CN111345814B - Analysis method, device, equipment and storage medium for electrocardiosignal center beat - Google Patents

Analysis method, device, equipment and storage medium for electrocardiosignal center beat Download PDF

Info

Publication number
CN111345814B
CN111345814B CN202010085728.4A CN202010085728A CN111345814B CN 111345814 B CN111345814 B CN 111345814B CN 202010085728 A CN202010085728 A CN 202010085728A CN 111345814 B CN111345814 B CN 111345814B
Authority
CN
China
Prior art keywords
point
loss
anchor
heart beat
scales
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202010085728.4A
Other languages
Chinese (zh)
Other versions
CN111345814A (en
Inventor
赵巍
胡静
贾东亚
王红梅
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangzhou Shiyuan Electronics Thecnology Co Ltd
Original Assignee
Guangzhou Shiyuan Electronics Thecnology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Guangzhou Shiyuan Electronics Thecnology Co Ltd filed Critical Guangzhou Shiyuan Electronics Thecnology Co Ltd
Priority to CN202010085728.4A priority Critical patent/CN111345814B/en
Publication of CN111345814A publication Critical patent/CN111345814A/en
Application granted granted Critical
Publication of CN111345814B publication Critical patent/CN111345814B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • A61B5/346—Analysis of electrocardiograms
    • A61B5/349—Detecting specific parameters of the electrocardiograph cycle
    • A61B5/366—Detecting abnormal QRS complex, e.g. widening
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Cardiology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Molecular Biology (AREA)
  • Pathology (AREA)
  • Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Biophysics (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)

Abstract

The embodiment of the invention discloses an analysis method, a device, equipment and a storage medium for electrocardiosignal center beat. The method comprises the following steps: preprocessing the acquired electrocardiosignals to obtain a plurality of sections of sample signals with set lengths; carrying out feature extraction on the sample signal to generate feature vectors of the sample signal on three scales; constructing anchor points with preset lengths according to the characteristic vectors, and calculating the reference point position of the heartbeat according to the anchor points; inputting the characteristic vector of each heart beat in the corresponding width range into a type identification module, sequentially calculating a convolution layer, an activation layer, a batch normalization layer and a linear full-connection layer of the type identification module, and outputting a five-dimensional vector, wherein the five-dimensional vector is used for representing the probability that the heart beats correspond to five heart beat types. The embodiment realizes the automatic synchronous training of the modules, and saves the training time.

