CN115956925B - QRS wave detection method and system based on multistage smooth envelope - Google Patents
QRS wave detection method and system based on multistage smooth envelope Download PDFInfo
- Publication number
- CN115956925B CN115956925B CN202310093325.8A CN202310093325A CN115956925B CN 115956925 B CN115956925 B CN 115956925B CN 202310093325 A CN202310093325 A CN 202310093325A CN 115956925 B CN115956925 B CN 115956925B
- Authority
- CN
- China
- Prior art keywords
- ecg
- data
- sequence
- feature
- final detection
- 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
Links
Images
Classifications
-
- 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
- Image Analysis (AREA)
Abstract
A QRS wave detection method and system based on multistage smoothing envelope comprises the following steps: constructing a clustering model according to ECGpre data generated by preprocessing, carrying out fine filtering on ECG signals on a designated cluster of the clustering model, generating fine-filtered ECG data and recording the fine-filtered ECG data as ECG f Data; for ECG f Detecting data and performing feature conversion to generate a final detection feature sequence; for final detection of feature sequence ECG F Further sorting; the multi-level smooth envelope is calculated by the multi-level smooth envelope and recorded as ECG E The method comprises the steps of carrying out a first treatment on the surface of the Multi-level smoothing envelope ECG E Element of (a) and final detection feature sequence ECG F And comparing, extracting a position sequence of the candidate QRS complex, and determining the position of the QRS complex. The loss in the process of converting the characteristic curve can be effectively reduced; improving the accuracy of the QRS complex detection.
Description
Technical Field
The invention belongs to the technical field of electrocardiosignal acquisition, and particularly relates to a QRS wave detection method and system based on multistage smoothing envelope.
Background
With the development of the Internet and portable electrocardio acquisition equipment and the maturation of an electrocardio signal automatic analysis system, the cost required by all-weather monitoring of cardiovascular patients is greatly reduced, and the forward progress of human health business is promoted. The accurate identification of the QRS complex serves as the basis for automatic analysis of the electrocardiographic signals and plays an important role in the whole link.
Currently, the main methods include a differential thresholding method, a template matching method, a neural network method, a wavelet transform method, and the like. The principle of the differential threshold method is simpler, the calculation is quick, normal electrocardiosignals can be effectively detected, but the detection effect on signals containing noise interference is not ideal, in order to keep lower time complexity and improve the accuracy of QRS wave detection, the QRS wave group detection is realized based on the differential threshold method in the prior art, but the algorithms still have some defects: firstly, in the pretreatment process of signals, noise is not treated finely enough, and the characteristic curve loss is more, so that the detection accuracy is reduced; and secondly, when the threshold value is evaluated, unified parameters are used, QRS complex classifications with different forms and different times are not processed, and malformed QRS complex is easy to miss.
Disclosure of Invention
Aiming at the problems, the invention provides a QRS wave detection method and a QRS wave detection system based on multistage smooth envelopment, which can solve the problems that noise is not treated finely enough, characteristic curve loss is more, detection accuracy is reduced, unified parameters are used when a threshold value is evaluated, QRS wave groups in different forms and at different times are not treated, and malformed QRS wave groups are easy to miss.
An embodiment of the present invention provides a QRS wave detection method based on a multistage smoothing envelope, including:
constructing a clustering model according to ECGpre data generated by preprocessing, carrying out fine filtering on ECG signals on a designated cluster of the clustering model, generating fine-filtered ECG data and recording the fine-filtered ECG data as ECG f Data;
for ECG f Detecting the data and performing feature conversion to generate a final detection feature sequence, and recording the final detection feature sequence as an ECG F ；
For final detection of feature sequence ECG F Further sorting; wherein, for the finalDetecting feature sequence ECG F Sorting model training, extracting final detection feature sequence ECG F Generating a feature matrix from feature vectors in the data set;
for the sorted final detection characteristic sequence ECG F Performing multistage smoothing envelope calculation to obtain final detection characteristic sequence ECG F Is also denoted as ECG and is a multi-level smoothed envelope of (2) E ；
Multi-level smoothing envelope ECG E Element of (a) and final detection feature sequence ECG F And comparing, extracting a position sequence of the candidate QRS complex, and determining the position of the QRS complex.
Optionally, before the step of constructing the clustering model according to the ECGpre data generated by the preprocessing, the method further includes: acquiring ECG data; the ECG data is preprocessed and preprocessed ECG data is generated, and the preprocessed ECG data is denoted as ECGpre data.
