets.

(4) Delete ReferId and TargetIdx1 from UnMatchIdx.

(5) Steps (2)-(4) are repeated until the TargetIdx is empty, and the template merging ends.

In an optional implementation manner, when the above-mentioned sorting operation includes a template sorting operation, the above-mentioned performing the sorting operation on the above-mentioned data template includes:

The multiple templates are sorted according to the descending order of the number of template members of each template in the multiple templates.

In the case that the above-mentioned sorting operation includes a template deletion operation, the above-mentioned sorting operation for the above-mentioned data template includes:

When the number of templates M exceeds the first number threshold T1, calculate the age value of each of the above-mentioned templates, and the above-mentioned age value is the latest added template in the above-mentioned template set and the time interval data obtained from the above-mentioned heartbeat data to be processed;

According to the age value and the first duration parameter of each of the above-mentioned templates, calculate the age parameter corresponding to each of the above-mentioned templates;

Obtain a template whose age value is greater than the age parameter corresponding to the age value; determine the templates with the largest age values in the top N from the templates whose age value is greater than the age parameter as the template to be deleted, and delete the template to be deleted from the template set ; N=M-T1.

In this embodiment of the present application, the number of templates can be managed and controlled in real time. The above-mentioned first number threshold T1 can be set, and when the template number M exceeds the first number threshold T1, the first round of template deletion process can be started, and the template with a larger age value can be deleted by comparing and evaluating the age value of the template and the age parameter. .

In a specific embodiment, the first round of template deletion process may include:

Determine the template identifier to be searched CandidateIdx1∈[0,T-1], T is the number of templates;

Calculate the age value Life(t) of different templates, t∈CandidateIdx1, where the age value is specifically the latest added template in the template set and the time interval data obtained from the above-mentioned heartbeat data to be processed, that is, the latest template member heartbeat. Time interval from matching heartbeat to newly detected heartbeat (represented by sample points);

The corresponding age parameters can be calculated according to different age values, which can be calculated by the following formula:

The first duration parameter includes a ₁ and b ₁ , which are fixed duration and variable duration respectively. Optionally, the calculation method of the above-mentioned age parameter can be set as required, such as modifying the above-mentioned first duration parameter, which is not limited here;

Search all template indexes OldIdx satisfying Life(t)>Life _Thre (t);

Calculate the number N of templates to be deleted (DeleteNum), DeleteNum=number of templates-T1, find out the templates with the oldest age in the first DeleteNum in OldIdx, and delete them from the template library.

Further optionally, performing the sorting operation on the above-mentioned data template further includes:

When the above-mentioned template quantity M exceeds the second quantity threshold T2, the template to be searched is obtained from the above-mentioned template, and the number of template members of the above-mentioned template to be searched is less than the preset number of members threshold; The above-mentioned second quantity threshold T2 is greater than the above-mentioned th a quantity threshold T1;

Calculate the age value of each of the above-mentioned templates to be searched; calculate the age reference value according to the number of the above-mentioned templates to be searched and the second duration parameter;

Obtain the template whose age value is greater than the age reference value, determine the templates with the largest age value in the top N as templates to be deleted from the templates whose age value is greater than the age reference value, and delete the template to be deleted from the template set.

Specifically, when the number of templates exceeds the second number threshold T2 (>T1), a second round of template deletion process may be started. In this embodiment of the present application, the template with a smaller number of template members may be deleted first. For example, taking the preset number of members as 3 as an example, the second round of template deletion process may include:

Determine the identifier of the template to be searched, CandidateIdx2∈{the identifier of the template whose number of template members is less than 3}.

Calculate the age value Life(t) of different templates, t∈CandidateIdx2, which is the same as the previous one, and will not be repeated here;

Different from the first round of template deletion, the age reference value calculated in this stage is a fixed age parameter, which can be as follows:

The second duration parameter includes a ₂ and b ₂ , which are fixed duration and variable duration respectively. Optionally, the calculation method of the above-mentioned age reference value can be set as required, such as modifying the above-mentioned second duration parameter, which is not limited here. ; CandidateNum is the number of templates contained in CandidateIdx2;

Search all template indexes OldIdx satisfying Life(t)>Life _Thre ;

Calculate the number N of templates to be deleted (DeleteNum), DeleteNum=number of templates-T1, find out the template indexes of the oldest DeleteNum templates in OldIdx, and delete them from the template library.

Other template sorting processing may also be set and executed as required, which is not limited in this embodiment of the present application.

106. Obtain the clustering result of the heartbeat data to be processed according to the template matching result and the template sorting result.

