CN111493858A - Single-lead specific main wave identification and positioning method based on cluster analysis - Google Patents
Single-lead specific main wave identification and positioning method based on cluster analysis Download PDFInfo
- Publication number
- CN111493858A CN111493858A CN202010180896.1A CN202010180896A CN111493858A CN 111493858 A CN111493858 A CN 111493858A CN 202010180896 A CN202010180896 A CN 202010180896A CN 111493858 A CN111493858 A CN 111493858A
- Authority
- CN
- China
- Prior art keywords
- main wave
- specific
- interval
- wave
- class
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000007621 cluster analysis Methods 0.000 title claims abstract description 25
- 238000004458 analytical method Methods 0.000 claims abstract description 62
- 238000011156 evaluation Methods 0.000 claims description 37
- 230000000877 morphologic effect Effects 0.000 claims description 22
- 208000000418 Premature Cardiac Complexes Diseases 0.000 claims description 12
- 239000013589 supplement Substances 0.000 claims description 11
- 230000016507 interphase Effects 0.000 claims description 9
- 230000036279 refractory period Effects 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000006073 displacement reaction Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 230000001502 supplementing effect Effects 0.000 claims description 3
- 238000002203 pretreatment Methods 0.000 claims description 2
- 238000003745 diagnosis Methods 0.000 abstract description 5
- 206010015856 Extrasystoles Diseases 0.000 description 6
- 206010003119 arrhythmia Diseases 0.000 description 6
- 230000006793 arrhythmia Effects 0.000 description 6
- 230000001746 atrial effect Effects 0.000 description 4
- 230000002861 ventricular Effects 0.000 description 4
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 230000036244 malformation Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 208000024891 symptom Diseases 0.000 description 1
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]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
- A61B5/024—Detecting, measuring or recording pulse rate or heart rate
- A61B5/02405—Determining heart rate variability
-
- 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
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B2560/00—Constructional details of operational features of apparatus; Accessories for medical measuring apparatus
- A61B2560/02—Operational features
- A61B2560/0223—Operational features of calibration, e.g. protocols for calibrating sensors
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B2576/00—Medical imaging apparatus involving image processing or analysis
- A61B2576/02—Medical imaging apparatus involving image processing or analysis specially adapted for a particular organ or body part
- A61B2576/023—Medical imaging apparatus involving image processing or analysis specially adapted for a particular organ or body part for the heart
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Cardiology (AREA)
- Physics & Mathematics (AREA)
- Animal Behavior & Ethology (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Pathology (AREA)
- Veterinary Medicine (AREA)
- Physiology (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
A single-lead specific main wave identification and positioning method based on cluster analysis comprises the following steps: a) pre-treating; b) based on the selected characteristics, performing self-adaptive cluster analysis on the main wave; c) and comprehensively analyzing to determine the type and specific position of the specific main wave. The invention determines the type and the position of the single-lead specific main wave by applying a clustering analysis tool in multiple stages and layers, thereby providing reference for further diagnosis of doctors. The method is an unsupervised learning method, does not need label data, can identify and position specific waveforms of individual electrocardiograms, and is easy to understand, easy to realize and high in precision.
Description
Technical Field
The invention provides a cluster analysis-based single-lead specific main wave identification and positioning method, and relates to the field of electrocardiogram intelligent diagnosis.
Background
Electrocardiographic examination is a common item of physical examination, and if a patient is suspected to have arrhythmia symptoms, the patient usually goes to a hospital to make an electrocardiogram, but the problem is difficult to find in the electrocardiogram of minutes or tens of minutes in the hospital, and at this time, a doctor may wear a wearable electrocardiogram detector for two weeks or more, which may generate an electrocardiogram of hundreds of hours, and the doctor takes a second and a second for examination, which is very time-consuming.
The study on the arrhythmia by using a machine learning method such as a support vector machine, KNN and the like in the early stage, the study on the arrhythmia by using a deep learning method is also proposed by researchers along with the rise of the deep learning in recent years, for example, the Wu Endada Bangtian machine learning group abroad proposes to use CNN to detect the arrhythmia at the cardiologist level, the domestic Gastrian and the like study on the multi-lead arrhythmia intelligent diagnosis method by using CNN, the study on the multi-lead arrhythmia intelligent diagnosis method by using two-dimensional CNN or L is also proposed, and related invention patents are applied.