Optionally, a cluster model is constructed according to ECGpre data generated by preprocessing, ECG signals on a designated cluster of the cluster model are subjected to fine filtering, and the fine filtered ECG data is generated and recorded as ECG f Data, comprising:
establishing a square distribution diagram for ECGpre data, selecting a maximum value and a minimum value proportionally, carrying out data normalization to obtain normalized data, and recording the normalized data as ECG HN1 。
Optionally, the ECG is extracted HN1 And data characteristics of ECGpre; wherein the data characteristics include zero-crossing rate, kurtosis, skewness, and shannon entropy values of the sequence.
Optionally, the method further comprises:
extracting fine filtering clustering feature vectors of the test data set, determining the optimal clustering quantity through an Elbow method, and marking a clustering matrix as an ECG (electronic organ health System) FEA-F’ ；
Constructing a clustering model, acquiring an ECG data set in the clustering model, determining the window width of Wiener filtering on each cluster in the ECG data set, and simultaneously determining parameters of a Butterworth filter;
fine filtering of clustered data specified by an ECG dataset to generate ECG f Data.
Alternatively to this, the method may comprise,for ECG f Detecting the data and performing feature conversion to generate a final detection feature sequence, and recording the final detection feature sequence as an ECG F Comprising:
computing ECG f The first-order difference and the absolute value of the data are respectively marked as ECGfd and ECGfad;
windowing and smoothing the ECGfad to generate a sequence after windowing and smoothing, and marking the sequence as ECGfs;
the windowed smoothed sequence ECGfs is filtered by a Chebyshev-II filter, and the filtered sequence is recorded as an ECG F’ ；
For filtered serial ECG F’ Establishing a square distribution diagram, selecting a maximum value and a minimum value according to a proportion, carrying out data normalization to obtain normalized data, and recording the normalized data as ECG F 。
Optionally, for the final detected feature sequence ECG F Sorting model training, extracting final detection feature sequence ECG F Generates a feature matrix comprising:
separate extraction of sequential ECG F’ And final detection of feature sequence ECG F A feature vector, wherein the feature vector comprises a zero-crossing rate, kurtosis, skewness and shannon entropy values of the sequence;
training of final detection feature sequence ECG F Sorting model, extracting final detection characteristic sequence ECG F Feature vectors in the data set of (2) generate feature matrixes and cluster through a K-Means clustering algorithm to generate a clustering model for sorting.
Optionally, for the sorted final detected feature sequence ECG F Performing multistage smoothing envelope calculation to obtain final detection characteristic sequence ECG F Is also denoted as ECG and is a multi-level smoothed envelope of (2) E Comprising:
determining a final detection feature sequence ECG F The window width of the multiple windows on the corresponding cluster after sorting is smooth and the corresponding punishment coefficient is adopted; determining cluster envelope parameters; analysis of envelope calculation to calculate a multistage smoothed envelope ECG E 。
Optionally, the multi-level smoothing envelope is includedElement and final detection feature sequence ECG F Comparing, extracting a sequence of locations of the candidate QRS complex, determining QRS complex locations, including:
multi-level smoothing envelope ECG E Element of (a) and final detection feature sequence ECG F The same positions of the elements in the two layers are compared;
recording a multi-level smoothed envelope ECG E Medium to larger than final detected signature sequence ECG F A candidate QRS complex location sequence is generated.
Optionally, the method further comprises:
dividing the candidate QRS complex position sequence into a plurality of continuous fragments, and taking the position corresponding to the maximum value in the characteristic curve in the fragments at different positions as the QRS complex position;
and carrying out merging correction on the QRS complex position according to the RR refractory period parameter, and determining the final position of the QRS complex.
Based on the same inventive concept, another aspect of the embodiments of the present invention further provides a QRS wave detection system based on a multistage smoothing envelope, including:
a fine filtering module for constructing a cluster model according to ECGpre data generated by preprocessing, carrying out fine filtering on the ECG signals on the designated cluster of the cluster model, generating fine filtered ECG data and recording the fine filtered ECG data as ECG f Data;
feature conversion module for ECG f Detecting the data and performing feature conversion to generate a final detection feature sequence, and recording the final detection feature sequence as an ECG F ；
Sorting module for ECG of final detected feature sequence F Further sorting; wherein for the final detected feature sequence ECG F Further sorting, comprising: for final detection of feature sequence ECG F Sorting model training, extracting final detection feature sequence ECG F Generating a feature matrix from feature vectors in the data set;
an envelope calculation module for sorting the final detection characteristic sequence ECG F Performing multistage smoothing envelope calculation to obtain final detection characteristic sequence ECG F Is also denoted as ECG and is a multi-level smoothed envelope of (2) E ；
A position determining module for smoothing the multi-level envelope ECG E Element of (a) and final detection feature sequence ECG F And comparing, extracting a position sequence of the candidate QRS complex, and determining the position of the QRS complex.