In the embodiment of the present application, template matching can be used to output template matching results, and at the same time, it can support the sorting of template libraries, and template merging can be introduced to merge similar templates; The age value adaptively deletes templates with large age and small capacity, so as to avoid the problem that the number of templates increases too quickly in the case of large interference, and has anti-interference ability; according to the template matching results and template sorting results, the heartbeat data can be obtained in time The real-time clustering results of the template are more standardized.

On the basis of the foregoing embodiments, the following uses specific application scenarios to illustrate the method for clustering electrocardiogram and heartbeat data in the embodiments of the present application.

The embodiments of the present application can be used to remotely monitor the real-time heartbeat clustering of a patient's heart activity state. The patient wears a mobile ECG collection box equipped with a remote data transmission function, and the collection box collects the patient's ECG signal and uploads it to the server. On the server side, the data is received and parsed into 12-lead Holter real-time data. After the 12-lead Holter real-time data is detected by real-time R wave, new heartbeat data information is obtained, and then the 12-lead Holter data and heartbeat data information are transmitted to the ECG signal real-time clustering module for heartbeat clustering.

After the real-time heartbeat clustering algorithm reads the data, it constructs a dual-lead Holter signal from the 12-lead Holter signal according to the main analysis lead number and the secondary analysis lead number, and then completes the signal preprocessing and ECG segment division. The former is obtained. Filter the Holter signal, which obtains the QRS waveform data of the newly detected heartbeat.

In the signal quality evaluation link, the algorithm will combine the saved recent signals and the corresponding recent heartbeat information to calculate the signal quality parameters.

For example, three cases are given below to describe the real-time clustering result with heartbeat offset, the template merging result with template offset, and the template deletion result when the number of templates is too large, which are only for illustration here.

(1) Real-time clustering results with heartbeat offset:

Considering the sampling rate fs=256Hz, set the matching threshold RE _Thre1 =0.1, and set the heartbeat offset search range ShiftLtm1=[-30,30]. Calculate the relative error RE(s,t) between the newly detected heartbeat QRS waveform and all templates in the template library under different offset conditions, find the minimum relative error RE _min =0.07, and the corresponding offset - 15, which matches template number 10. Since RE _min < RE _Thre1 , it is judged that the matching is successful, and the matching template information is updated with the newly detected heartbeat information after offset correction. For details, please refer to a schematic diagram of a template matching result after heartbeat offset search shown in FIG. 4 . Figure 4 shows the newly detected heartbeat waveform before correction and the template member heartbeat waveform contained in the matching template.

(2) Template merging results with template offsets:

Set the matching threshold RE _Thre2 = 0.12, the number of templates T = 20, the template offset search range ShiftLtm2 = [-10, 10], the unsorted template identification list UnMatchIdx = {0, 1, 2, ..., 19}, the reference template number ReferId=0, processing template number list TargetIdx={1,2,...,19}. After the offset search, find a list of similar processing template numbers that satisfies RE(s,t)<RE _{Thre2TargetIdx1} ={5,12}, then use the similar processing template information and the corresponding offset to update the reference template, and then update the unidentified template. Arrange the template identifier list UnMatchIdx={1,...,4,6,...,11,13,...,19}, and continue the template merging process. For details, refer to a schematic diagram of a template matching result after template merging when there is a template offset shown in FIG. 5 . Figure 5 shows the template member waveforms after the template merge. The three heartbeat groups in the figure represent the reference template member waveforms without offset correction and the similar target templates numbered 5 and 12 (the above-mentioned similar processing templates). ) member waveform.

(3) Template deletion results when there are too many templates:

Set the first round template deletion parameters: T1=200, a1=1500, b1=1500. Set the parameters for the second round of template deletion: T2=300, a2=34800, b2=23000. During the real-time clustering process, the number of templates in the template library will continue to increase due to the introduction of sporadic interference. When the number of templates exceeds T1, the first template deletion process starts. When encountering large interference, the number of templates increases rapidly. At this time, the newly added templates cannot be deleted in time during the first template deletion process. When the number of templates exceeds T2, the second template deletion process begins. For details, please refer to a schematic diagram of a template matching result after template deletion when the number of templates is too large as shown in FIG. 6 . FIG. 6 shows several deleted templates, which may have signals of different waveforms.