Therefore, based on the latest research result, a single-lead specific dominant wave identification and positioning method based on cluster analysis is provided, and the type and the position of the lead specific dominant wave are determined by applying a cluster analysis tool in multiple stages and multiple levels, so that reference is provided for further diagnosis of doctors. The method is an unsupervised learning method, does not need label data, and can identify and position individual electrocardiograms, thereby better solving two problems of the conventional supervised learning method, being easy to understand and realize and having higher precision.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and invents a single-guide specific main wave identification and positioning method based on cluster analysis.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a single-lead specific main wave identification and positioning method based on cluster analysis comprises the following steps: a) pre-treating; b) based on the selected characteristics, performing self-adaptive cluster analysis on the main wave; c) and comprehensively analyzing to determine the type and specific position of the specific main wave. The pretreatment method of a) comprises the following steps:
step 1: obtaining the approximate position of each main wave on the single-lead data based on a conventional analysis method; the analysis method is a P-T calculation method or a QRS wave starting point and end point positioning method based on regularized least square regression learning;
step 2: removing baseline drift and high-frequency error of original data by using MODWT; the MODWT is a highly redundant non-orthogonal transform, the sample capacity can be an arbitrary value, the MODWT has displacement invariance, through comprehensive comparison and analysis, a sym4 wavelet is selected and analyzed to the layer 6, and d2, d3, d4, d5 and d6 are selected as reconstruction signals, so that baseline drift and high-frequency errors are removed;
and step 3: calibrating, namely determining the specific position and time limit of the existing main wave, and acquiring a candidate main wave position which is possibly missed, wherein the calibrating and determining the specific position and time limit of the existing main wave are divided into two steps:
(1) for all initial dominant wave positions: and replacing the original initial main wave position by an absolute value maximum point in a small interval near a certain initial main wave position, and obtaining the main wave main body direction according to all the main wave directions. For individual main wave directions inconsistent with the main direction, changing or keeping the original direction according to the amplitude sequencing of the main wave directions in the main direction and the amplitude condition of the main wave in the opposite direction, thereby determining the specific position of the main wave, and determining each main wave time limit based on the slope change condition and the QRS normal time limit range;
(2) outside the refractory period of the existing main wave position, finding out an absolute value maximum value point in a certain interval near each point as a candidate missing main wave position, and providing preparation for targeted supplement according to a clustering analysis result in a later period;
preferably, the method comprises the following steps: the b) self-adaptive clustering analysis method for the main wave based on the selected characteristics comprises the following steps:
step 1: based on the main wave morphological characteristics, the main wave is subjected to self-adaptive cluster analysis, possible missing candidate main waves are supplemented, and possible specific main wave positions are found out:
(1) obtaining waveform data in a certain range at two ends of each main wave point, subtracting the corresponding amplitude of the main wave point, and then taking an absolute value to obtain waveform data to be gathered;
(2) adopting a self-adaptive clustering method for waveform data to be clustered to obtain the classification number corresponding to the highest average evaluation index, and carrying out clustering analysis on the classification number to obtain the category evaluation index corresponding to each classification;
(3) determining whether to enter a shape clustering module according to the overall evaluation index, each classification evaluation index and each classification average time limit: when the average time limit is larger, the evaluation index is required to be larger; on the contrary, when the average time limit is smaller, the evaluation index requirement can be properly lowered, and the specific parameters are determined according to the actual analysis data;
(4) when the waveform can be divided, the missing main wave position of the supplement candidate is considered, firstly, the centers of various types and the minimum and maximum distances between each point in the types and the center are calculated; secondly, calculating the distance between each candidate main wave position and the center of each class, taking the class with the minimum distance as a candidate class, and inspecting whether the distance is within the maximum distance range in the class, wherein the distance is slightly larger but has small difference or the distance from one class is far smaller than the distance from the adjacent class, so that the candidate main wave position is classified into the class; finally, recording the remaining candidate main wave positions;
(5) classifying, namely determining two types of specific main wave categories, wherein the number of the categories is two, the average time limit corresponding to a certain category is the largest, and the given requirements of the wide and large malformed main wave time limits are met, and then taking the category as the specific main wave category; if the number of categories is more than two, sorting the various average time limits from large to small, and taking all categories of which the average time limits meet the requirements as specific main waves;
(6) individual calibrations are performed on the specific dominant wave in the forward direction. Performing cluster analysis on the forward main wave to determine whether the forward main wave can be divided into two parts; if the main wave has the forward specific main wave, determining whether to remove the main wave according to whether the forward main wave can be divided into two parts and the time limit of the individual main wave, and if the main wave does not have the forward specific main wave, determining whether to supplement the individual main wave according to whether the forward main wave can be divided into two parts and the time limit of the individual main wave;
(7) the waveform is not separable, the special condition that the negative main wave exists is considered, and the negative main wave meets the time limit requirement, and the main wave is determined as the specific main wave;
(8) if the specific main wave does not meet the requirement, the specific main wave based on morphological clustering analysis is empty;
step 2: based on the main wave interval characteristics, carrying out self-adaptive cluster analysis on the main wave interval, supplementing possible missing candidate main waves, and finding out possible specific main wave positions:
(1) acquiring first-order differential data of a main wave position, namely interval data;
(2) judging whether interval data are uniform or not according to interval range, if so, the interval is not separable, and if not, the following steps are continued;
(3) adopting a self-adaptive clustering method to obtain the classification number corresponding to the highest average evaluation index, and carrying out clustering analysis according to the classification number to obtain the evaluation index corresponding to each classification;
(4) if the average evaluation index is greater than a given value or the evaluation index in a certain class is greater than a given value, the interval can be divided;
(5) when the interval is available, the missing main wave position of the supplement candidate is considered, firstly, the centers of various classes and the minimum and maximum distances between each point in the classes and the center are calculated; secondly, calculating the distance between the front interval and the rear interval of each candidate main wave and the center of each class, taking the class with the minimum distance as a candidate class, and inspecting whether the distance is within the range of the maximum distance in the class, wherein the distance is slightly larger but has small difference or the distance from one class is far smaller than the distance from the adjacent class, and classifying the front interval and the rear interval of the position of the candidate main wave into the class; finally, recording the remaining candidate main wave positions;
(6) classifying, and determining the main wave class of the specific interval, taking the rapid premature beat as an example: the number of categories is two, the interval corresponding to a certain category is the minimum, and the given interval requirement is met, the category is taken as a specific interval category, and the interval positions plus 1 are taken as the corresponding specific main wave positions; if the number of categories is more than two, sorting the average intervals from small to large, taking the interval positions corresponding to all categories with the average interval meeting the requirements plus 1 as the specific main wave position, and the escape and conduction block types are just opposite;
(7) if the evaluation index is smaller than the given value, the specific main wave based on interval clustering analysis is empty;
preferably, the method comprises the following steps: the c) comprehensive analysis and the method for determining the specific main wave type and the specific position comprise the following steps: determining the type and position of a specific main wave based on the main wave morphology clustering analysis result and the main wave interval clustering result;
the result is as follows: when the morphological clustering analysis result and the interval clustering analysis result do not exist, the specific main wave is an empty set;
and a second result: when the morphological clustering analysis result and the interphase clustering analysis result exist one time, the corresponding specific main wave type and the specific position are selected;
and a third result: and when both the morphological clustering analysis result and the interphase clustering analysis result exist, taking the intersection of the morphological clustering analysis result and the interphase clustering analysis result as a final result.
Drawings
FIG. 1 is a flow chart of the implementation steps of a single-guide specific dominant wave identification and positioning method based on cluster analysis according to the present invention;
FIG. 2 is a diagram showing an example of the specific dominant wave positions identified by the electrocardiogram of "sporadic ventricular premature beats";
FIG. 3 is an example of the specific dominant wave positions identified by the electrocardiogram of "frequent ventricular premature beats";
FIG. 4 is a diagram showing an example of the specific dominant wave positions identified by electrocardiogram of "sporadic atrial premature beats";
FIG. 5 is an example of the specific dominant wave positions identified by the electrocardiogram of "frequent atrial premature beats".
Detailed Description
The technical scheme of the present invention will be further described in detail with reference to the accompanying drawings, and fig. 1 shows a method for identifying and positioning a single-waveguide specific main wave based on cluster analysis, which includes the following steps:
a) pretreatment:
step 1: obtaining the approximate position of each main wave on the single-lead data based on a conventional analysis method;
at present, the main wave location methods of electrocardiogram are very numerous, such as the classic P-T algorithm or the QRS wave starting point and end point location method based on regularized least squares regression learning (patent number: 201610369281.7) which is self-developed by the company
Step 2: removing baseline drift and high-frequency error of original data by using MODWT;
MODWT is a highly redundant non-orthogonal transform, with sample volumes of arbitrary value, with displacement invariance, well suited for processing electrocardiographic data. In actual use, by comprehensive comparison and analysis, a sym4 wavelet is selected, the level 6 is analyzed, and d2, d3, d4, d5 and d6 are selected as reconstruction signals, so that baseline drift and high-frequency errors are removed;
and step 3: calibrating, namely determining the specific position and time limit of the existing main wave and obtaining the candidate main wave position which is possibly omitted;
because of the inherent defects of single lead coupling, mutual calibration between leads cannot be performed like multi-lead coupling, and the primary wave initially positioned may have the phenomena of direction reversal (for example, the primary wave should take an R wave to become an S wave), omission and the like, which all affect the later-stage clustering analysis result. For this purpose, it is carried out in two steps:
(1) for all initial dominant wave positions: replacing an original initial main wave position with an absolute value maximum point in a small interval near a certain initial main wave position, obtaining main wave main body directions according to all main wave directions, changing or keeping the original direction of individual main wave directions inconsistent with the main body directions according to amplitude sequencing of the main wave directions and main wave amplitude conditions in the opposite direction, determining the specific position of the main wave, and determining each main wave time limit (note: distance between two peak valley points) based on the slope change condition and QRS normal time limit range;
(2) outside the refractory period of the existing main wave position, finding out the maximum value point of the absolute value in a certain interval near each point, using the maximum value point as a candidate missing main wave position, and providing preparation for targeted supplement according to a clustering analysis result in a later period;
b) based on the selected characteristics, performing adaptive clustering analysis on the main wave:
step 1: based on the main wave morphological characteristics, the main wave is subjected to self-adaptive cluster analysis, possible missing candidate main waves are supplemented, and possible specific main wave positions are found out:
(1) obtaining waveform data within a certain range at two ends of each main wave point (note: because the normal main wave range is about 0.06s-0.08s, according to the actual analysis experience, the interval is about 0.07 s), subtracting the corresponding amplitude of the main wave point, and then taking an absolute value (note: considering the wide and large malformed main wave in the positive and negative directions) to obtain the waveform data to be gathered;
(2) adopting a self-adaptive clustering method for waveform data to be clustered to obtain the classification number corresponding to the highest average evaluation index, and carrying out clustering analysis on the classification number to obtain the category evaluation index corresponding to each classification;
(3) determining whether to enter a shape clustering module according to the overall evaluation index, each classification evaluation index and each classification average time limit: when the average time limit is larger, the evaluation index is required to be larger; conversely, when the average time limit is smaller, the evaluation index requirement can be lowered appropriately. The specific parameters are determined according to actual analysis data;
(4) when the waveform can be divided, the missing main wave position of the supplement candidate is considered, firstly, the centers of various types and the minimum and maximum distances between each point in the types and the center are calculated; secondly, calculating the distance between each candidate main wave position and the center of each class, taking the class with the minimum distance as a candidate class, inspecting whether the distance is within the maximum distance range in the class, and if the distance is within the range or slightly larger but has small difference or the distance from one class is far smaller than the distance from the adjacent class, classifying the candidate main wave position into the class; finally, recording the remaining candidate main wave positions;
(5) and classifying and determining the specific main wave category. If the number of the categories is two, the average time limit corresponding to a certain category is the maximum, and the given requirement of the broad deformity main wave time limit is met, the category is taken as a specific main wave category; if the number of the categories is more than two, sorting the various average time limits from large to small, and taking all categories of which the average time limits meet the requirements as specific main waves;
(6) individual calibrations are performed on the specific dominant wave in the forward direction. And carrying out cluster analysis on the forward main wave to determine whether the forward main wave can be halved. If the main wave has the forward specificity, determining whether to remove the main wave according to whether the forward main wave can be divided into two parts and the time limit of the individual main wave, and if the main wave does not have the forward specificity, determining whether to supplement the individual main wave according to whether the forward main wave can be divided into two parts and the time limit of the individual main wave;
(7) if the waveform is inseparable, considering the special condition of having a negative main wave, if the negative main wave meets the time limit requirement, determining the main wave as a specific main wave;
(8) if the requirement is not met, the specific main wave based on the morphological clustering analysis is null;
step 2: based on the main wave interval characteristics, carrying out self-adaptive cluster analysis on the main wave interval, supplementing possible missing candidate main waves, and finding out possible specific main wave positions:
(1) acquiring first-order differential data of a main wave position, namely interval data;
(2) judging whether interval data are uniform or not according to interval range, if so, the interval is not separable, and if not, the following steps are continued;
(3) adopting a self-adaptive clustering method to obtain the classification number corresponding to the highest average evaluation index, and carrying out clustering analysis according to the classification number to obtain the evaluation index corresponding to each classification;
(4) if the average evaluation index is greater than a given value, such as 0.85, or the evaluation index within a class is greater than a given value, such as 0.98, the interval may be divided;
(5) when the interval is divisible, the missing dominant wave positions of the complementary candidates are considered. Firstly, calculating various centers and the minimum and maximum distances between each point in the various centers and the center; secondly, calculating the distance between the front interval and the rear interval of each candidate main wave and the center of each class, taking the class with the minimum distance as a candidate class, inspecting whether the distance is within the range of the maximum distance in the class, and if the distance is within the range or slightly larger but has small difference or the distance from one class is far smaller than the distance from the adjacent class, classifying the front interval and the rear interval of the position of the candidate main wave as the class; finally, recording the remaining candidate main wave positions;
(6) and classifying and determining the dominant wave class of the specific interval. Taking a rapid premature beat as an example: if the number of the categories is two, the interval corresponding to a certain category is the minimum and meets the given interval requirement, the category is taken as a specific interval category, and the interval positions plus 1 are taken as the corresponding specific main wave positions; if the number of the categories is more than two, sorting the average intervals of the categories from small to large, and adding 1 to interval positions corresponding to all categories of which the average intervals meet the requirements to obtain the specific main wave position. Escape and conduction block types are just opposite;
(7) if the evaluation index is smaller than the given value, the specific main wave based on interval clustering analysis is empty;
c) comprehensively analyzing, and determining the type and specific position of the specific main wave: determining the type and position of a specific main wave based on the main wave morphology clustering analysis result and the main wave interval clustering result;
the result is as follows: when the morphological clustering analysis result and the interval clustering analysis result do not exist, the specific main wave is an empty set;
and a second result: when the morphological clustering analysis result and the interphase clustering analysis result exist one time, the corresponding specific main wave type and the specific position are selected;
and a third result: and when both the morphological clustering analysis result and the interphase clustering analysis result exist, taking the intersection of the morphological clustering analysis result and the interphase clustering analysis result as a final result.