Description

Analysis method, device, equipment and storage medium for electrocardiosignal center beat
Technical Field
The embodiment of the invention relates to the technical field of signal processing, in particular to an analysis method, device, equipment and storage medium for electrocardiosignal center beat.
Background
At present, the number of cardiovascular disease patients in China is about 2.6 hundred million, the cardiovascular death rate accounts for the first part of the death of urban and rural residents, and the number of sick people is still continuously increased. The cardiac electrical signals reflect the electrophysiological processes of heart activity. Because of low cost and convenient use, the examination based on the electrocardio signals is widely used for examination and diagnosis of cardiovascular diseases. A typical electrocardiographic waveform is shown in figure 1. The normal electrocardiosignals generally consist of P waves, QRS complexes and T waves, and sometimes u waves. Where the P wave represents the electrical activity of atrial contraction and the QRS wave and T represent the electrical activity of ventricular contraction and relaxation, respectively.
The electrocardiogram analysis algorithm is an important component of diagnosis and analysis equipment such as an electrocardiogram machine, an electrocardiogram monitor and the like. The electrocardio analysis algorithm can detect various diseases and give an alarm in time by measuring and classifying electrocardiosignals. In the actual use process, the electrocardiographic analysis generally includes first performing heart beat position detection (also referred to as QRS detection), then performing reference point detection (such as QRS wave start point and QRS wave end point) and type identification (normal, supraventricular abnormality, ventricular abnormality, etc.) on the detected heart beat position, and finally outputting an analysis result. Once the heartbeat position output by the heartbeat position detection module changes, the performance of subsequent reference point detection and heartbeat type identification can be seriously influenced. The existing algorithm usually designs different functional modules for different steps and tests the functional modules respectively. The method has the advantages that the research and development process of a single module is simple, the performance of each module can be tested by marking information after the input and output modes before each module are specified, but the method has the defect that the joint debugging process among the modules is complicated.
Disclosure of Invention
The invention provides an analysis method, device, equipment and storage medium for confirming central beating of an electrocardiosignal, and aims to solve the technical problem that the joint debugging process among a plurality of modules is complicated based on step refinement in the prior art.
In a first aspect, an embodiment of the present invention provides an analysis method for a cardiac signal center beat, including:
preprocessing the acquired electrocardiosignals to obtain a plurality of sections of sample signals with set lengths;
performing feature extraction on the sample signal to generate feature vectors of the sample signal on three scales, wherein the lengths of the feature vectors on the three scales are one half, one quarter and one eighth of the set length respectively;
constructing an anchor point with a preset length according to the characteristic vector corresponding to the sample signal, and calculating the reference point position of the heartbeat according to the anchor point;
inputting the characteristic vector of each heart beat in the corresponding reference point position range into a type identification module, sequentially calculating a convolution layer, an activation layer, a batch normalization layer and a linear full-connection layer of the type identification module, and outputting a five-dimensional vector, wherein the five-dimensional vector is used for representing the probability that the heart beats correspond to five heart beat types.
Wherein, the constructing an anchor point with a preset length according to the feature vector corresponding to the sample signal and calculating the reference point position of the heartbeat according to the anchor point comprise:
for each feature vector corresponding to each scale, taking each sample point on the feature vector as a center, constructing an anchor point with the length of 9 sample points, wherein the fraction of the anchor point is the value of the central sample point of the anchor point;
inputting the feature vector corresponding to the anchor point into a heart beat position detection and measurement module to obtain the fraction of the heart beat probability, the variable quantity from the anchor point to the boundary box and the continuous time proportion of each wave band;
sorting all the anchor points according to the scores from large to small to generate a check list;
sequentially screening candidate regions of the anchor points in the inspection list, adding the anchor points screened each time into the candidate list as candidate regions, deleting the anchor points and other anchor points within 0.2 seconds of the anchor points in the inspection list, wherein the anchor points screened each time are the anchor points with the highest score, the starting points of the anchor points are more than 0, and the end points of the anchor points are smaller than the length of the corresponding feature vectors in the inspection list;
calculating the midpoint and the width of a heart beat boundary box according to the midpoint and the width of the anchor points in the candidate list and the variation, and calculating the starting point and the end point of the heart beat boundary box according to the midpoint and the width of the heart beat boundary box;
and calculating the positions of the five datum points of the heartbeat according to the starting point, the width and the time proportion of the heartbeat boundary frame.
Wherein, the performing feature extraction on the sample signal to generate feature vectors of the sample signal on three scales, the lengths of the feature vectors on the three scales are respectively one half, one quarter and one eighth of the set length, and the method includes:
inputting the sample signal into an initial feature extraction module to obtain an initial feature extraction result, carrying out feature extraction on the initial feature extraction result according to corresponding scale grades by the feature extraction modules corresponding to three scales, wherein the lengths of feature vectors on the three scales are respectively one half, one quarter and one eighth of the set length;
and outputting the feature extraction result corresponding to each scale to the corresponding convolution layer and synthesizing the feature extraction results on the adjacent scales to generate the feature vectors of the sample signal on the three scales.
Wherein the total loss of the type recognition module in the training process is calculated by the following formula:
loss＝det_loss+bbx_loss+fiducial_loss+cls_loss
wherein, loss is total loss, det _ loss is heart beat position detection loss, bbx _ loss is heart beat boundary box range detection loss, fit _ loss is reference point position detection loss and cls _ loss is heart beat type classification loss.
Wherein the heart beat position detection loss is calculated by the following formula:
Figure GDA0004073600780000031
wherein, score i Represents the output score (0) of the ith anchor point at the heart beat position detection and measurement module<score i <1)，f i Indicating the type of anchor point, f when the absolute value of the difference between the anchor point position and the true heart beat position is less than 0.15 second i =1, otherwise f i ＝0。
Wherein, the heart beat boundary box range detection loss bbx _ loss is calculated by the following formula:
Figure GDA0004073600780000032
the fiducial position detection loss, fidual _ loss, is calculated by the following formula:
Figure GDA0004073600780000033
Figure GDA0004073600780000034
Figure GDA0004073600780000035
/>
Figure GDA0004073600780000036
wherein, the start point and the end point of the heart beat corresponding to the jth anchor point are respectively
Figure GDA0004073600780000037
And &>
Figure GDA0004073600780000038
The P wave starting point, the P wave end point, the QRS wave starting point, the QRS wave end point and the T wave end point corresponding to the jth anchor point are respectively
Figure GDA0004073600780000041
And &>
Figure GDA0004073600780000042
Wherein the type classification penalty is calculated by the following formula:
Figure GDA0004073600780000043
wherein
Figure GDA0004073600780000044
Indicating the probability that the jth anchor belongs to a type k heartbeat>
Figure GDA0004073600780000045
Figure GDA0004073600780000046
Indicating whether the anchor point belongs to the kth type and if so->
Figure GDA0004073600780000047
Or vice versa>
Figure GDA0004073600780000048
In a second aspect, an embodiment of the present invention further provides an apparatus for analyzing a cardiac signal center beat, including:
the preprocessing unit is used for preprocessing the acquired electrocardiosignals to obtain a plurality of sections of sample signals with set lengths;
the characteristic extraction unit is used for extracting characteristics of the sample signal and generating characteristic vectors of the sample signal on three scales, and the lengths of the characteristic vectors on the three scales are respectively one half, one quarter and one eighth of the set length;
the reference point detection unit is used for constructing an anchor point with a preset length according to the characteristic vector corresponding to the sample signal and calculating the position of the reference point of the heartbeat according to the anchor point;
the heart beat classifying unit is used for inputting each feature vector of the heart beats in the corresponding reference point position range into the type identification module, sequentially calculating the convolution layer, the activation layer, the batch normalization layer and the linear full-connection layer of the type identification module and outputting a five-dimensional vector, wherein the five-dimensional vector is used for representing the probability that the heart beats correspond to five heart beat types.