The methods used in ECG template matching can be roughly divided into two categories, one is the template matching method, and the other is the feature parameter method. The template matching method is to directly match the newly acquired ECG signal with the existing template, and then determine the similarity between the two by calculating the correlation coefficient. If the correlation coefficient is greater than a given threshold, the ECG signal is considered to be similar to a specific template, otherwise a new template is established. In general, the template matching method requires a prior determination of a known template library for comparison, and may not be adaptable when encountering rare special cases. The characteristic parameter method will perform signal transformation on the ECG signal in advance before template matching, usually using wavelet transformation. The transformed signal has a good degree of discrimination in certain frequency bands or regions, so it can be used for template matching. However, the signal transformation method adopted by the characteristic parameter method has a large amount of calculation, so the algorithm takes a long time. In addition, there are some methods, but they are only applicable to offline clustering scenarios and cannot be applied to real-time clustering scenarios.

Compared with general methods, the ECG heartbeat data clustering method in the embodiment of the present application has the following advantages:

(1) No signal transformation is required, and template matching is performed directly on the basis of the original filtered data, with high computational efficiency;

(2) Multi-lead joint matching, avoiding the situation of template mixing when single lead matching is used, and the matching accuracy is high;

(3) Introduce an offset search mechanism to avoid the problem of too many templates caused by heartbeat offset and template offset, and reduce template information storage overhead;

(4) The template deletion mechanism is introduced. When there are too many templates in the template library, the templates with large age and small capacity will be deleted adaptively according to the age of the template, so as to avoid the problem that the number of templates increases too quickly in the case of large interference. , with anti-interference ability.

Based on the description of the above embodiments of the method for clustering ECG heartbeat data, an embodiment of the present application further discloses a device for clustering ECG heartbeat data. Please refer to the schematic structural diagram of an electrocardiogram and heartbeat data clustering device shown in FIG. 7 , the electrocardiogram and heartbeat data clustering device 700 includes a template matching module 710, a template sorting module 720 and a clustering module 730, wherein:

The above template matching module 710 is used for:

Obtaining heartbeat data to be processed, and comparing the above-mentioned heartbeat data to be processed with multiple data templates;

If there is a target data template matching the above-mentioned heartbeat data to be processed, use the information of the above-mentioned heartbeat data to be processed to update the information of the above-mentioned target data template;

If there is no target data template matching the above-mentioned heartbeat data to be processed, a new data template is established according to the above-mentioned heartbeat data to be processed, and the above-mentioned newly-added data template corresponds to the newly-added template identifier; the above-mentioned newly-added data template is determined to be the same as the above-mentioned The above-mentioned target data template matched by the heartbeat data to be processed;

outputting the template matching result of the above-mentioned heartbeat data to be processed, the above-mentioned template matching result including the template identifier of the above-mentioned target data template;

The above-mentioned template sorting module 720 is configured to perform sorting operations on the above-mentioned data templates whenever the number of the above-mentioned heartbeat data to be processed reaches a preset number threshold, and the above-mentioned sorting operations include a template merging operation, a template sorting operation and/ Or template delete operation to obtain template sorting results;

The clustering module 730 is configured to obtain the clustering result of the heartbeat data to be processed according to the template matching result and the template sorting result.

According to an embodiment of the present application, each step involved in the method shown in FIG. 1 may be performed by each module in the apparatus 700 for clustering electrocardiogram and heartbeat data shown in FIG. 7 , which will not be repeated here.

The electrocardiogram heartbeat data clustering device 700 in this embodiment of the present application can acquire heartbeat data to be processed, and compare the heartbeat data to be processed with multiple data templates; matching target data template, use the information of the heartbeat data to be processed to update the information of the target data template; if there is no target data template matching the heartbeat data to be processed, according to the heartbeat data to be processed establishing a new data template, where the new data template corresponds to a new template identifier; determining the new data template as the target data template matching the to-be-processed heartbeat data; outputting the to-be-processed heartbeat The template matching result of the data, the template matching result includes the template identifier of the target data template; whenever the number of the processed heartbeat data to be processed reaches a preset number threshold, the data template is sorted operation, the sorting operation includes a template merging operation, a template sorting operation and/or a template deletion operation, and a template sorting result is obtained; according to the template matching result and the template sorting result, the clustering of the heartbeat data to be processed is obtained As a result, multiple heartbeats with similar morphologies can be aggregated into one template through template matching, and the templates can be integrated in time, so that real-time clustering can be realized without complex signal transformation, and the data processing efficiency is higher.

Based on the descriptions of the foregoing method embodiments and apparatus embodiments, the embodiments of the present application further provide an electronic device. Referring to FIG. 8 , the electronic device 800 at least includes a processor 801 , an input device 802 , an output device 803 and a computer storage medium 804 . The processor 801, the input device 802, the output device 803, and the computer storage medium 804 in the electronic device 800 may be connected through a bus or other means.