The above-mentioned one or at least one technical solution in the embodiments of the present application has at least the following technical effects:
determining the window width of Wiener filtering on each cluster in the ECG data set and simultaneously determining the parameters of the Butterworth filter; the data on the clusters designated by the ECG data set is subjected to fine filtering, so that loss in the process of converting the characteristic curve is effectively reduced, and the QRS complex detection is facilitated.
By sorting the feature sequences more finely, different types of sequences can be separated, for example: and carrying out envelope calculation on the data of each type of sequence respectively by using the ECG signal containing the deformed QRS wave, thereby improving the accuracy of QRS wave group detection.
By using the multistage smoothing envelope for detection, the characteristic change of the QRS complex in different time widths can be fully considered, and the detection accuracy of the QRS complex is further improved.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention may be realized and attained by the structure particularly pointed out in the written description and drawings.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a QRS wave detection method based on a multistage smoothing envelope;
fig. 2 is a schematic diagram of a QRS wave detection system based on a multistage smoothing envelope;
FIG. 3 is a schematic diagram of a data preprocessing flow in an embodiment of the present invention;
FIG. 4 is a schematic flow chart of a fine filtering clustering feature in an embodiment of the invention;
FIG. 5 is a schematic diagram of a fine filtering clustering model generation flow in an embodiment of the present invention;
FIG. 6 is a schematic diagram of a fine filtering process according to an embodiment of the present invention;
FIG. 7 is a schematic diagram of a detection feature conversion process according to an embodiment of the present invention;
FIG. 8 is a schematic diagram of a feature sequence sorting flow in an embodiment of the invention;
FIG. 9 is a schematic diagram of a multi-level smoothing envelope calculation flow in an embodiment of the invention;
fig. 10 is a schematic diagram of QRS complex extraction flow in an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, an embodiment of the present application provides a QRS wave detection method based on a multistage smoothing envelope, including:
s101, constructing a clustering model according to ECGpre data generated by preprocessing, carrying out fine filtering on ECG signals on a designated cluster of the clustering model, generating fine-filtered ECG data and recording the fine-filtered ECG data as ECG f Data;
s102 for ECG f Detecting the data and performing feature conversion to generate a final detection feature sequence, and recording the final detection feature sequence as an ECG F ；
S103, final inspectionCharacterization sequence ECG F Further sorting; wherein for the final detected feature sequence ECG F Further sorting, comprising: for final detection of feature sequence ECG F Sorting model training, extracting final detection feature sequence ECG F Generating a feature matrix from feature vectors in the data set;
s104, sorting the final detection characteristic sequence ECG F Performing multistage smoothing envelope calculation to obtain final detection characteristic sequence ECG F Is also denoted as ECG and is a multi-level smoothed envelope of (2) E ；
S105, smoothing the envelope ECG in multiple stages E Element of (a) and final detection feature sequence ECG F And comparing, extracting a position sequence of the candidate QRS complex, and determining the position of the QRS complex.
The detection method in the embodiment of the present disclosure is applied to an electrocardiographic detection device, which may be a medical electronic device or a hardware device that uses the detection method. For example, the device may be a computer, an electrocardiograph, or other electronic devices.
Wherein, referring to fig. 3, before step S101, further includes: acquiring ECG data; the ECG data is preprocessed and preprocessed ECG data is generated, and the preprocessed ECG data is denoted as ECGpre data.
Specifically, ECG data 301 of a fixed duration is acquired, and the fixed duration may be set to 10s to 15 to resample 302 the fixed duration ECG data to a fixed frequency, and may be resampled to a fixed frequency of 125 hz, and then passed through a notch filter to remove power frequency interference 303. The data is further filtered 304 by selecting a bandpass filter with a passband of 0.5 to 45 hz, and the data generated after the further filtering is recorded as ECGpre data by selecting a 4 th order Butterworth filter.