Wherein the reference point detecting unit includes:
an anchor point calculation module, configured to construct, for each feature vector corresponding to each scale, an anchor point with a length of 9 sample points by taking each sample point on the feature vector as a center, where a score of the anchor point is a value of a central sample point of the anchor point;
the data output module is used for inputting the feature vector corresponding to the anchor point into the heartbeat position detection and measurement module to obtain the fraction of the heartbeat probability, the variable quantity from the anchor point to the boundary box and the continuous time proportion of each wave band;
the anchor point sequencing module is used for sequencing all the anchor points from large to small according to the scores to generate a check list;
the candidate region screening module is used for sequentially screening the candidate regions of the anchor points in the inspection list, adding the anchor points screened each time as candidate regions into the candidate list, deleting the anchor points and other anchor points within 0.2 second of the distance from the anchor points in the inspection list, wherein the anchor points screened each time are the anchor points with the highest score, the starting points of the anchor points are greater than 0, and the end points of the anchor points are smaller than the lengths of the corresponding feature vectors in the inspection list;
the range calculation module is used for calculating the midpoint and the width of the heart beat boundary box according to the midpoint and the width of the anchor points in the candidate list and the variation, and calculating the starting point and the ending point of the heart beat boundary box according to the midpoint and the width of the heart beat boundary box;
and the datum point calculating module is used for calculating the positions of the five datum points of the heartbeat according to the starting point, the width and the time proportion of the heartbeat boundary frame.
Wherein the feature extraction unit includes:
the characteristic extraction module is used for inputting the sample signal into the initial characteristic extraction module to obtain an initial characteristic extraction result, the characteristic extraction modules corresponding to three scales perform characteristic extraction on the initial characteristic extraction result according to corresponding scale grades, and the lengths of characteristic vectors on the three scales are respectively one half, one quarter and one eighth of the set length;
and the feature vector generation module is used for outputting the feature extraction result corresponding to each scale to the corresponding convolution layer and synthesizing the feature extraction results on the adjacent scales to generate the feature vectors of the sample signal on the three scales.
Wherein the total loss of the type recognition module in the training process is calculated by the following formula:
loss＝det_loss+bbx_loss+fiducial_loss+cls_loss
wherein, loss is total loss, det _ loss is heart beat position detection loss, bbx _ loss is heart beat boundary box range detection loss, fit _ loss is reference point position detection loss and cls _ loss is heart beat type classification loss.
Wherein the heart beat position detection loss is calculated by the following formula:
Figure GDA0004073600780000051
wherein, score i Represents the output score (0) of the ith anchor point at the heart beat position detection and measurement module<score i <1)，f i Indicating the type of anchor point, f when the absolute value of the difference between the anchor point position and the true heartbeat position is less than 0.15 seconds i =1, otherwise f i ＝0。
Wherein, the heart beat boundary box range detection loss bbx _ loss is calculated by the following formula:
Figure GDA0004073600780000052
the fiducial position detection loss, fidual _ loss, is calculated by the following formula:
Figure GDA0004073600780000061
Figure GDA0004073600780000062
Figure GDA0004073600780000063
Figure GDA0004073600780000064
wherein, the start point and the end point of the heart beat corresponding to the jth anchor point are respectively
Figure GDA0004073600780000065
And &>
Figure GDA0004073600780000066
The starting point of P wave, the ending point of P wave, the starting point of QRS wave, the ending point of QRS wave and the ending point of T wave corresponding to the jth anchor point are respectively
Figure GDA0004073600780000067
And &>
Figure GDA0004073600780000068
Wherein the type classification penalty is calculated by the following formula:
Figure GDA0004073600780000069
wherein the content of the first and second substances,
Figure GDA00040736007800000610
indicating the probability that the jth anchor belongs to a type k heartbeat>
Figure GDA00040736007800000611
/>
Figure GDA00040736007800000612
Indicating whether the anchor point belongs to the kth type and if so->
Figure GDA00040736007800000613
Or vice versa>
Figure GDA00040736007800000614
In a third aspect, an embodiment of the present invention further provides an apparatus, where the apparatus includes:
one or more processors;
a memory for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement the method for analyzing a center beat of an electrocardiographic signal according to any one of the first aspect.
In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method for analyzing a cardiac beat of an electrocardiograph signal according to any one of the first aspect.
According to the analysis method, the device, the equipment and the storage medium for the electrocardiosignal center beat, a plurality of sections of sample signals with set lengths are obtained by preprocessing the collected electrocardiosignals; performing feature extraction on the sample signal to generate feature vectors of the sample signal on three scales, wherein the lengths of the feature vectors on the three scales are respectively one half, one quarter and one eighth of the set length; constructing an anchor point with a preset length according to the characteristic vector corresponding to the sample signal, and calculating the reference point position of the heartbeat according to the anchor point; inputting the characteristic vector of each heart beat in the corresponding reference point position range into a type identification module, sequentially calculating a convolution layer, an activation layer, a batch normalization layer and a linear full-connection layer of the type identification module, and outputting a five-dimensional vector, wherein the five-dimensional vector is used for representing the probability that the heart beats correspond to five heart beat types. According to the deep learning-based method, the types of the datum positions of the heart beats are output through an end-to-end full-automatic analysis process, automatic synchronous training of all modules is achieved, the training process is simplified, the comprehensive performance of each link is kept, and the training time is saved.
Drawings
FIG. 1 is a schematic diagram of a structure of an ECG signal;
FIG. 2 is a flowchart of a method for analyzing a cardiac beat according to an embodiment of the present invention;
FIGS. 3-6 are schematic diagrams illustrating a variation of an electrocardiosignal processing procedure according to an embodiment of the present invention;
FIG. 7 is a schematic diagram of data flow during processing of an ECG signal according to an embodiment of the present invention;
fig. 8 is a flow chart of feature extraction of an analysis method for cardiac electrical signal center beat according to a second embodiment of the present invention;
FIG. 9 is a schematic diagram of data flow during feature extraction according to a second embodiment of the present invention;
10-13 are schematic diagrams of data processing flow of the feature extraction module according to the second embodiment of the present invention;
FIG. 14 is a flowchart illustrating the calculation of the reference point position according to the second embodiment of the present invention;
FIG. 15 is a schematic diagram of a data processing flow for calculating the position of the datum point according to a second embodiment of the present invention;
FIG. 16 is a schematic diagram of a data processing flow of a second-center beat classification according to an embodiment of the present invention;
fig. 17 is a schematic structural diagram of an analysis apparatus for cardiac signal center beat according to a third embodiment of the present disclosure;
fig. 18 is a schematic structural diagram of an apparatus according to a fourth embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are for purposes of illustration and not limitation. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
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 application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is explicitly and implicitly understood by one skilled in the art that the embodiments described herein can be combined with other embodiments.
Example one
Fig. 2 is a flowchart of an analysis method for cardiac electrical signal center beat according to an embodiment of the present invention. The analysis method for the electrocardiosignal center beat provided in the embodiment can be executed by a detection analysis device for the QRS wave in the electrocardiosignal, the detection device for the QRS wave in the electrocardiosignal can be realized in a software and/or hardware mode, and the analysis device for the electrocardiosignal center beat can be formed by two or more physical entities or one physical entity. For example, the analysis device for the cardiac signal center beat can be a mobile phone, an industrial control computer, and the like.
As shown in fig. 2, the method for analyzing a cardiac signal center beat provided in the first embodiment includes the following steps:
step S110: preprocessing the acquired electrocardiosignals to obtain a plurality of sections of sample signals with set lengths.