Examples
Description of the embodiments
In order to test the effectiveness of the method, specific waveform identification and positioning tests are carried out on the 130 remaining single-lead electrocardiograms, and the accuracy reaches over 95 percent. The figure takes the identification and positioning of four single-lead electrocardiogram data specific waveforms as an example;
basic parameters of data: duration: 60 seconds, frequency: 250 Hz;
during positioning, the positioning and positioning operations are all operated in the same set of parameters in order to better reflect the robustness of the program. The specific parameters are set as follows: a refractory period: 0.3 second; morphological clustering interval: 0.07 second; morphological clustering overall evaluation threshold: 0.85; morphological clustering classification evaluation threshold: 0.95; morphological clustering wide malformation class average time limit threshold: 0.12 second; interval clustering leveling threshold: 0.2 second; interval clustering overall evaluation threshold: 0.85; interval clustering classification evaluation threshold: 0.95; interval clustering premature beat class interval threshold: 90 times/sec.
Calculation results
The specific dominant wave that identifies the location is indicated in red "○":
1) one example of the specific main wave identification and location of the electrocardiogram of 'accidental ventricular premature beat' is shown in figure 2;
2) an example of the specific main wave identification and location of the electrocardiogram of frequent ventricular premature beats is shown in figure 3;
3) an example of specific main wave identification and location of 'sporadic atrial premature beat' is shown in figure 4;
4) an example of the specific main wave identification and location of the electrocardiogram of frequent atrial premature beats is shown in fig. 5;
as can be seen from FIGS. 2-5, the specific waveforms of each electrocardiogram are identified and located accurately.
The above embodiments are only for illustrating the technical idea of the present invention, and the protection scope of the present invention is not limited thereby, and any modifications made on the basis of the technical scheme according to the technical idea of the present invention fall within the protection scope of the present invention.
Claims (3)
1. A single-lead specific main wave identification and positioning method based on cluster analysis comprises the following steps: a) pre-treating; b) based on the selected characteristics, performing self-adaptive cluster analysis on the main wave; c) comprehensive analysis is carried out to determine the type and specific position of the specific main wave, and the method is characterized in that: the pretreatment method of a) comprises the following steps:
step 1: obtaining approximate positions of main waves on single-lead data based on a conventional analysis method, wherein the analysis method is a P-T calculation method or a QRS wave starting point and end point positioning method based on regularized least square regression learning;
step 2: removing baseline drift and high-frequency errors of original data by using MODWT (modified wavelet transform), wherein the MODWT is highly redundant non-orthogonal transform, the sample capacity can be any value, and has displacement invariance, selecting sym4 wavelet through comprehensive comparison analysis, analyzing to the 6 th layer, and selecting d2, d3, d4, d5 and d6 as reconstruction signals, thereby removing the baseline drift and the high-frequency errors;
and step 3: calibrating, namely determining the specific position and time limit of the existing main wave, and acquiring a candidate main wave position which is possibly missed, wherein the calibrating and determining the specific position and time limit of the existing main wave are divided into two steps:
for all initial dominant wave positions: replacing an original initial main wave position with an absolute value maximum value point in a small interval near a certain initial main wave position, obtaining main wave main body directions according to all main wave directions, changing or keeping the original direction of individual main wave directions inconsistent with the main body directions according to amplitude sequencing of the main wave directions in the main direction and main wave amplitude conditions in the opposite direction, determining the specific position of the main wave, and determining each main wave time limit based on the slope change condition and the QRS normal time limit range;
and outside the refractory period of the existing main wave position, finding out the maximum absolute value point in a certain interval near each point as a candidate missing main wave position, and providing preparation for targeted supplement according to a clustering analysis result in a later period.