Referring to FIG. 4, a histogram is created for ECGpre data, and the maximum and minimum values are selected proportionally for data normalization to obtain normalized data, which is denoted as ECG HN1 ；
Specifically, the orthometric normalization 401 is a method of normalizing data according to a orthometric distribution diagram of the data, selecting values at a certain ratio of the beginning to the end in the histogram as the minimum and the maximum of the data normalization, and then substituting the values into a maximum and minimum normalization formula.
Specifically, first, a histogram of a fixed number of groups (e.g., 100) is generated using the preprocessed data, and then values corresponding to groups of a certain ratio (e.g., 0.02 and 0.01) before and after are obtained as minimum and maximum values, respectively designated as MINs HN1 And MAX HN1 . Then through the maximum and minimum normalization formula
Normalization is performed, wherein the ECG HN1 The first orthonormal ECG data is represented.
Extracting ECG HN1 And ECGpre data features; wherein the characteristics include zero-crossing rate, kurtosis, skewness, and shannon entropy values of the sequence. For example: and calculating the characteristics of zero crossing rate, kurtosis, skewness, shannon entropy and the like of the sequences when the threshold values are 0.05, 0.01 and 0.001 respectively. The threshold zero crossing rate is further set on the basis of solving the zero crossing rate, and is recorded as a zero crossing point when the threshold is met, so that the influence of small fluctuation on zero crossing rate calculation can be eliminated to a certain extent. The feature vectors 402 extracted from the note data are:
wherein Fzcr1, fzcr2 and Fzcr3 respectively represent zero-crossing rates of ECGpre when the threshold values are 0.05, 0.01 and 0.001; fzcr4, fzcr5, fzcr6, respectively represent ECG HN1 Zero-crossing rate at threshold values of 0.05, 0.01, 0.001; fkur1, fkur2 represent ECGpre and ECG, respectively HN1 Kurtosis value of (a); fske1 and Fske2 represent ECGpre and ECG, respectively HN1 Is a bias value of (1); fsha1, fsha2 represent and ECGpre and ECG, respectively HN1 Is a shannon entropy value.
Specifically, referring to fig. 5 and 6, data preprocessing and fine filtering cluster feature extraction 501 operations are sequentially performed on each piece of data of a data set, a feature matrix for clustering is obtained by extracting fine filtering cluster feature vectors of a test data set, determining the optimal number of clusters by an Elbow method,
the clustering matrix is noted as ECG FEA-F’ The method comprises the steps of carrying out a first treatment on the surface of the Wherein the ECG FEA-Fq Representing a fine filtering clustering feature vector corresponding to the q-th data in the data set; q represents the total number of data sets.
Constructing a clustering model, acquiring an ECG data set in the clustering model, determining window width 601 of Wiener filtering on each cluster in the ECG data set, and simultaneously determining parameters 602 of a Butterworth filter;
fine filtering 603 of the clustered data specified by the ECG dataset to generate ECG f Data.
Specifically, the Elbow method is first used to determine the optimal number of clusters; clustering is then performed using a K-Means clustering algorithm, resulting in a cluster model for fine filtering and saving 502. In order to reduce the damage to the effective features of the QRS complex while suppressing noise, mainly myoelectric interference, as much as possible, a certain number of ECG data sets are selected near the cluster center of the fine filtering cluster model, and more fine setting is required, a part of data in the test data set can be selected near each cluster center, and the width of the Wiener filter window corresponding to each cluster is set.
Specifically, the window width of Wiener filtering is determined on each cluster, for example: 3. 5, 7, 11, etc., reduces the impairment of QRS complex features in the ECG signal while suppressing noise; the parameters of the Butterworth filter are then determined on each cluster, with the Butterworth filter parameters being determined to be 4 th order and 0.5 to 45 hz.