The computer storage medium 804 may be stored in the memory of the electronic device 800 . The computer storage medium 804 is used for storing computer programs, and the computer program includes program instructions. The processor 801 is used for executing the program instructions stored in the computer storage medium 804 . The processor 801 (or called CPU (Central Processing Unit, central processing unit)) is the computing core and the control core of the electronic device 800, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions to thereby Implement the corresponding method flow or corresponding function; in an embodiment, the processor 801 in the embodiment of the present application may be used to perform a series of processing, including some or all of the methods in the embodiment shown in FIG. 1 , and so on.

Embodiments of the present application further provide a computer storage medium (Memory), where the above-mentioned computer storage medium is a memory device in the electronic device 800 for storing programs and data. It can be understood that, the computer storage medium here may include both the built-in storage medium in the electronic device 800 , and certainly also the extended storage medium supported by the electronic device 800 . The computer storage medium provides storage space in which the operating system of the electronic device 800 is stored. In addition, one or more instructions suitable for being loaded and executed by the processor 801 are also stored in the storage space, and these instructions may be one or more computer programs (including program codes). It should be noted that the computer storage medium here can be a high-speed RAM memory, or a non-volatile memory (non-volatile memory), such as at least one magnetic disk memory; optionally, at least one storage medium located far away from the aforementioned processor can also be used. computer storage media.

In one embodiment, one or more instructions stored in the computer storage medium can be loaded and executed by the processor 801 to implement the corresponding steps in the foregoing embodiment; in specific implementation, one or more instructions in the computer storage medium can be Any steps of the method shown in FIG. 1 are loaded and executed by the processor 801, and are not repeated here.

Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the devices and modules described above can refer to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

In the several embodiments provided in this application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the division of the module is only for one logical function division, and there may be other division methods in actual implementation, for example, multiple modules or components may be combined or integrated into another system, or some features may be ignored, or not implement. The shown or discussed mutual coupling, or direct coupling, or communication connection may be through some interfaces, indirect coupling or communication connection of devices or modules, and may be in electrical, mechanical or other forms.

Modules described as separate components may or may not be physically separated, and components shown as modules may or may not be physical modules, that is, may be located in one place, or may be distributed to multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution in this embodiment.

In the above-mentioned embodiments, it may be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented in software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the procedures or functions according to the embodiments of the present application are generated in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable device. The computer instructions may be stored in or transmitted over a computer-readable storage medium. The computer instructions can be sent from one website site, computer, server, or data center to another by wire (eg, coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (eg, infrared, wireless, microwave, etc.) A website site, computer, server or data center for transmission. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that includes an integration of one or more available media. The available media may be read-only memory (ROM), or random access memory (RAM), or magnetic media, such as floppy disks, hard disks, magnetic tapes, magnetic disks, or optical media, such as, A digital versatile disc (DVD), or a semiconductor medium, for example, a solid state disk (SSD) and the like.

Claims

A method for clustering electrocardiogram heartbeat data, comprising:

Obtaining heartbeat data to be processed, and comparing the heartbeat data to be processed with multiple data templates;

If there is a target data template matching the heartbeat data to be processed, update the information of the target data template using the information of the heartbeat data to be processed;

If there is no target data template matching the heartbeat data to be processed, a new data template is created according to the heartbeat data to be processed, and the new data template corresponds to the new template identifier; determining the target data template that matches the heartbeat data to be processed;

outputting a template matching result of the heartbeat data to be processed, where the template matching result includes a template identifier of the target data template;

Whenever the number of the processed heartbeat data to be processed reaches a preset number threshold, a sorting operation is performed on the data template, and the sorting operation includes a template merging operation, a template sorting operation and/or a template deleting operation, Get the template finishing result;

According to the template matching result and the template sorting result, the clustering result of the heartbeat data to be processed is obtained.
The electrocardiogram heartbeat data clustering method according to claim 1, wherein the comparing the to-be-processed heartbeat data with a plurality of data templates in a template library comprises:

respectively acquiring multiple relative errors between the waveform of the heartbeat data to be processed and the representative waveforms of the multiple data templates;

The minimum error among the plurality of relative errors is acquired, and if the minimum error is smaller than a preset error threshold, it is determined that the target data template corresponding to the minimum error matches the heartbeat data to be processed.
The method for clustering ECG heartbeat data according to claim 2, wherein when the sorting operation includes a template merging operation, the sorting operation on the data template comprises:

Calculate the relative error between the processing template and the reference template, compare the relative error with a preset matching threshold, the reference template is one template in the unsorted template, and the processing template is the unsorted template except the Templates other than reference templates;

acquiring a processing template with the relative error less than the preset matching threshold, combining the processing template with the relative error less than the preset matching threshold with the corresponding reference template, and recording the obtained template as an organized template;