2. The method for identifying and positioning the single-guide specific main wave based on the cluster analysis according to claim 1, wherein: the b) self-adaptive clustering analysis method for the main wave based on the selected characteristics comprises the following steps:
step 1: based on the main wave morphological characteristics, the main wave is subjected to self-adaptive cluster analysis, possible missing candidate main waves are supplemented, and possible specific main wave positions are found out:
obtaining waveform data in a certain range at two ends of each main wave point, subtracting the corresponding amplitude of the main wave point, and then taking an absolute value to obtain waveform data to be gathered;
adopting a self-adaptive clustering method for waveform data to be clustered to obtain the classification number corresponding to the highest average evaluation index, and carrying out clustering analysis on the classification number to obtain the category evaluation index corresponding to each classification;
determining whether to enter a shape clustering module according to the overall evaluation index, each classification evaluation index and each classification average time limit: when the average time limit is larger, the evaluation index is required to be larger; on the contrary, when the average time limit is smaller, the evaluation index requirement can be properly lowered, and the specific parameters are determined according to the actual analysis data;
when the waveform can be divided, the missing main wave position of the supplement candidate is considered, firstly, the centers of various types and the minimum and maximum distances between each point in the types and the center are calculated; secondly, calculating the distance between each candidate main wave position and the center of each class, taking the class with the minimum distance as a candidate class, and inspecting whether the distance is within the maximum distance range in the class, wherein the distance is slightly larger but has small difference or the distance from one class is far smaller than the distance from the adjacent class, so that the candidate main wave position is classified into the class; finally, recording the remaining candidate main wave positions;
classifying, namely determining two types of specific main wave categories, wherein the number of the categories is two, the average time limit corresponding to a certain category is the largest, and the given requirements of the wide and large malformed main wave time limits are met, and then taking the category as the specific main wave category; if the number of categories is more than two, sorting the various average time limits from large to small, and taking all categories of which the average time limits meet the requirements as specific main waves;
carrying out individual calibration on the forward specific main wave, carrying out cluster analysis on the forward main wave, and determining whether the forward main wave can be divided into two parts; if the main wave has the forward specific main wave, determining whether to remove the main wave according to whether the forward main wave can be divided into two parts and the time limit of the individual main wave, and if the main wave does not have the forward specific main wave, determining whether to supplement the individual main wave according to whether the forward main wave can be divided into two parts and the time limit of the individual main wave;
the waveform is not separable, the special condition that the negative main wave exists is considered, and the negative main wave meets the time limit requirement, and the main wave is determined as the specific main wave;
if the specific main wave does not meet the requirement, the specific main wave based on morphological clustering analysis is empty;
step 2: based on the main wave interval characteristics, carrying out self-adaptive cluster analysis on the main wave interval, supplementing possible missing candidate main waves, and finding out possible specific main wave positions:
acquiring first-order differential data of a main wave position, namely interval data;
judging whether interval data are uniform or not according to interval range, if so, the interval is not separable, and if not, the following steps are continued;
adopting a self-adaptive clustering method to obtain the classification number corresponding to the highest average evaluation index, and carrying out clustering analysis according to the classification number to obtain the evaluation index corresponding to each classification;
if the average evaluation index is greater than a given value or the evaluation index in a certain class is greater than a given value, the interval can be divided;
when the interval is available, the missing main wave position of the supplement candidate is considered, firstly, the centers of various classes and the minimum and maximum distances between each point in the classes and the center are calculated; secondly, calculating the distance between the front interval and the rear interval of each candidate main wave and the center of each class, taking the class with the minimum distance as a candidate class, and inspecting whether the distance is within the range of the maximum distance in the class, wherein the distance is slightly larger but has small difference or the distance from one class is far smaller than the distance from the adjacent class, and classifying the front interval and the rear interval of the position of the candidate main wave into the class; finally, recording the remaining candidate main wave positions;
classifying, and determining the main wave class of the specific interval, taking the rapid premature beat as an example: the number of categories is two, the interval corresponding to a certain category is the minimum, and the given interval requirement is met, the category is taken as a specific interval category, and the interval positions plus 1 are taken as the corresponding specific main wave positions; if the number of categories is more than two, sorting the average intervals from small to large, taking the interval positions corresponding to all categories with the average interval meeting the requirements plus 1 as the specific main wave position, and the escape and conduction block types are just opposite;
and if the evaluation index is smaller than the given value, the specific main wave based on interval clustering analysis is empty.
3. The method for identifying and positioning the single-guide specific main wave based on the cluster analysis according to claim 1, wherein: the c) comprehensive analysis and the method for determining the specific main wave type and the specific position comprise the following steps: determining the type and position of a specific main wave based on the main wave morphology clustering analysis result and the main wave interval clustering result;
the result is as follows: when the morphological clustering analysis result and the interval clustering analysis result do not exist, the specific main wave is an empty set;
and a second result: when the morphological clustering analysis result and the interphase clustering analysis result exist one time, the corresponding specific main wave type and the specific position are selected;
and a third result: and when both the morphological clustering analysis result and the interphase clustering analysis result exist, taking the intersection of the morphological clustering analysis result and the interphase clustering analysis result as a final result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010180896.1A CN111493858B (en) | 2020-03-16 | 2020-03-16 | Single-guide-joint specific main wave identification and positioning method based on cluster analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010180896.1A CN111493858B (en) | 2020-03-16 | 2020-03-16 | Single-guide-joint specific main wave identification and positioning method based on cluster analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111493858A true CN111493858A (en) | 2020-08-07 |
| CN111493858B CN111493858B (en) | 2022-12-09 |
Family
ID=71848228
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010180896.1A Active CN111493858B (en) | 2020-03-16 | 2020-03-16 | Single-guide-joint specific main wave identification and positioning method based on cluster analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111493858B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112869752A (en) * | 2021-02-10 | 2021-06-01 | 武汉大学 | Electrocardiosignal acquisition device and quality grade evaluation and QRS wave detection method |
| CN113111972A (en) * | 2021-05-07 | 2021-07-13 | 杭州博日科技股份有限公司 | Melting curve Tm value determination method and device based on hierarchical clustering and electronic equipment |
| CN115486854A (en) * | 2022-09-15 | 2022-12-20 | 浙江好络维医疗技术有限公司 | Single lead electrocardiogram ventricular premature beat identification method aiming at dry electrode acquisition |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019100565A1 (en) * | 2017-11-27 | 2019-05-31 | 乐普（北京）医疗器械股份有限公司 | Method and device for self-learning dynamic electrocardiography analysis employing artificial intelligence |
| CN109875548A (en) * | 2019-03-24 | 2019-06-14 | 浙江好络维医疗技术有限公司 | A kind of Characteristics of electrocardiogram waveform clustering method based on multi-lead comprehensive analysis |
| CN109893119A (en) * | 2019-03-24 | 2019-06-18 | 浙江好络维医疗技术有限公司 | A kind of P wave recognition positioning method based on multi-lead clustering |
| CN110236529A (en) * | 2019-07-19 | 2019-09-17 | 浙江好络维医疗技术有限公司 | A kind of multi-lead arrhythmia cordis intelligent diagnosing method based on MODWT and LSTM |
| CN110367969A (en) * | 2019-07-05 | 2019-10-25 | 复旦大学 | A kind of improved electrocardiosignal K-Means Cluster |
-
2020
- 2020-03-16 CN CN202010180896.1A patent/CN111493858B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019100565A1 (en) * | 2017-11-27 | 2019-05-31 | 乐普（北京）医疗器械股份有限公司 | Method and device for self-learning dynamic electrocardiography analysis employing artificial intelligence |
| CN109875548A (en) * | 2019-03-24 | 2019-06-14 | 浙江好络维医疗技术有限公司 | A kind of Characteristics of electrocardiogram waveform clustering method based on multi-lead comprehensive analysis |
| CN109893119A (en) * | 2019-03-24 | 2019-06-18 | 浙江好络维医疗技术有限公司 | A kind of P wave recognition positioning method based on multi-lead clustering |
| CN110367969A (en) * | 2019-07-05 | 2019-10-25 | 复旦大学 | A kind of improved electrocardiosignal K-Means Cluster |
| CN110236529A (en) * | 2019-07-19 | 2019-09-17 | 浙江好络维医疗技术有限公司 | A kind of multi-lead arrhythmia cordis intelligent diagnosing method based on MODWT and LSTM |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112869752A (en) * | 2021-02-10 | 2021-06-01 | 武汉大学 | Electrocardiosignal acquisition device and quality grade evaluation and QRS wave detection method |
| CN112869752B (en) * | 2021-02-10 | 2022-02-01 | 武汉大学 | Electrocardiosignal acquisition device and quality grade evaluation and QRS wave detection method |
| CN113111972A (en) * | 2021-05-07 | 2021-07-13 | 杭州博日科技股份有限公司 | Melting curve Tm value determination method and device based on hierarchical clustering and electronic equipment |
| CN113111972B (en) * | 2021-05-07 | 2023-02-24 | 杭州博日科技股份有限公司 | Melting curve Tm value determination method and device based on hierarchical clustering and electronic equipment |
| CN115486854A (en) * | 2022-09-15 | 2022-12-20 | 浙江好络维医疗技术有限公司 | Single lead electrocardiogram ventricular premature beat identification method aiming at dry electrode acquisition |
| CN115486854B (en) * | 2022-09-15 | 2024-04-30 | 浙江好络维医疗技术有限公司 | Single-lead electrocardiograph ventricular premature beat identification method for dry electrode acquisition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111493858B (en) | 2022-12-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111493858B (en) | Single-guide-joint specific main wave identification and positioning method based on cluster analysis | |
| CN109117730B (en) | Real-time electrocardiogram atrial fibrillation judgment method, device and system and storage medium | |
| US7769434B2 (en) | Method of physiological data analysis and measurement quality check using principal component analysis | |
| CN109893119B (en) | P wave identification and positioning method based on multi-lead clustering analysis | |
| CN109758145B (en) | Automatic sleep staging method based on electroencephalogram causal relationship | |
| WO2009091583A1 (en) | Method and apparatus for predicting patient outcomes form a physiological segmentable patient signal | |
| CN104834921B (en) | Electrocardio just/abnormal big data processing method and processing device | |
| KR102632272B1 (en) | Classification method of atrial fibrillation and congestive heart failure using a convolutional artificial neural network | |
| CN109875548B (en) | Electrocardiogram characteristic waveform clustering method based on multi-lead comprehensive analysis | |
| Jones et al. | Improving ECG classification interpretability using saliency maps | |
| CN112932498A (en) | T wave morphology classification system with strong generalization capability based on deep learning | |
| CN111261278A (en) | Heart disease detection method based on deep learning model of three-dimensional image | |
| Xie et al. | Research on heartbeat classification algorithm based on CART decision tree | |
| CN109893124B (en) | Difficult electrocardiogram dominant wave identification and positioning method based on multi-lead clustering analysis | |
| CN114190950B (en) | Electrocardiogram intelligent analysis method for noise-containing label and electrocardiograph | |
| CN113069124A (en) | Electrocardio monitoring method based on CNN-ET model | |
| Kim et al. | Lightweight Convolutional Neural Network for Real-Time Arrhythmia Classification on Low-Power Wearable Electrocardiograph | |
| CN113729724B (en) | Electrocardiogram measuring method and device | |
| CN113940682B (en) | Atrial fibrillation recognition method based on statistical characteristics | |
| CN111493859B (en) | Single-lead wide and large malformation main wave identification and positioning method based on dtw-hierarchical clustering analysis | |
| Gregg et al. | 12-Lead ECG interpretation by database comparison | |
| Kovalchuk et al. | A novel feature vector for ECG classification using deep learning | |
| CN111513702A (en) | Method for detecting electrocardiographic data and storage medium | |
| Wang et al. | Accurately Identifying Coronary Atherosclerotic Heart Disease through Merged Beats of Electrocardiogram | |
| Wen et al. | ECG beat classification using GreyART network |
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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 310012 block B, 5 / F, building e, Paradise Software Park, No.3 xidoumen Road, Xihu District, Hangzhou City, Zhejiang Province Applicant after: ZHEJIANG HELOWIN MEDICAL TECHNOLOGY Co.,Ltd. Address before: 310012 A building 7D, Paradise Software Park, 3 West Road, Hangzhou, Zhejiang, Xihu District Applicant before: ZHEJIANG HELOWIN MEDICAL TECHNOLOGY Co.,Ltd. |
|
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 310012 Block A, 13/F, Building E, Paradise Software Park, No. 3, Xidoumen Road, Xihu District, Hangzhou, Zhejiang Applicant after: ZHEJIANG HELOWIN MEDICAL TECHNOLOGY CO.,LTD. Address before: 310012 block B, 5 / F, building e, Paradise Software Park, No.3 xidoumen Road, Xihu District, Hangzhou City, Zhejiang Province Applicant before: ZHEJIANG HELOWIN MEDICAL TECHNOLOGY CO.,LTD. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: Building D, 14th Floor, Block C, Tian Tang Software Park, No. 3 Xidoumen Road, Gudang Street, Xihu District, Hangzhou City, Zhejiang Province 310012 Patentee after: ZHEJIANG HELOWIN MEDICAL TECHNOLOGY CO.,LTD. Country or region after: China Address before: 310012 Block A, 13/F, Building E, Paradise Software Park, No. 3, Xidoumen Road, Xihu District, Hangzhou, Zhejiang Patentee before: ZHEJIANG HELOWIN MEDICAL TECHNOLOGY CO.,LTD. Country or region before: China |