In some alternative embodiments, the ECG is targeted f Detecting the data and performing feature conversion to generate a final detection feature sequence, and recording the final detection feature sequence as an ECG F Comprising:
referring to FIG. 7, the ECG is calculated f First order difference of dataAnd an absolute value 701, the first order difference and the absolute value are respectively referred to as ECGfd and ECGfad;
windowing smoothing 702 is performed on ECGfad to generate a sequence after windowing smoothing, which is marked as ECGfs;
the windowed smoothed sequence ECGfs is filtered 703 by a Chebyshev-II filter, and the filtered sequence is recorded as an ECG F’ 。
Specifically, windowed smoothing is an operation of convolving a signal with a particular window; for example, we select a Boxzen window with a window width of 12 as a convolution check to perform convolution operation on ECGfad, so as to obtain a sequence after windowing and smoothing, which is marked as ECGfs. After the windowing smoothing is finished, a plurality of high T wave influence detection results exist, a 2-35 Hz bandwidth Chebyshev-II type filter is selected for noise suppression, and the filtered sequence is ECG F’ . For filtered serial ECG F’ Establishing a square distribution diagram, selecting a maximum value and a minimum value according to a proportion, carrying out data normalization to obtain normalized data, and recording the normalized data as ECG F 。
In some alternative embodiments, to improve the accuracy of the detection algorithm, the detected feature sequence obtained in the previous step needs to be further sorted, after the feature sequence obtained by conversion, when the abnormal QRS complex or the high T wave is included, there are some smaller protrusions in the feature sequence, and if uniform detection parameters are used, there is a larger detection error rate on the type of data, wherein, for the final detected feature sequence, ECG F Sorting model training, extracting final detection feature sequence ECG F Generates a feature matrix comprising:
referring to fig. 8, the sequential ECG is extracted separately F’ And final detection of feature sequence ECG F A feature vector 801, wherein the feature vector comprises zero-crossing rate at a threshold of 0.02, 0.01, 0.005, kurtosis, skewness, and shannon entropy values of the sequence; the feature vectors extracted from the note data are as follows:
wherein, the liquid crystal display device comprises a liquid crystal display device,
、/>
、/>
respectively indicate->
Zero-crossing rate at threshold values of 0.02, 0.01, 0.005; />
、/>
、/>
Respectively indicate->
Zero-crossing rate at threshold values of 0.02, 0.01, 0.005;
、/>
respectively indicate->
And->
Kurtosis value of (a); />
、/>
Respectively indicate->
And
is a bias value of (1); />
、/>
Respectively indicate->
And->
Is a shannon entropy value.
Specifically, the feature vectors 802 of the sorting feature sequences of the dataset are extracted, and feature matrices for sorting clustering are obtained, which are marked as:
wherein the ECG FEA-Sp Representing sorting clustering feature vectors corresponding to the p-th data in the data set; p represents the total number of datasets.
Training of final detection feature sequence ECG F Sorting model, extracting final detection characteristic sequence ECG F Feature vectors in the data set of (2) generate feature matrixes and cluster through a K-Means clustering algorithm to generate a clustering model for sorting.
In some alternative embodiments, the characteristic sequence ECG is detected for the final detection F Performing multistage smoothing envelope calculation to obtain final detection characteristic sequence ECG F Is also denoted as ECG and is a multi-level smoothed envelope of (2) E Comprising:
referring to fig. 9, the final detection feature sequence ECG is determined F The window width of the multiple windows on the corresponding cluster after sorting is smooth and the corresponding punishment coefficient is adopted; cluster envelope parameter determination 901; analysis of envelope calculation to calculate a multistage smoothed envelope ECG E 。
Specifically, the final detection feature sequence ECG needs to be determined first F Sorting the corresponding clustersThe window width of the multi-window smoothing and the corresponding penalty coefficient; and then after the sorting clusters of the data to be detected are determined, reading envelope calculation parameters of the corresponding clusters, and superposing a multi-window smooth curve to obtain a detection envelope.
Specifically, for example: is provided with one-dimensional sequence
Comprises->
The individual elements are->
A smooth window with a radius of
Penalty coefficients corresponding to each window are +.>
。
The first step is to calculate the smooth curve corresponding to each window, and the formula is as follows:
wherein, the liquid crystal display device comprises a liquid crystal display device,
indicate->
Smooth window->
；/>
Indicate->
Smooth radius of the smooth window +.>
Indicate->
Penalty coefficients for the individual smoothing windows; />
Representing the acquisition sequence->
Is>
To->
Element fragments of (2); />
Representing the calculation->
Is the average value of (2); when->
And->
Less than 1 or greater than->
At this time, filling is performed using 0.
And step two, superposing the smooth curves to obtain a multistage smooth envelope, wherein the formula is as follows:
wherein, the liquid crystal display device comprises a liquid crystal display device,
indicate->
An element; />
Indicate->
Line smooth curve->
The elements.
In some alternative embodiments, referring to FIG. 10, elements in the multi-level smoothing envelope are combined with the final detected feature sequence ECG F In contrast, a sequence of locations 1001 of candidate QRS complexes is extracted, determining QRS complex locations, including:
multi-level smoothing envelope ECG E Element of (a) and final detection feature sequence ECG F The same positions of the elements in the two layers are compared;
recording a multi-level smoothed envelope ECG E Medium to larger than final detected signature sequence ECG F A candidate QRS complex location sequence is generated.