The preprocessing mainly includes the waveform adjustment of signals such as resampling and filtering, and specifically, for the most original electrocardiosignals, the electrocardiosignals are firstly resampled to 256Hz (fs = 256), and then are subjected to band-pass filtering by using a filter with a pass band ranging from 0.5Hz to 40 Hz. For an electrocardiographic signal of an individual, the length of an electrocardiographic signal segment used for analysis is generally about 10 seconds, that is, for an electrocardiographic signal of an individual, 2560 sample points exist in the electrocardiographic signal with the length of 10 seconds, which is obtained in this embodiment by performing the integration, and the electrocardiographic signal after resampling is assumed to be s i I =1, \ 8230;, n (n = 2560), fig. 3 and 4 show the cardiac electrical signal before and after filtering. Comparing fig. 3 and 4, it can be seen that the band pass filtering filters out "glitches" in the signal.
Step S120: and performing feature extraction on the sample signal to generate feature vectors of the sample signal on three scales, wherein the lengths of the feature vectors on the three scales are respectively one half, one quarter and one eighth of the set length.
The lengths of the feature vectors in the three dimensions are respectively one half, one quarter and one eighth of the set length, and for simplicity, the feature vectors are respectively defined as a first layer, a second layer and a third layer. Where the first layer is used for detection of pacemaker signals, the second layer for detection of normal width beats, and the third layer for detection of wide malformed beats, such as ventricular premature beats, the set of the three constituting a full beat.
Step S130: and constructing an anchor point with a preset length according to the characteristic vector corresponding to the sample signal, and calculating the reference point position of the heartbeat according to the anchor point.
Based on the values of the anchor points with the set length, the sample signals can be scored and sorted, and the candidate regions of the suspected heart beats are roughly identified, as shown in fig. 5, in which the positions of a plurality of suspected QRS waves are identified and marked with "x", and in addition to the part with the larger peak, a considerable number of the heart waves with the smaller peak are marked.
After the suspected QRS wave is preliminarily confirmed, the candidate region and the feature vector which are correspondingly obtained are input into a candidate region detection module, and whether the signal of the candidate region is the QRS wave can be further accurately judged. On the basis of the candidate regions shown in fig. 5, the signal marked by "x" in fig. 6 is judged as QRS wave, and in fig. 3-6, the wave corresponding to "·" is the actual QRS wave, so that it is clear that all QRS waves are accurately judged in fig. 6 after the above processing.
In the process of heartbeat detection, the candidate region detection module can more accurately judge the width range corresponding to each heartbeat, the width range is expressed by data through the datum point, and then the heartbeat type is judged based on the corresponding feature vector in the width range.
Step S140: inputting the characteristic vector of each heart beat in the corresponding reference point position range into a type identification module, sequentially calculating a convolution layer, an activation layer, a batch normalization layer and a linear full-connection layer of the type identification module, and outputting a five-dimensional vector, wherein the five-dimensional vector is used for representing the probability that the heart beats correspond to five heart beat types.
The feature vectors in the reference point position range corresponding to each heart beat are input into the type identification module, calculation is sequentially carried out on the convolution layer, the activation layer, the batch normalization layer and the linear full-connection layer, a five-dimensional vector result can be obtained for each heart beat, the five-dimensional vector shows the probability that the heart beat may correspond to the five-dimensional heart beat type, the type with the highest probability can be used as the heart beat type of the heart beat on the basis of the five-dimensional vector, and if the probabilities corresponding to the multiple types are close, the heart beat can be marked independently for manual judgment.
The data flow process is further visualized in fig. 7, so that the most original electrocardiosignals are obtained, then the electrocardiosignals are preprocessed to obtain preliminary signals capable of performing feature extraction, and for the results of the feature extraction, the heart beat and the corresponding width detection, the results of the heart beat detection, the width measurement and the adjustment extraction are integrated, and the classification of the heart beats can be realized.
On the whole, a plurality of sections of sample signals with set lengths are obtained by preprocessing the acquired electrocardiosignals; performing feature extraction on the sample signal to generate feature vectors of the sample signal on three scales, wherein the lengths of the feature vectors on the three scales are respectively one half, one quarter and one eighth of the set length; constructing an anchor point with a preset length according to the feature vector corresponding to the sample signal, and calculating the reference point position of the heartbeat according to the anchor point; inputting the characteristic vector of each heartbeat in the corresponding reference point position range for calculating the heartbeat into a type identification module, sequentially calculating a convolution layer, an activation layer, a batch normalization layer and a linear full-connection layer of the type identification module, and outputting a five-dimensional vector, wherein the five-dimensional vector is used for representing the probability that the heartbeats correspond to five heartbeat types. According to the deep learning-based method, the types of the datum positions of the heart beats are output through an end-to-end full-automatic analysis process, automatic synchronous training of all modules is achieved, the training process is simplified, the comprehensive performance of each link is kept, and the training time is saved.
Example two
In the present embodiment, the embodiment is embodied on the basis of the above-mentioned embodiment, especially the step S120 and the step S130 are embodied, it should be noted that the embodiment presents the embodiment of the step S120 and the step S130 at the same time, and the two are not necessarily implemented at the same time, but are integrated processing for describing the solution, and in the actual processing process, the embodiment of the step S120 and the step S130 may exist as an independent implementation manner, or may be integrated with each other. In this embodiment, the whole signal processing process is fully described in a more detailed layer. As a whole, step S110, the relevant steps in fig. 8 and 14, and step S140 are included.
Step S110: preprocessing the acquired electrocardiosignals to obtain a plurality of sections of sample signals with set lengths.
Step S121: inputting the sample signal into an initial feature extraction module to obtain an initial feature extraction result, carrying out feature extraction on the initial feature extraction result according to corresponding scale grades by the feature extraction modules corresponding to three scales, wherein the lengths of feature vectors on the three scales are respectively one half, one quarter and one eighth of the set length.
Step S122: and outputting the feature extraction result corresponding to each scale to the corresponding convolution layer and generating the feature vectors of the sample signals on the three scales by integrating the feature extraction results on the adjacent scales.
The data processing in steps S121 and S122 can refer to fig. 9, where C0, C1, C2, and C3 respectively represent an initial feature extraction module and a feature extraction module corresponding to three scales, and after entering C0 (initial feature extraction module), the sample signal flows in the direction shown in fig. 9, and corresponding processing is performed in the flow process as shown in fig. 9. Taking the first layer as an example, after the sample signal enters C0, the data processing flow shown in fig. 10 (including convolutional layer, batch normalization layer, activation layer, and pooling layer) is performed; the extraction result of C0 is output to C1 (the first layer extraction module), C1 obtains the corresponding processing result according to the data processing flow shown in fig. 11 (including two convolution layers, two batch normalization layers, and two activation layers), the processing result is output to the convolution layers corresponding to C2 and C1, and the feature vector f corresponding to the first layer is obtained through data synthesis of two convolutions and one upsampling shown in fig. 9 1 . Feature extraction of the second and third layers refer to fig. 12 and 13, respectively. The data processing procedure of the second and third layers is similar, except that some of the processing parameters involved are different, for example, the number of channels (C) of the convolutional layer of C0, the convolutional kernel size (k), the step size(s) and the padding size (p) are 64, 7, 1 and 3, respectively, and the parameter combinations of the two convolutional layers of C1, C2 and C3 themselves may be different, referring specifically to fig. 11-13. Finally extractingIn the feature vector, the jth sample point (i.e. position j) on the ith scale (ith layer) feature vector is recorded as
Figure GDA0004073600780000111
Step S131: and for each feature vector corresponding to the scale, constructing an anchor point with the length of 9 sample points by taking each sample point on the feature vector as a center, wherein the fraction of the anchor point is the value of the central sample point of the anchor point.
Step S132: and inputting the characteristic vector corresponding to the anchor point into a heart beat position detection and measurement module to obtain the fraction of the heart beat probability, the variable quantity from the anchor point to the boundary box and the continuous time proportion of each waveband.
Step S133: and sequencing all the anchor points from large to small according to the scores to generate a check list.