The above steps are repeated until the number of uncollated templates is zero.
The method for clustering ECG heartbeat data according to claim 3, wherein, in the case that the sorting operation includes a template sorting operation, the sorting operation on the data template comprises:

The multiple templates are sorted according to the descending order of the number of template members of each template in the multiple templates.
The method for clustering ECG heartbeat data according to any one of claims 2-4, wherein, in the case that the sorting operation includes a template deletion operation, the sorting operation on the data template comprises:

When the number M of templates exceeds the first number threshold T1, the age value of each template is calculated, and the age value is the time interval between the newly added template in the template set and the acquisition of the heartbeat data to be processed data;

According to the age value of each template and the first duration parameter, calculate the age parameter corresponding to each template;

Obtain the template whose age value is greater than the age parameter corresponding to the age value; determine from the templates whose age value is greater than the age parameter that the first N templates with the largest age value are templates to be deleted, and select the templates from the template set Delete the template to be deleted; N=M-T1.
The method for clustering electrocardiogram heartbeat data according to claim 5, wherein in the case that the sorting operation is a template deletion operation, the performing sorting operation on the data template further comprises:

When the number of templates M exceeds the second number threshold T2, the template to be searched is obtained from the template, and the number of template members of the template to be searched is less than the preset number of members threshold; the second number threshold T2 is greater than the first number threshold T1;

Calculate the age value of each of the templates to be searched; calculate the age reference value according to the number of the templates to be searched and the second duration parameter;

Obtain the template whose age value is greater than the age reference value, determine that the templates with the largest age value in the top N are templates to be deleted from the templates whose age value is greater than the age reference value, and delete all templates from the template set. Describe the template to be deleted.
The method for clustering electrocardiogram heartbeat data according to claim 1 or 6, wherein before acquiring the heartbeat data to be processed, the method further comprises:

obtaining electrocardiogram data, dividing the electrocardiogram data, and obtaining heart beat segments;

The acquiring the heartbeat data to be processed includes:

The to-be-processed heartbeat data of a preset signal length is acquired from the heartbeat segment.
The method for clustering electrocardiogram heartbeat data according to claim 7, characterized in that, before acquiring the electrocardiogram data, the method further comprises:

Acquiring the collected multi-lead ECG data; constructing dual-lead ECG data according to the main analysis lead number and the sub-analysis lead number in the multi-lead ECG data;

Band-pass filtering is performed on the two-lead electrocardiogram data to obtain the electrocardiogram data;

The dividing the electrocardiogram data to obtain a plurality of heart beat segments includes:

The electrocardiogram data is divided according to the R wave position information obtained by detection, and a plurality of heartbeat segments of equal length centered on the R wave position are obtained.
The method for clustering electrocardiogram heartbeat data according to claim 1, wherein before said comparing the heartbeat data to be processed with a plurality of data templates in a template library, the method further comprises:

calculating the signal-to-noise ratio of the heartbeat data to be processed;

determining the signal quality parameter of the heartbeat data to be processed according to the signal-to-noise ratio of the heartbeat data to be processed;

The creating a new data template according to the heartbeat data to be processed includes:

When the signal quality parameter of the heartbeat data to be processed is the first quality threshold, a new data template is established according to the heartbeat data to be processed.
An electrocardiogram heartbeat data clustering device, characterized in that it comprises a template matching module, a template sorting module and a clustering module, wherein:

The template matching module is used for:

Obtaining heartbeat data to be processed, and comparing the heartbeat data to be processed with multiple data templates;

If there is a target data template matching the heartbeat data to be processed, update the information of the target data template using the information of the heartbeat data to be processed;

If there is no target data template matching the heartbeat data to be processed, a new data template is created according to the heartbeat data to be processed, and the new data template corresponds to the new template identifier; determining the target data template that matches the heartbeat data to be processed;

outputting a template matching result of the heartbeat data to be processed, where the template matching result includes a template identifier of the target data template;

The template sorting module is configured to perform sorting operations on the data templates whenever the number of processed heartbeat data to be processed reaches a preset number threshold, and the sorting operations include template merging operations, template sorting operations operation and/or template deletion to obtain template sorting results;

The clustering module is configured to obtain the clustering result of the heartbeat data to be processed according to the template matching result and the template sorting result.
An electronic device, characterized in that it comprises a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor causes the processor to execute any one of claims 1 to 9 The steps of the ECG heartbeat data clustering method described in the item.
A computer-readable storage medium, characterized in that it stores a computer program, and when the computer program is executed by a processor, causes the processor to execute the electrocardiogram heartbeat data according to any one of claims 1 to 9 Steps of the clustering method.