In particular, in obtaining a multi-level smoothed envelope
Then, the elements are combined with the final detection characteristic sequence
The elements at the same position are compared one by one and +.>
Middle is greater than->
Obtaining a candidate QRS complex position sequence, denoted +.>
。
Dividing the candidate QRS complex position sequence into a plurality of continuous segments 1002, and taking the position corresponding to the maximum value in the characteristic curve in the segments at different positions as the QRS complex position;
the QRS complex location is combined and corrected according to the RR refractory period parameters to determine the QRS complex final location 1003.
Will be
The continuous positions in the process are regulated and divided into a plurality of continuous fragments, and the fragments are marked as:
wherein, the liquid crystal display device comprises a liquid crystal display device,
indicate->
QRS complex location segments of segment successive locations; />
Representing the total number of consecutive position segments.
Traversing
In each of which is obtained +.>
The position of the maximum at the corresponding position is taken as the position of the QRS complex and is marked as +.>
。
Further filtering of the detected QRS complex locations,
there may be a small number of adjacent QRS complex locations that do not correspond to the basic rhythm, screening for portions where some adjacent QRS complex locations are close, which may be portions where adjacent locations are less than 0.15s in time, and selecting the QRS complex location where the value is greatest as the post-merger QRS complex location, determining the QRS complex final location 1004.
This application disclosesAnother aspect of the embodiments relates to a QRS wave detection system based on a multi-level smoothing envelope, see fig. 2, comprising: a fine filtering module for constructing a cluster model according to ECGpre data generated by preprocessing, carrying out fine filtering on the ECG signals on the designated cluster of the cluster model, generating fine filtered ECG data and recording the fine filtered ECG data as ECG f Data;
feature conversion module for ECG f Detecting the data and performing feature conversion to generate a final detection feature sequence, and recording the final detection feature sequence as an ECG F ；
Sorting module for ECG of final detected feature sequence F Further sorting; wherein for the final detected feature sequence ECG F Further sorting, comprising: for final detection of feature sequence ECG F Sorting model training, extracting final detection feature sequence ECG F Generating a feature matrix from feature vectors in the data set;
an envelope calculation module for sorting the final detection characteristic sequence ECG F Performing multistage smoothing envelope calculation to obtain final detection characteristic sequence ECG F Is also denoted as ECG and is a multi-level smoothed envelope of (2) E ；
A position determining module for smoothing the multi-level envelope ECG E Element of (a) and final detection feature sequence ECG F And comparing, extracting a position sequence of the candidate QRS complex, and determining the position of the QRS complex.
Although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (11)
1. A QRS wave detection method based on a multistage smoothing envelope, the method comprising: constructing a clustering model according to ECGpre data generated by preprocessing, and carrying out ECG signals on a designated cluster of the clustering modelFine filtering to generate and record fine filtered ECG data as ECG f Data;
for ECG f Detecting the data and performing feature conversion to generate a final detection feature sequence, and recording the final detection feature sequence as an ECG F ；
For final detection of feature sequence ECG F Further sorting; wherein for the final detected feature sequence ECG F Sorting model training, extracting final detection feature sequence ECG F Generating a feature matrix from feature vectors in the data set;
for the sorted final detection characteristic sequence ECG F Performing multistage smoothing envelope calculation to obtain final detection characteristic sequence ECG F Is also denoted as ECG and is a multi-level smoothed envelope of (2) E ；
Multi-level smoothing envelope ECG E Element of (a) and final detection feature sequence ECG F And comparing, extracting a position sequence of the candidate QRS complex, and determining the position of the QRS complex.
2. The method according to claim 1, wherein before the step of constructing a cluster model from the ECGpre data generated by the preprocessing, further comprises: acquiring ECG data; the ECG data is preprocessed and preprocessed ECG data is generated, and the preprocessed ECG data is denoted as ECGpre data.
3. The method according to any one of claims 1 or 2, wherein a cluster model is constructed from the ECGpre data generated by the preprocessing, the ECG signals on a designated cluster of the cluster model are subjected to fine filtering, and the fine filtered ECG data is generated and recorded as ECG f Data, comprising:
establishing a square distribution diagram for ECGpre data, selecting a maximum value and a minimum value proportionally, carrying out data normalization to obtain normalized data, and recording the normalized data as ECG HN1 。
4. The method of claim 3, wherein the ECG is extracted HN1 And data characteristics of ECGpre; wherein the data featuresIncluding zero-crossing rate, kurtosis, skewness, and shannon entropy values of the sequence.