Step S134: and sequentially screening the candidate areas of the anchor points in the inspection list, adding the anchor points screened each time into the candidate list as the candidate areas, deleting the anchor points and other anchor points within 0.2 seconds of the distance from the anchor points in the candidate list, wherein the anchor points screened each time are the anchor points with the highest score, the starting points of which are greater than 0 and the end points of which are less than the length of the corresponding feature vector in the inspection list, and the scores of the anchor points are taken as the scores of the corresponding candidate areas.
For the determination of the candidate region, steps S131 to S133 can be described by the following mathematical language: a. with each sample point on the feature vector
Figure GDA0004073600780000112
Centered, an anchor point of length 9 sample points was constructed>
Figure GDA0004073600780000113
E.g. sample points>
Figure GDA0004073600780000114
The corresponding anchor point is->
Figure GDA0004073600780000115
The score of the anchor point is->
Figure GDA0004073600780000116
(sample at anchor center). (anchor points on the layer 1 feature vector correspond to candidate regions with the length of 17 sample points on the original signal, namely 0.067 second, and are used for detecting pacemaker signals; anchor points on the layer 2 feature vector correspond to candidate regions with the length of 33 sample points on the original signal, namely 0.13 second, and are used for detecting normal width heartbeats; and anchor points on the layer 3 feature vector correspond to candidate regions with the length of 65 sample points on the original signal, namely 0.25 second, and are used for detecting wide and malformed heartbeats). B. c. Selecting the point with highest score in the checking list, wherein the anchor point starting point is greater than 0, and the anchor point ending point is less than the characteristic graph length (l) i ) Adds the anchor point of (a) as a candidate area into the candidate list and deletes the anchor point in the check list, and all other anchor points within 0.2 seconds (normally, the shortest interval between two heartbeats is 0.2 seconds) from the anchor point. d. And c, repeating the step c until the checking list is empty, taking the anchor points in the candidate list as the positions of the candidate areas, and taking the values of the anchor points as the scores of the candidate areas and outputting the scores.
Step S135: and calculating the midpoint and the width of the cardioid beat boundary box according to the midpoint and the width of the anchor points in the candidate list and the variable quantity, and calculating the starting point and the end point of the cardioid beat boundary box according to the midpoint and the width of the cardioid beat boundary box.
Step S136: and calculating the positions of the five reference points of the heart beat according to the starting point, the width and the time proportion of the heart beat boundary frame.
The overall calculation process on the basis of the feature vectors with respect to the heartbeat is as follows: firstly, anchor points are established on a feature vector, and then 1) the probability fraction (score) of each anchor point belonging to the heartbeat, 2) the variation (de lta) from the anchor point to a heartbeat boundary box and 3) the proportion (duration rate) of 4 part durations in the heartbeat boundary box are predicted simultaneously: p wave start to P wave end (P) dur ) P wave end to QRS wave start (PR) seg ) QRS wave start to QRS wave end (QRS) int ) QRS wave end to T wave end (ST) int ). Then obtaining the heart beat position (beat location) according to the probability score of each anchor point, obtaining the range (bounding box) of the heart beat boundary frame according to the variable quantity (de lta) of the boundary frame, and finally obtaining 5 reference point positions (coronary points) of the current heart beat by using the range of the heart beat boundary frame and the proportion of the duration time of each part: p wave starting point (P) onset ) End point of P wave (P) offset ) Onset of QRS wave (QRS) onset ) QRS wave end point (QRS) offset ) And T wave end point (T) offset ). The specific process is as follows:
a. with each sample point on the feature vector
Figure GDA0004073600780000121
Centered, an anchor point of length 9 sample points was constructed>
Figure GDA0004073600780000122
E.g. a sample point->
Figure GDA0004073600780000123
The corresponding anchor point is->
Figure GDA0004073600780000124
The score of the anchor point is->
Figure GDA0004073600780000125
(sample at anchor center). (the anchor point on the layer 1 feature vector corresponds to a candidate region of 17 sample points length on the original signal, i.e., 0.067 seconds, for analysis of pacemaker signals; the anchor point on the layer 2 feature vector corresponds to a candidate region of 33 sample points length on the original signal, i.e., 0.13 seconds, for analysis of normal and supraventricular abnormal heartbeats; the anchor point on the layer 3 feature vector corresponds to a candidate region of 65 sample points length on the original signal, i.e., 0.25 seconds, for analysis of ventricular abnormal and fused heartbeats).
b. And inputting the characteristic vectors corresponding to the anchor points into a heart beat position detection and measurement module, and simultaneously obtaining the heart beat probability fraction, the change value from the anchor points to the boundary box and the proportion of the duration time of each wave band.
c. For the predicted heart beat probability scores, the probability scores are sorted from large to small to generate a check list. The anchor point with the highest score is selected in the check list as a candidate area to be added into the candidate list, and the anchor point and all other anchor points within 0.2 seconds (normally, the shortest interval between two heartbeats is 0.2 seconds) of the anchor point are deleted in the check list. The above steps are repeated until the check list is empty or the remaining anchor values are less than 0.5. And finally, taking the midpoint of the anchor point in the candidate list as the heart beat position.
d. For anchor point to heart beat bounding box variance, first, the midpoint of the anchor point (anchor) is used center ) And width (anchor) width ) And its amount of change (delta) center And delta width ) Computing predicted mid-points of cardioid bounding boxes
Figure GDA0004073600780000131
And width->
Figure GDA0004073600780000132
Finally, the starting point of the predicted heartbeat bounding box is determined>
Figure GDA0004073600780000133
And endpoint>
Figure GDA0004073600780000134
Position, as shown in equation (2).
Figure GDA0004073600780000135
e. According to the starting point and the width of the heart beat bounding box and the proportion of the duration of 4 parts in the bounding box, the positions of 5 reference points are predicted by using the formula (3):
Figure GDA0004073600780000136
step S140: inputting each feature vector of the heart beats in the corresponding width range into a type identification module, sequentially calculating a convolution layer, an activation layer, a batch normalization layer and a linear full-connection layer of the type identification module, and outputting a five-dimensional vector, wherein the five-dimensional vector is used for representing the probability that the heart beats correspond to five heart beat types.
For the heart beat position detected by the heart beat position detection and measurement module, firstly, the feature vector of the corresponding position on the feature extraction module is extracted, and then the feature vector is sent to the heart beat type identification module for classification. After the computation of the convolution layer (convo l ut i on), the activation layer (re l u), the batch normal i zat i on) and the linear full-link layer (l i near), a 5-dimensional vector is output, which respectively represents the probability that the heartbeat belongs to 5 types: sinus beating, supraventricular abnormal beating, ventricular abnormal beating, fusion beating and other types of beating. And finally, taking the type with the highest probability as the type of the heart beat.
EXAMPLE III
Fig. 17 is a schematic structural diagram of an analysis apparatus for cardiac electrical signal center beat according to a third embodiment of the present invention. Referring to fig. 17, the apparatus for analyzing a cardiac signal center beat includes: a preprocessing unit 310, a feature extraction unit 320, a reference point detection unit 330, and a heartbeat classification unit 340.
The preprocessing unit 310 is configured to preprocess the acquired electrocardiographic signals to obtain a plurality of segments of sample signals with a set length; a feature extraction unit 320, configured to perform feature extraction on the sample signal, and generate feature vectors of the sample signal on three scales, where lengths of the feature vectors on the three scales are half, quarter, and eighth of the set length, respectively; the reference point detection unit 330 is configured to construct an anchor point with a preset length according to the feature vector corresponding to the sample signal, and calculate a reference point position of the heartbeat according to the anchor point; the heartbeat classifying unit 340 is configured to input the feature vector of each heartbeat in the corresponding reference point position range into the type identification module, and output a five-dimensional vector through calculation of the convolution layer, the activation layer, the batch normalization layer and the linear full-link layer of the type identification module in sequence, where the five-dimensional vector is used to characterize the probability that the heartbeat belongs to the five heartbeat types.
On the basis of the above embodiment, the reference point detecting unit 330 includes:
an anchor point calculation module, configured to construct, for each feature vector corresponding to each scale, an anchor point with a length of 9 sample points by taking each sample point on the feature vector as a center, where a score of the anchor point is a value of a central sample point of the anchor point;
the data output module is used for inputting the feature vector corresponding to the anchor point into the heartbeat position detection and measurement module to obtain the fraction of the heartbeat probability, the variable quantity from the anchor point to the boundary box and the continuous time proportion of each wave band;