5. A detection method according to claim 3, further comprising:
extracting fine filtering clustering feature vectors of the test data set, determining the optimal clustering quantity through an Elbow method, and marking a clustering matrix as an ECG (electronic organ health System) FEA-F’ ；
Constructing a clustering model, acquiring an ECG data set in the clustering model, determining the window width of Wiener filtering on each cluster in the ECG data set, and simultaneously determining parameters of a Butterworth filter;
fine filtering of clustered data specified by an ECG dataset to generate ECG f Data.
6. The method of claim 1, wherein the method is directed to ECG f Detecting the data and performing feature conversion to generate a final detection feature sequence, and recording the final detection feature sequence as an ECG F Comprising:
computing ECG f The first-order difference and the absolute value of the data are respectively marked as ECGfd and ECGfad;
windowing and smoothing the ECGfad to generate a sequence after windowing and smoothing, and marking the sequence as ECGfs;
the windowed smoothed sequence ECGfs is filtered by a Chebyshev-II filter, and the filtered sequence is recorded as an ECG F’ ；
For filtered serial ECG F’ Establishing a square distribution diagram, selecting a maximum value and a minimum value according to a proportion, carrying out data normalization to obtain normalized data, and recording the normalized data as ECG F 。
7. The method of claim 1, wherein the final detected signature sequence ECG is F Sorting model training, extracting final detection feature sequence ECG F Generates a feature matrix comprising:
separate extraction of sequential ECG F’ And final detectionFeature sequence ECG F A feature vector, wherein the feature vector comprises a zero-crossing rate, kurtosis, skewness and shannon entropy values of the sequence;
training of final detection feature sequence ECG F Sorting model, extracting final detection characteristic sequence ECG F Feature vectors in the data set of (2) generate feature matrixes and cluster through a K-Means clustering algorithm to generate a clustering model for sorting.
8. The method of claim 1, wherein the final detected signature sequence ECG is F Performing multistage smoothing envelope calculation to obtain final detection characteristic sequence ECG F Is also denoted as ECG and is a multi-level smoothed envelope of (2) E Comprising:
determining a final detection feature sequence ECG F The window width of the multiple windows on the corresponding cluster after sorting is smooth and the corresponding punishment coefficient is adopted; determining cluster envelope parameters; analysis of envelope calculation to calculate a multistage smoothed envelope ECG E 。
9. The method of claim 1, wherein elements in the multi-level smoothed envelope are combined with the final detected signature sequence ECG F Comparing, extracting a sequence of locations of the candidate QRS complex, determining QRS complex locations, including:
multi-level smoothing envelope ECG E Element of (a) and final detection feature sequence ECG F The same positions of the elements in the two layers are compared;
recording a multi-level smoothed envelope ECG E Medium to larger than final detected signature sequence ECG F A candidate QRS complex location sequence is generated.
10. The method of detecting according to claim 9, characterized in that the method further comprises:
dividing the candidate QRS complex position sequence into a plurality of continuous fragments, and taking the position corresponding to the maximum value in the characteristic curve in the fragments at different positions as the QRS complex position;
and carrying out merging correction on the QRS complex position according to the RR refractory period parameter, and determining the final position of the QRS complex.