the anchor point sequencing module is used for sequencing all the anchor points from large to small according to the scores to generate a check list;
the candidate region screening module is used for sequentially screening the candidate regions of the anchor points in the inspection list, adding the anchor points screened each time as candidate regions into the candidate list, deleting the anchor points and other anchor points within 0.2 second of the distance from the anchor points in the inspection list, wherein the anchor points screened each time are the anchor points with the highest score, the starting points of the anchor points are greater than 0, and the end points of the anchor points are smaller than the lengths of the corresponding feature vectors in the inspection list;
the range calculation module is used for calculating the midpoint and the width of the heart beat boundary box according to the midpoint and the width of the anchor points in the candidate list and the variation, and calculating the starting point and the ending point of the heart beat boundary box according to the midpoint and the width of the heart beat boundary box;
and the datum point calculating module is used for calculating the positions of the five datum points of the heartbeat according to the starting point, the width and the time proportion of the heartbeat boundary frame.
On the basis of the above embodiment, the feature extraction unit 320 includes:
the characteristic extraction module is used for inputting the sample signal into the initial characteristic extraction module to obtain an initial characteristic extraction result, the characteristic extraction modules corresponding to three scales perform characteristic extraction on the initial characteristic extraction result according to corresponding scale grades, and the lengths of characteristic vectors on the three scales are respectively one half, one quarter and one eighth of the set length;
and the feature vector generation module is used for outputting the feature extraction result corresponding to each scale to the corresponding convolution layer and synthesizing the feature extraction results on the adjacent scales to generate the feature vectors of the sample signal on the three scales.
On the basis of the above embodiment, the total loss of the type identification module in the training process is calculated by the following formula:
loss＝det_loss+bbx_loss+fiducial_loss+cls_loss
wherein, loss is total loss, det _ loss is heart beat position detection loss, bbx _ loss is heart beat boundary box range detection loss, fit _ loss is reference point position detection loss and cls _ loss is heart beat type classification loss.
On the basis of the above embodiment, the heartbeat position detection loss is calculated by the following formula:
Figure GDA0004073600780000151
wherein, score i Represents the output score (0) of the ith anchor point at the heart beat position detection and measurement module<score i <1)，f i Indicating the type of anchor point, f when the absolute value of the difference between the anchor point position and the true heart beat position is less than 0.15 second i =1, otherwise f i ＝0。
On the basis of the above embodiment, the heart beat bounding box range detection loss bbx _ loss is calculated by the following formula:
Figure GDA0004073600780000152
the fiducial point position detection loss fidcial _ loss is calculated by the following formula:
Figure GDA0004073600780000153
Figure GDA0004073600780000154
Figure GDA0004073600780000155
Figure GDA0004073600780000161
wherein, the start point and the end point of the heart beat corresponding to the jth anchor point are respectively
Figure GDA0004073600780000162
And &>
Figure GDA0004073600780000163
The P wave starting point, the P wave end point, the QRS wave starting point, the QRS wave end point and the T wave end point corresponding to the jth anchor point are respectively
Figure GDA0004073600780000164
And &>
Figure GDA0004073600780000165
On the basis of the above embodiment, the type classification penalty is calculated by the following formula:
Figure GDA0004073600780000166
wherein the content of the first and second substances,
Figure GDA0004073600780000167
indicating that the jth anchor belongs to the kth classProbability of a heart beat->
Figure GDA0004073600780000168
Figure GDA0004073600780000169
Indicating whether the anchor point belongs to the kth type and if so->
Figure GDA00040736007800001610
Or vice versa>
Figure GDA00040736007800001611
The analysis device for the electrocardiosignal center beat provided by the embodiment of the invention is contained in the detection equipment for the electrocardiosignal center beat, can be used for executing the analysis method for the electrocardiosignal center beat provided by any embodiment, and has corresponding functions and beneficial effects.
Example four
Fig. 18 is a schematic structural diagram of an apparatus according to a fourth embodiment of the present invention, which may be various electrocardiographs and electrocardiographs in a specific product presentation, and more specifically, may be an apparatus to which the analysis method of the cardiac signal center beat described in the foregoing embodiments is applied. As shown in fig. 18, the apparatus includes a processor 410, a memory 420, an input device 430, an output device 440, and a communication device 450; the number of the processors 410 in the device may be one or more, and one processor 410 is taken as an example in fig. 18; the processor 410, the memory 420, the input device 430, the output device 440, and the communication device 450 of the detection apparatus for cardiac electrical signal center beating may be connected by a bus or other means, and fig. 18 illustrates the connection by the bus as an example.
The memory 420 serves as a computer-readable storage medium, and can be used for storing software programs, computer-executable programs, and modules, such as program instructions/modules corresponding to the analysis method for the cardiac beat signal center in the embodiment of the present invention (for example, the preprocessing unit 310, the feature extraction unit 320, the reference point detection unit 330, and the cardiac beat classification unit 340 in the analysis device for the cardiac beat signal center). The processor 410 executes software programs, instructions and modules stored in the memory 420 to execute various functional applications and data processing of the device, that is, to implement the analysis method of the cardiac electrical signal center beat.
The memory 420 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required for at least one function; the storage data area may store data created according to use of the device, and the like. Further, the memory 420 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some examples, memory 420 may further include memory located remotely from processor 410, which may be connected to devices through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The input means 430 may be used to receive input numeric or character information and to generate key signal inputs relating to user settings and function controls of the device. The output device 440 may include a display device such as a display screen. The communication device 450 is used for data communication with the image capturing module.
The equipment comprises an analysis device for the electrocardiosignal center beat, can be used for executing an analysis method for any electrocardiosignal center beat, and has corresponding functions and beneficial effects.
EXAMPLE five
Embodiments of the present invention also provide a storage medium containing computer-executable instructions, which when executed by a computer processor, perform a method for analyzing a cardiac electrical signal center beat, the method comprising:
preprocessing the acquired electrocardiosignals to obtain a plurality of sections of sample signals with set lengths;
performing feature extraction on the sample signal to generate feature vectors of the sample signal on three scales, wherein the lengths of the feature vectors on the three scales are respectively one half, one quarter and one eighth of the set length;
constructing an anchor point with a preset length according to the characteristic vector corresponding to the sample signal, and calculating the reference point position of the heartbeat according to the anchor point;
inputting the characteristic vector of each heart beat in the corresponding reference point position range into a type identification module, sequentially calculating a convolution layer, an activation layer, a batch normalization layer and a linear full-connection layer of the type identification module, and outputting a five-dimensional vector, wherein the five-dimensional vector is used for representing the probability that the heart beats correspond to five heart beat types.
Of course, the storage medium provided by the embodiment of the present invention contains computer-executable instructions, and the computer-executable instructions are not limited to the operations of the method described above, and may also perform related operations in the analysis method for cardiac electrical signal center beat provided by any embodiment of the present invention.
From the above description of the embodiments, it is obvious for those skilled in the art that the present invention can be implemented by software and necessary general hardware, and certainly can be implemented by hardware, but the former is a better embodiment in many cases. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as a floppy disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a FLASH Memory (FLASH), a hard disk or an optical disk of a computer, and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device) to execute the methods according to the embodiments of the present invention.
It should be noted that, in the embodiment of the analysis apparatus for cardiac electrical signal center beat, the included units and modules are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, specific names of the functional units are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present invention.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims (9)