11. A QRS wave detection system based on a multistage smoothing envelope, comprising:
a fine filtering module for constructing a cluster model according to ECGpre data generated by preprocessing, carrying out fine filtering on the ECG signals on the designated cluster of the cluster model, generating fine filtered ECG data and recording the fine filtered ECG data as ECG f Data;
feature conversion module for ECG f Detecting the data and performing feature conversion to generate a final detection feature sequence, and recording the final detection feature sequence as an ECG F ；
Sorting module for ECG of final detected feature sequence F Further sorting; wherein for the final detected feature sequence ECG F Further sorting, comprising: for final detection of feature sequence ECG F Sorting model training, extracting final detection feature sequence ECG F Generating a feature matrix from feature vectors in the data set;
an envelope calculation module for sorting the final detection characteristic sequence ECG F Performing multistage smoothing envelope calculation to obtain final detection characteristic sequence ECG F Is also denoted as ECG and is a multi-level smoothed envelope of (2) E ；
A position determining module for smoothing the multi-level envelope ECG E Element of (a) and final detection feature sequence ECG F And comparing, extracting a position sequence of the candidate QRS complex, and determining the position of the QRS complex.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310093325.8A CN115956925B (en) | 2023-02-10 | 2023-02-10 | QRS wave detection method and system based on multistage smooth envelope |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310093325.8A CN115956925B (en) | 2023-02-10 | 2023-02-10 | QRS wave detection method and system based on multistage smooth envelope |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115956925A CN115956925A (en) | 2023-04-14 |
| CN115956925B true CN115956925B (en) | 2023-06-06 |
Family
ID=85894573
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310093325.8A Active CN115956925B (en) | 2023-02-10 | 2023-02-10 | QRS wave detection method and system based on multistage smooth envelope |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115956925B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4838278A (en) * | 1987-02-26 | 1989-06-13 | Hewlett-Packard Company | Paced QRS complex classifier |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8954140B2 (en) * | 2012-09-27 | 2015-02-10 | Samsung Electronics Co., Ltd. | Method and system for determining QRS complexes in electrocardiogram signals |
| US9241650B2 (en) * | 2013-12-15 | 2016-01-26 | Lifewatch Technologies Ltd. | Patient specific QRS complex classification for arrhythmia detection |
| CN110013247B (en) * | 2019-05-24 | 2020-06-05 | 东北大学 | Detection, differentiation and positioning method for electrocardiogram P wave and T wave |
| CN113229826B (en) * | 2020-11-25 | 2022-07-26 | 联想(北京)有限公司 | QRS wave detection method and device and electronic equipment |
| US20220202344A1 (en) * | 2020-12-29 | 2022-06-30 | Draegerwerk Ag & Co. Kgaa | Apparatus and Method for Electrocardiogram (ECG) Signal Analysis and Heart Block Detection |
| CN114176602B (en) * | 2021-12-30 | 2024-04-26 | 重庆康如来科技有限公司 | Method for simultaneously positioning electrocardiograph P wave, QRS wave and T wave based on deep learning multi-target detection |
-
2023
- 2023-02-10 CN CN202310093325.8A patent/CN115956925B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4838278A (en) * | 1987-02-26 | 1989-06-13 | Hewlett-Packard Company | Paced QRS complex classifier |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115956925A (en) | 2023-04-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109948647B (en) | Electrocardiogram classification method and system based on depth residual error network | |
| CN106473750B (en) | Personal identification method based on photoplethysmographic optimal period waveform | |
| CN114266276B (en) | Motor imagery electroencephalogram signal classification method based on channel attention and multi-scale time domain convolution | |
| CN109124620B (en) | Atrial fibrillation detection method, device and equipment | |
| CN114366124B (en) | Epileptic electroencephalogram identification method based on semi-supervised deep convolution channel attention list classification network | |
| CN109948396B (en) | Heart beat classification method, heart beat classification device and electronic equipment | |
| CN114469124B (en) | Method for identifying abnormal electrocardiosignals in movement process | |
| CN113768519B (en) | Method for analyzing consciousness level of patient based on deep learning and resting state electroencephalogram data | |
| US20230301581A1 (en) | Fusion signal processing for maternal uterine activity detection | |
| CN111685774B (en) | OSAHS Diagnosis Method Based on Probability Integrated Regression Model | |
| CN111317499A (en) | Heart sound signal processing method based on wavelet technology | |
| CN112949639A (en) | Congenital heart disease heart sound intelligent segmentation and classification algorithm, device and storage medium | |
| CN115363586A (en) | Psychological stress grade assessment system and method based on pulse wave signals | |
| CN113796889A (en) | Auxiliary electronic stethoscope signal discrimination method based on deep learning | |
| CN113436203A (en) | Automatic ventricular membrane segmentation method | |
| CN110811591A (en) | Heart failure grading method based on heart rate variability | |
| CN115956925B (en) | QRS wave detection method and system based on multistage smooth envelope | |
| CN115429284A (en) | Electrocardiosignal classification method, electrocardiosignal classification system, computer equipment and readable storage medium | |
| CN115281676A (en) | Fatigue detection method based on GRU neural network and ECG signal | |
| CN115374815A (en) | Automatic sleep staging method based on visual Transformer | |
| CN114757236A (en) | Electroencephalogram signal denoising optimization method and system based on TQWT and SVMD | |
| CN112464711A (en) | MFDC-based electroencephalogram identity identification method, storage medium and identification device | |
| CN112270359A (en) | One-dimensional sequence ascending clustering method and system | |
| CN114403831B (en) | PPG waveform pulse extraction method and device | |
| CN116091489B (en) | Electrocardiogram image recognition method and system based on deep learning |
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 |