1. An analysis method for electrocardiosignal center beat is characterized by comprising the following steps:
preprocessing the acquired electrocardiosignals to obtain a plurality of sections of sample signals with set lengths;
performing feature extraction on the sample signal to generate feature vectors of the sample signal on three scales, wherein the lengths of the feature vectors on the three scales are one half, one quarter and one eighth of the set length respectively;
constructing an anchor point with a preset length according to the characteristic vector corresponding to the sample signal, and calculating the reference point position of the heartbeat according to the anchor point;
inputting the characteristic vector of each heart beat in the corresponding reference point position range into a type identification module, sequentially calculating a convolution layer, an activation layer, a batch normalization layer and a linear full-connection layer of the type identification module, and outputting a five-dimensional vector, wherein the five-dimensional vector is used for representing the probability that the heart beat corresponds to five heart beat types;
wherein, the performing feature extraction on the sample signal to generate feature vectors of the sample signal on three scales, the lengths of the feature vectors on the three scales are respectively one half, one quarter and one eighth of the set length, and the method includes:
inputting the sample signal into an initial feature extraction module to obtain an initial feature extraction result, carrying out feature extraction on the initial feature extraction result according to corresponding scale grades by the feature extraction modules corresponding to three scales, wherein the lengths of feature vectors on the three scales are respectively one half, one quarter and one eighth of the set length;
and outputting the feature extraction result corresponding to each scale to the corresponding convolution layer and generating the feature vectors of the sample signals on the three scales by integrating the feature extraction results on the adjacent scales.
2. The analysis method according to claim 1, wherein the constructing an anchor point with a preset length according to the feature vector corresponding to the sample signal, and calculating the reference point position of the heartbeat according to the anchor point comprises:
for the feature vector corresponding to each scale, constructing an anchor point with the length of 9 sample points by taking each sample point on the feature vector as a center, wherein the fraction of the anchor point is the value of the central sample point of the anchor point;
inputting the feature vector corresponding to the anchor point into a heart beat position detection and measurement module to obtain the fraction of the heart beat probability, the variable quantity from the anchor point to the boundary box and the continuous time proportion of each wave band;
sequencing all the anchor points from large to small according to the scores to generate a check list;
sequentially screening candidate regions of the anchor points in the inspection list, adding the anchor points screened each time into the candidate list as candidate regions, deleting the anchor points and other anchor points within 0.2 seconds of the anchor points in the inspection list, wherein the anchor points screened each time are the anchor points with the highest score, the starting points of the anchor points are more than 0, and the end points of the anchor points are smaller than the length of the corresponding feature vectors in the inspection list;
calculating the midpoint and the width of a heart beat boundary box according to the midpoint and the width of the anchor points in the candidate list and the variation, and calculating the starting point and the end point of the heart beat boundary box according to the midpoint and the width of the heart beat boundary box;
and calculating the positions of the five datum points of the heartbeat according to the starting point, the width and the time proportion of the heartbeat boundary frame.
3. The analysis method according to claim 1, wherein the total loss of the type recognition module during the training process is calculated by the following formula:
loss＝det_loss+bbx_loss+fiducial_loss+cls_loss
wherein, loss is total loss, det _ loss is heart beat position detection loss, bbx _ loss is heart beat boundary box range detection loss, fit _ loss is reference point position detection loss and cls _ loss is heart beat type classification loss.
4. The analysis method according to claim 3, wherein the heart beat position detection loss is calculated by the following formula:
Figure FDA0004077385950000021
wherein, score i Represents the output score (0) of the ith anchor point at the heart beat position detection and measurement module<score i <1)，f i Indicating the type of anchor point, f when the absolute value of the difference between the anchor point position and the true heart beat position is less than 0.15 second i =1, otherwise f i ＝0。
5. The analysis method according to claim 3, wherein the heart beat bounding box range detection loss bbx _ loss is calculated by the following formula:
Figure FDA0004077385950000022
the fiducial position detection loss, fidual _ loss, is calculated by the following formula:
Figure FDA0004077385950000023
Figure FDA0004077385950000031
Figure FDA0004077385950000032
Figure FDA0004077385950000033
wherein, the starting point and the end point of the real solid beat corresponding to the jth anchor point are respectively
Figure FDA0004077385950000034
And &>
Figure FDA0004077385950000035
The jth anchor point corresponds to the real P wave starting point, P wave end point, QRS wave starting point, QRS wave end point and T wave end point respectively
Figure FDA0004077385950000036
And &>
Figure FDA0004077385950000037
The jth anchor point corresponds to the start point and the end point of the predicted heartbeat and is/are respectively->
Figure FDA0004077385950000038
And &>
Figure FDA0004077385950000039
The corresponding predicted P wave starting point, P wave end point, QRS wave starting point, QRS wave end point and T wave end point of the jth anchor point are ^ and/or ^>
Figure FDA00040773859500000310
And &>
Figure FDA00040773859500000311
6. The analytical method of claim 3, wherein the type classification penalty is calculated by the formula:
Figure FDA00040773859500000312
wherein the content of the first and second substances,
Figure DA00040773859557377734
indicating the probability that the jth anchor belongs to a type k heartbeat>
Figure DA00040773859557422012
Figure FDA00040773859500000313
Represents whether the anchor point belongs to the kth type and if so->
Figure FDA00040773859500000314
Or vice versa>
Figure FDA00040773859500000315
7. An analysis device for cardiac signal center beat, comprising:
the preprocessing unit is used for preprocessing the acquired electrocardiosignals to obtain a plurality of sections of sample signals with set lengths;
the characteristic extraction unit is used for extracting characteristics of the sample signal and generating characteristic vectors of the sample signal on three scales, and the lengths of the characteristic vectors on the three scales are respectively one half, one quarter and one eighth of the set length;
the reference point detection unit is used for constructing an anchor point with a preset length according to the characteristic vector corresponding to the sample signal and calculating the reference point position of the heartbeat according to the anchor point;
the heart beat classification unit is used for inputting the feature vector of each heart beat in the corresponding reference point position range into the type identification module, sequentially calculating a convolution layer, an activation layer, a batch normalization layer and a linear full-connection layer of the type identification module, and outputting a five-dimensional vector, wherein the five-dimensional vector is used for representing the probability that the heart beat corresponds to five heart beat types;
wherein the feature extraction unit includes:
the characteristic extraction module is used for inputting the sample signal into the initial characteristic extraction module to obtain an initial characteristic extraction result, the characteristic extraction modules corresponding to three scales perform characteristic extraction on the initial characteristic extraction result according to corresponding scale grades, and the lengths of characteristic vectors on the three scales are respectively one half, one quarter and one eighth of the set length;
and the characteristic vector generation module is used for outputting the characteristic extraction result corresponding to each scale to the corresponding convolution layer and synthesizing the characteristic extraction results on adjacent scales to generate the characteristic vectors of the sample signal on three scales.
8. An apparatus, characterized in that the apparatus comprises:
one or more processors;
a memory for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement a method of analyzing a center beat of cardiac electrical signals as recited in any of claims 1-6.
9. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out a method for analyzing a center beat of an electrocardiogram signal as claimed in any one of claims 1 to 6.
CN202010085728.4A 2020-02-11 2020-02-11 Analysis method, device, equipment and storage medium for electrocardiosignal center beat Active CN111345814B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010085728.4A CN111345814B (en) 2020-02-11 2020-02-11 Analysis method, device, equipment and storage medium for electrocardiosignal center beat

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010085728.4A CN111345814B (en) 2020-02-11 2020-02-11 Analysis method, device, equipment and storage medium for electrocardiosignal center beat

Publications (2)

Publication Number Publication Date
CN111345814A CN111345814A (en) 2020-06-30
CN111345814B true CN111345814B (en) 2023-04-18

Family

ID=71192212

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010085728.4A Active CN111345814B (en) 2020-02-11 2020-02-11 Analysis method, device, equipment and storage medium for electrocardiosignal center beat

Country Status (1)

Country Link
CN (1) CN111345814B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114343659A (en) * 2022-01-05 2022-04-15 北京信息科技大学 Electrocardio characteristic signal extraction method, computer device and storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108932452A (en) * 2017-05-22 2018-12-04 中国科学院半导体研究所 Arrhythmia cordis beat classification method based on multiple dimensioned convolutional neural networks
CN110393519A (en) * 2019-08-19 2019-11-01 广州视源电子科技股份有限公司 Analysis method, device, storage medium and the processor of electrocardiosignal
CN110720894A (en) * 2019-11-08 2020-01-24 广州视源电子科技股份有限公司 Atrial flutter detection method, device, equipment and storage medium

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8755875B2 (en) * 2008-05-09 2014-06-17 Siemens Medical Solutions Usa, Inc. System for heart performance characterization and abnormality detection

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108932452A (en) * 2017-05-22 2018-12-04 中国科学院半导体研究所 Arrhythmia cordis beat classification method based on multiple dimensioned convolutional neural networks
CN110393519A (en) * 2019-08-19 2019-11-01 广州视源电子科技股份有限公司 Analysis method, device, storage medium and the processor of electrocardiosignal
CN110720894A (en) * 2019-11-08 2020-01-24 广州视源电子科技股份有限公司 Atrial flutter detection method, device, equipment and storage medium

Also Published As

Publication number Publication date
CN111345814A (en) 2020-06-30

Similar Documents

Publication Publication Date Title
CN106108889B (en) Electrocardiogram classification method based on deep learning algorithm
EP3313275B1 (en) Method and system using mathematical analysis and machine learning related to a disease
US11350868B2 (en) Electrocardiogram information processing method and electrocardiogram workstation system
US6024705A (en) Automated seismic detection of myocardial ischemia and related measurement of cardiac output parameters
CN107837082A (en) Electrocardiogram automatic analysis method and device based on artificial intelligence self study
US11783924B2 (en) ECG information processing method and ECG workstation
CN108511055B (en) Ventricular premature beat recognition system and method based on classifier fusion and diagnosis rules
US8992435B2 (en) System and method for classifying a heart sound
CN113095302B (en) Depth model for arrhythmia classification, method and device using same
WO2007130997A2 (en) Method and device for filtering, segmenting, compressing and classifying oscillatory signals
CN107252313A (en) The monitoring method and system of a kind of safe driving, automobile, readable storage medium storing program for executing
CN103970975A (en) Electrocardio data processing method and electrocardio data processing system
CN111839494A (en) Heart rate monitoring method and system
CN111345814B (en) Analysis method, device, equipment and storage medium for electrocardiosignal center beat
Hermawan et al. Development of ECG signal interpretation software on Android 2.2
KR102216047B1 (en) Method and apparatus for classifying cardiac arrhythmia using an auto associative neural network
CN111345815B (en) Method, device, equipment and storage medium for detecting QRS wave in electrocardiosignal
JP4718033B2 (en) ECG information processing apparatus and ECG information processing method
CN115553784B (en) Coronary heart disease assessment method and system based on electrocardio and heart sound signal coupling analysis
KR20180128634A (en) R-peak detection method and system using portable bio-information measuring terminal
CN112842355A (en) Electrocardiosignal heart beat detection and identification method based on deep learning target detection
US20220262516A1 (en) Atrial Fibrillation Prediction Model And Prediction System Thereof
CN114081502B (en) Non-invasive heart disease diagnosis method and device based on machine learning
Vondrak et al. Detection of myocardial infarction using analysis of vectorcardiographic loops
KR102578047B1 (en) A system for constructing an algorithm for predicting hyperkalemia through an electrocardiogram and a method for constructing an algorithm for predicting hyperkalemia through an electrocardiogram using the same

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant