CN115486854B - Single-lead electrocardiograph ventricular premature beat identification method for dry electrode acquisition - Google Patents
Single-lead electrocardiograph ventricular premature beat identification method for dry electrode acquisition Download PDFInfo
- Publication number
- CN115486854B CN115486854B CN202211122514.5A CN202211122514A CN115486854B CN 115486854 B CN115486854 B CN 115486854B CN 202211122514 A CN202211122514 A CN 202211122514A CN 115486854 B CN115486854 B CN 115486854B
- Authority
- CN
- China
- Prior art keywords
- wave
- sum
- beat
- heart beat
- point
- 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
- 208000000418 Premature Cardiac Complexes Diseases 0.000 title claims abstract description 52
- 230000002861 ventricular Effects 0.000 title claims abstract description 52
- 238000000034 method Methods 0.000 title claims abstract description 40
- 230000000877 morphologic effect Effects 0.000 claims abstract description 15
- 206010015856 Extrasystoles Diseases 0.000 claims abstract description 14
- 238000001514 detection method Methods 0.000 claims abstract description 13
- 230000000747 cardiac effect Effects 0.000 claims abstract description 9
- 238000007781 pre-processing Methods 0.000 claims abstract description 5
- 230000007423 decrease Effects 0.000 claims description 12
- 238000012360 testing method Methods 0.000 claims description 11
- 238000000718 qrs complex Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 9
- 230000011218 segmentation Effects 0.000 claims description 9
- 238000005070 sampling Methods 0.000 claims description 7
- 238000012549 training Methods 0.000 claims description 7
- 101100129500 Caenorhabditis elegans max-2 gene Proteins 0.000 claims description 6
- 101100083446 Danio rerio plekhh1 gene Proteins 0.000 claims description 6
- 101100116390 Schizosaccharomyces pombe (strain 972 / ATCC 24843) ded1 gene Proteins 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 230000003247 decreasing effect Effects 0.000 claims 1
- 238000011156 evaluation Methods 0.000 claims 1
- 238000004422 calculation algorithm Methods 0.000 abstract description 6
- 229920000915 polyvinyl chloride Polymers 0.000 description 28
- 239000004800 polyvinyl chloride Substances 0.000 description 28
- 238000002565 electrocardiography Methods 0.000 description 8
- 230000008859 change Effects 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000003909 pattern recognition Methods 0.000 description 3
- 238000012567 pattern recognition method Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000033764 rhythmic process Effects 0.000 description 2
- 206010002383 Angina Pectoris Diseases 0.000 description 1
- 206010008479 Chest Pain Diseases 0.000 description 1
- 206010033557 Palpitations Diseases 0.000 description 1
- 208000009729 Ventricular Premature Complexes Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 206010003119 arrhythmia Diseases 0.000 description 1
- 230000006793 arrhythmia Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000004660 morphological change Effects 0.000 description 1
- 230000002028 premature Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
-
- 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
-
- 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
- A61B5/35—Detecting specific parameters of the electrocardiograph cycle by template matching
-
- 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
- A61B5/352—Detecting R peaks, e.g. for synchronising diagnostic apparatus; Estimating R-R interval
-
- 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
- A61B5/355—Detecting T-waves
-
- 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
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- 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
- A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
-
- 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
- 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
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
Abstract
A method for identifying ventricular premature beat of single lead electrocardiograph for dry electrode acquisition, comprising the steps of: step one: acquiring and preprocessing single-lead electrocardiosignals; step two: the 9 lines of the ECG heart beat waveform state are positioned; step three: detecting special ventricular premature beats based on medical knowledge; step four: extracting approximate waveform characteristics of the heart beat; step five: on the basis of detecting the special ventricular premature beat for the first time based on medical knowledge in the third step, detecting the ventricular premature beat continuously based on LightGBM models of approximate waveforms, and the sixth step: and (5) evaluating a model. The ventricular premature beat identification method provided by the invention is used for identifying the complex PVC cardiac beat by combining 9 pieces of medical knowledge in the ventricular premature beat medical field and combining a model detection algorithm similar to waveform morphological characteristics, and solves the problem that the method depends on the sectional positioning of an electrocardiographic waveform, so that the accuracy and the efficiency of ventricular premature beat detection can be effectively improved.
Description
Technical Field
The invention relates to a single-lead electrocardiograph ventricular premature beat identification method aiming at dry electrode acquisition, belonging to the field of electrocardiograph data processing.
Background
Ventricular premature beat (Premature ventricular contraction, PVC), one type of common arrhythmia, refers to abnormal heart beats caused by premature onset of impulses at ectopic pacing sites, which may occur on the basis of sinus or ectopic heart rhythm, and long-term delays can lead to chest distress, palpitations, angina and even death. The PVC electrocardiogram is mainly characterized by a pre-emergent broad-deformity QRS-T complex, T wave number and QRS main wave direction are opposite, and related P waves are not generated before the QRS main wave, and whether the characteristics are distorted or not is an important condition for judging whether a patient has ventricular premature beat or not. Traditional identification of PVC electrocardiographs relies on extensive knowledge and personal experience in the field of physician cardiology, and current medical techniques are used to analyze electrocardiographic waveforms to make identification results. However, PVC may cause various morphological changes on an electrocardiogram due to different factors such as basic disease of a patient and physical constitution of an individual, and a large amount of detection data also increases the subsequent recognition work intensity of a doctor, and may cause misdiagnosis of the doctor under the condition of long-time work. In order to reduce the burden of doctors, the auxiliary doctors improve the efficiency and accuracy of PVC electrocardiogram recognition, and the automatic recognition of PVC electrocardiogram gradually becomes a research hot spot.
Currently, automatic identification of a single-lead electrocardiogram for dry electrode acquisition (PVC) generally comprises five stages (such as FIG. 1) of electrocardiosignal acquisition, preprocessing denoising, electrocardiographic waveform segmentation positioning, feature extraction and automatic classification, and the method can be divided into three main types: rule-based recognition methods, waveform-based pattern recognition methods, and deep learning-based recognition methods.
The single-lead electrocardiogram based on the dry electrode acquisition method for identifying the PVC comprises the following basic steps: the boundary point position and peak point of wavelet forms such as P wave and QRS wave in the heart beat are detected by the electrocardio waveform segmentation positioning algorithm, then the interval and amplitude of the wavelets are described, and the heart beat conforming to the corresponding rhythm and form in medicine can be detected as the PVC heart beat. The characteristics extracted by the method can be interpreted strongly, the time complexity is low, and the method has a good recognition effect on typical regular PVC heart beats, but an electrocardiogram is a weak physiological electric signal detected and recorded through body surface electrodes, the voltage amplitude of the weak physiological electric signal is usually not more than 5mv, so that the electrocardiograph is extremely easy to be interfered by noise when acquiring electrocardiograph information, the electrocardiograph is distorted, the PVC heart beats cannot be accurately recorded into a medical definition form, and the automatic classification method of PVC based on rules cannot achieve satisfactory degree.
The PVC pattern recognition method based on waveforms for the single-lead electrocardiogram acquired by the dry electrode comprises the following basic steps: according to the electrocardiosignal waveform segmentation positioning algorithm, waveform morphological characteristic parameters such as T wave amplitude, QT interval, slope and the like of the electrocardiosignal are extracted, and other researches focus on the fact that the electrocardiosignal has different performances in one or more characteristic spaces to identify PVC (polyvinyl chloride), such as wavelet transformation characteristics and the like, and then the PVC is input into a pattern identification algorithm to detect each heart beat. The method has the advantages that the detection effect is overall and is more advantageous, the characteristic space recognition and calculation process is more complicated, the automatic recognition of the morphological characteristics of PVC is dependent on the early-stage correct morphological segmentation positioning of the electrocardiographic waveform, the accuracy of detecting the R wave crest value point is higher, the positioning of morphological boundary points of wavelets such as P waves, T waves and the like is troublesome, the method is difficult to be used as the characteristic of the electrocardiograph, and if the peak value point and the boundary point used for positioning cannot reflect the real attribute of the electrocardiograph, the automatic recognition precision and efficiency cannot be applied to practice necessarily.
The basic idea of the deep learning-based PVC identification method for the single-lead electrocardiogram acquired by the dry electrode is to input an electrocardiogram signal, automatically extract electrocardiogram characteristics, convert the electrocardiogram signal from an original sample space into a deeper and more abstract space through layer-by-layer conversion, and detect the PVC signal by utilizing the obtained characteristics. Although the effect is better, the deep learning needs more training data, and in the huge electrocardiosignals, the quantity of complex PVC signals is smaller, on the other hand, the automatically learned characteristics are difficult to explain, and when misjudgment and missed judgment of samples occur, the model design is difficult to adjust and change.
Disclosure of Invention
The method aims at the single-lead electrocardiosignals acquired by the dry electrode, firstly solves the problem that one of the PVC methods based on medical knowledge or waveform-based pattern recognition cannot accurately and rapidly detect complex PVC signals, combines the medical knowledge and waveform pattern recognition by considering the characteristics of the two methods, and provides a ventricular premature beat automatic recognition method based on the combination of the medical knowledge and approximate waveform morphological characteristics for the single-lead electrocardiosignals acquired by the dry electrode; secondly, the method solves the problem that the extraction process of morphological characteristics in the waveform-based pattern recognition PVC method depends on the sectional positioning of electrocardiographic waveforms, and provides a characteristic extraction method of approximate waveforms of single electrocardiographic signals. And the set rules and the extracted approximate waveform features have strong interpretability, and the parameters and the model design are convenient to adjust and change aiming at different scenes.
The technical scheme of the invention is as follows: a method of single-lead electrocardiographic ventricular premature beat identification for dry electrode acquisition, the method comprising the steps of:
step one: and acquiring and preprocessing single-lead electrocardiosignals. According to a preset sampling rate, acquiring single-lead ECG samples of i training sets and j testing sets with fixed lengths for m seconds by adopting a dry electrode, and carrying out filtering denoising, myoelectricity removing, power frequency interference removing and baseline drift removing treatment on each sample.
Step two: the ECG heart beat waveform is located with 9 lines in state. Firstly, from each ECG sample preprocessed in the first step, the R wave position with the most obvious positioning characteristic and high detection accuracy is obtained for each heart beat; secondly, taking the detected R wave position as the center, taking a fixed window with a certain number of sampling points forwards and backwards as the basis for segmentation, representing a heart beat QRS complex represented by the R wave, and co-segmenting to obtain n heart beats; thirdly, weighting the 2 nd to n-1 th single heart beats to obtain a superposition wave corresponding to the n-2 heart beats of one sample ECG signal; and finally, carrying out the positions of a QRS wave starting point, a Q wave vertex, an R wave vertex, an S wave vertex, a QRS wave ending point, a T wave starting point, a T wave vertex, a T wave ending point and a main wave on all single heart beats and superimposed waves of each sample, and carrying out the positioning on the positions of 9 sub-wave key points.
Step three: detecting special ventricular premature beats based on medical knowledge, classifying each beat of a sample according to the background of the field and the ventricular premature beats, determining 9 rules from 9 positioning points of the second step to detect the special PVC beats, judging whether the beat signals accord with the 9 rules, if so, judging that the beat is the ventricular premature beats, and if not, continuing the fourth step;
Step four: and extracting approximate waveform characteristics of the heart beat. According to the point position of the R wave top point of the heart beat positioned in the second step, the QRS wave is approximately expressed by overlapping a few points before and after the point position, so that the most critical morphological characteristics of the QRS wave group of the heart beat are extracted, the complex flow of searching the starting point and the ending point of the QRS wave in the segmentation positioning stage of the electrocardio wave is replaced, the QRS wave morphological characteristics of each heart beat are described by a plurality of parameters such as slope, height, width, proportion and the like, and the characteristic set comprises a slope characteristic group, a height characteristic group and a width characteristic group.
Step five: and on the basis of detecting the special ventricular premature beat for the first time based on medical knowledge in the step three, detecting the ventricular premature beat continuously based on a LightGBM model of an approximate waveform. And (3) extracting the slope, height and width characteristic groups of the non-ventricular premature beat detected in the step (III), and inputting the characteristic groups into a LightGBM model for training by combining corresponding cardiac beat labels to obtain a ventricular premature beat detection model.
Step six: and (5) evaluating a model. The medical knowledge and the trained ventricular premature beat recognition model are applied to sample data of a test set, the final output result is compared with a label of the sample of the test set, and the recall rate and the precision are used for evaluating the performance of the model.
As preferable: the 9 rules in the third step are respectively as follows:
rule 1: the product of the difference Deltah 1 between the amplitude of the heart beat R wave and the amplitude of the QRS wave starting point and the difference Deltah 2,△h1 between the amplitude of the T wave and the amplitude of the QRS wave starting point and Deltah 2 is smaller than 0;
Rule 2: the heart beat slope 1 is larger than the slope 4, the slope 2 is larger than the slope 4, and the slope 4 is larger than the slope 3;
The slope 1 and 2 is calculated by: if the main wave is on, calculating the slopes of the R wave, the Q wave and the S wave as slopes 1 and 2 respectively; if the main wave is downward and the R wave amplitude value is more than or equal to 0, calculating the slopes of the S wave, the R wave and the QRS wave termination points to be respectively 1 and 2; if the main wave is downward and the R wave amplitude value is smaller than 0, calculating the slopes of the starting point and the ending point of the S wave and the QRS wave to be respectively 1 and 2;
slope 3 calculation mode: slope of the T wave apex and T wave onset;
slope 4 calculation mode: slope of T wave apex and T wave termination point;
Rule 3: the amplitude of the main wave peak of the heart beat is simultaneously larger than or smaller than the amplitude of the start and the end of the QRS wave, and the amplitude of the T wave peak is also simultaneously larger than or smaller than the amplitude of the start and the end of the T wave;
Rule 4: the multiplication product of the main wave amplitude value of the heart beat and the amplitude value of the T wave is smaller than 0;
rule 5: the monotonicity of the front and the back of the heart beat T wave is satisfied;
When the T wave direction is downward, dividing the area from the starting point of the T wave to the vertex of the T wave into two sub-areas: the method comprises the steps of taking a voltage value of a T wave starting point as a maximum value max1 of the subarea 1, calculating a maximum value max2 of the subarea 2, and determining a voltage value of the T wave starting point; the area from the T-wave apex to the T-wave termination point is subdivided into two sub-areas: the voltage value of the T wave termination point is taken as the maximum value max4 of the subarea 4, and then the maximum value max3 of the subarea 3 is calculated; since the T wave direction decreases downward, the left part decreases monotonically, and the right part increases monotonically, at which time max1> =max2 and max4> =max3 need to be satisfied, i.e., the T wave ending point and starting point are the maximum values after the T wave front, respectively. Otherwise, when the T wave direction is upward, the same is true.
Rule 6: the relationship between the maximum value of the QRS complex and the maximum value of the T complex of the heart beat is satisfied;
when the T wave direction is downward, the minimum value of the T wave group is smaller than the minimum value of the QRS wave group, and the maximum value of the T wave group is smaller than the maximum value of the QRS wave group; the maximum value of the T complex is greater than the maximum value of the QRS complex and the minimum value of the T complex is greater than the minimum value of the QRS complex when the T complex is directed upward.
Rule 7: firstly, repositioning R waves aiming at heart beats with downward main wave direction and R wave peak amplitude smaller than 0: calculating the maximum point of the intermediate region of the Q wave and the S wave to be used as the peak of the R wave;
[1] For heart beats with the main wave downward direction and the amplitude of the R wave top point being more than or equal to 0, the difference h 1 between the R wave top point and the Q wave top point amplitude is calculated, the difference h 2 between the R wave top point and the S wave top point amplitude is calculated, and the width w from the R wave to the QRS wave end point is satisfied, wherein the value h 1/h2 w is less than or equal to the set threshold value TH 1.
[2] Heart beat with downward main wave direction and R wave peak amplitude less than 0:
a) Excluding heart beats with R waves larger than Q waves and S wave amplitudes;
b) The voltage fluctuations in the region from the QRS wave end point to the T wave start point, in region 1 and in region 4 of the T wave group are smaller: calculating the maximum value and the minimum value of the voltage of each region, and then making a difference with the voltage values of the starting point and the ending point of the corresponding region (if the region monotonically increases, the voltage of the starting point of the region is made different from the voltage of the minimum value, and if the region monotonically decreases, the voltage of the ending point of the region is made different from the voltage of the maximum value, and if the region monotonically decreases, the opposite is achieved), so that the difference is smaller than a set threshold value TH 2;
[3] for heart beat with main wave direction upward:
a) The difference between the amplitude of the termination point of the QRS and the amplitude of the S wave is smaller than or equal to a set threshold value TH 3;
b) The voltage fluctuations in the region from the QRS wave end point to the T wave start point, in region 1 and in region 4 of the T wave group are smaller: and calculating the maximum value and the minimum value of the voltage of each region, and then performing difference with the voltage values of the starting point and the ending point of the corresponding region (if the region monotonically increases, the voltage of the starting point of the region is the minimum value, and if the region monotonically decreases inversely), the voltage of the ending point of the region is the maximum value, so that the difference is smaller than the set threshold TH 4.
Rule 8: heart beat cosine similarity: the transverse index is the QRS wave group starting point to the T wave group ending point of the superimposed wave, and the longitudinal direction is the alignment of the main wave vertexes of the front heart beat and the superimposed wave (the alignment of the R wave vertexes when the main wave direction is upward, the alignment of the S wave vertexes when the main wave direction is downward);
a) For the main wave direction upward heart beat: the cosine similarity is smaller than a set threshold value TH 5.
B) Downward heart beat for main wave direction: the cosine similarity is smaller than a set threshold value TH 6.
Rule 9: for the main wave direction upward heart beat:
a) The R wave amplitude of the current heart beat is larger than that of the superimposed wave.
B) The R-wave amplitude of the current beat is greater than the maximum amplitude of the ST segment.
As preferable: the slope characteristic group in the fourth step is as follows:
Sum of R-wave vertex of current heart beat and slope of the previous 7 points (sum 1);
sum of R-wave peak and slope of the previous 7 points of the superimposed wave (sum 2);
sum of slopes of the R-wave vertex of the current beat and 7 points later (sum 3);
Sum of R-wave peak of superimposed wave and slope of 7 points later (sum 4);
difference, ratio and percentage of difference between sum1 and sum 2;
difference, ratio and percentage of difference between sum3 and sum 4;
as preferable: the height feature group in the fourth step is as follows:
The sum of absolute values of differences (sum 5) is correspondingly made at 15 points (the peak point of the R wave and 7 points on the front and back) of the R wave of the current heart beat and the superimposed wave;
Sum of absolute values of 15 points of R wave of the current heart beat (sum 6);
sum of absolute values of 201 points of the current heart beat (sum 7);
sum of absolute values of 15 points of the R wave of the superimposed wave (sum 8);
sum of 201 points absolute values of the superimposed wave (sum 9);
Ratio R1 of sum6 to sum 7;
ratio R2 of sum8 to sum 9;
difference, ratio and percentage of difference between R1 and R2;
The sum of absolute difference values (sum 10) is correspondingly made for each 201 points of the current heart beat and the superimposed wave;
Ratio of sum10 to sum 9;
Ratio of sum5 to sum 8;
as preferable: the width feature group in the fourth step is as follows:
R wavefront: drawing 7 transverse lines by 7 points in front of the R wave vertex of the current heart beat, respectively calculating the sum of the differences between 7 transverse indexes of the intersection point of the current heart beat and the R wave vertex index of the current heart beat as R wave front width (w 1) of the superimposed wave, and taking the sum of the differences between 7 transverse indexes of the current heart beat and the intersection point index of the superimposed wave as R wave front width difference characteristic (w 2);
R wave: and similarly, 7 transverse lines are stippled by 7 points behind the R wave vertex of the current heart beat to obtain the R wave width (w 3) and the R wave width difference value characteristic (w 4) of the superimposed wave.
The invention aims at the single-lead electrocardiosignal collected by a dry electrode, provides a ventricular premature beat identification method based on combination of medical knowledge and approximate waveform morphological characteristics, combines 9 pieces of medical knowledge in the ventricular premature beat medical field, detects special PVC heart beats, integrates a model detection algorithm of the approximate waveform morphological characteristics to identify complex PVC heart beats, solves the problem that the method depends on electrocardiographic waveform sectional positioning, and can effectively improve the accuracy and efficiency of ventricular premature beat detection. And the set rules and the extracted approximate waveform features have strong interpretability, and the parameters and the model design are convenient to adjust and change aiming at different scenes.
Drawings
FIG. 1 is an automated identification process for PVC electrocardiography;
FIG. 2 is a flowchart of the overall algorithm of the present invention;
FIG. 3 shows two specific ventricular premature beats and corresponding superimposed waves;
fig. 4 is a flow chart of rule recognition according to the present invention.
Detailed Description
The invention will be described in detail below with reference to the attached drawings: as shown in fig. 2, for a single-lead electrocardiographic signal acquired by a dry electrode, a ventricular premature beat identification method based on medical knowledge and approximate waveform morphology features, the method comprising the steps of:
Step one: and acquiring and preprocessing single-lead electrocardiosignals. According to the sampling rate of 250Hz, acquiring electrocardiosignals with fixed length for 30 seconds by using a dry electrode, training samples 14891 parts, wherein ventricular premature beat samples 4973 parts, non-ventricular premature beat samples 9918 parts, sinus cardiac beats 6652 parts, interference samples 3266 parts, test samples 60399 parts, wherein 1537 parts of ventricular premature beat data, and carrying out filtering denoising, myoelectricity removing, power frequency interference removing and baseline drift removing treatment on each sample.
Step two: the ECG heart beat waveform is located with 9 lines in state. Firstly, from each ECG sample preprocessed in the first step, the R wave position with the most obvious positioning characteristic and high detection accuracy is obtained for each heart beat; secondly, taking the detected R wave position as a center to forward take a 75ms sampling point and backward take a 125ms sampling point fixed window as a basis to divide, and representing a heart beat QRS complex represented by the R wave to obtain n heart beats in a total dividing way; thirdly, weighting the 2 nd to n-1 th single heart beats to obtain a superposition wave corresponding to the n-2 heart beats of one sample ECG signal; and finally, carrying out the positions of a QRS wave starting point, a Q wave vertex, an R wave vertex, an S wave vertex, a QRS wave ending point, a T wave starting point, a T wave vertex, a T wave ending point and a main wave on all single heart beats and superimposed waves of each sample, and carrying out the positioning on the positions of 9 sub-wave key points.
Step three: the detection of special ventricular premature beats is based on medical knowledge. As shown in fig. 3, two special ventricular premature beats and corresponding superimposed waves are adopted, each cardiac beat of a sample is classified according to the background of the field and the medical knowledge of ventricular premature beats, 9 rules are determined from 9 positioning points in the second step to detect the special PVC cardiac beats, whether the cardiac beat signals accord with the following 9 rules (specific rule block diagram is shown in fig. 4), if so, the cardiac beat is the ventricular premature beat, and if not, the fourth step is continued;
rule 1: the product of the difference Deltah 1 between the amplitude of the heart beat R wave and the amplitude of the QRS wave starting point and the difference Deltah 2,△h1 between the amplitude of the T wave and the amplitude of the QRS wave starting point and Deltah 2 is smaller than 0;
Rule 2: the heart beat slope 1 is larger than the slope 4, the slope 2 is larger than the slope 4, and the slope 4 is larger than the slope 3;
The slope 1 and 2 is calculated by: if the main wave is on, calculating the slopes of the R wave, the Q wave and the S wave as slopes 1 and 2 respectively; if the main wave is downward and the R wave amplitude value is more than or equal to 0, calculating the slopes of the S wave, the R wave and the QRS wave termination points to be respectively 1 and 2; if the main wave is downward and the R wave amplitude value is smaller than 0, calculating the slopes of the starting point and the ending point of the S wave and the QRS wave to be respectively 1 and 2;
slope 3 calculation mode: slope of the T wave apex and T wave onset;
slope 4 calculation mode: slope of T wave apex and T wave termination point;
Rule 3: the amplitude of the main wave peak of the heart beat is simultaneously larger than or smaller than the amplitude of the start and the end of the QRS wave, and the amplitude of the T wave peak is also simultaneously larger than or smaller than the amplitude of the start and the end of the T wave;
Rule 4: the multiplication product of the main wave amplitude value of the heart beat and the amplitude value of the T wave is smaller than 0;
rule 5: the monotonicity of the front and the back of the heart beat T wave is satisfied;
When the T wave direction is downward, dividing the area from the starting point of the T wave to the vertex of the T wave into two sub-areas: the method comprises the steps of taking a voltage value of a T wave starting point as a maximum value max1 of the subarea 1, calculating a maximum value max2 of the subarea 2, and determining a voltage value of the T wave starting point; the area from the T-wave apex to the T-wave termination point is subdivided into two sub-areas: the voltage value of the T wave termination point is taken as the maximum value max4 of the subarea 4, and then the maximum value max3 of the subarea 3 is calculated; since the T wave direction decreases downward, the left part decreases monotonically, and the right part increases monotonically, at which time max1> =max2 and max4> =max3 need to be satisfied, i.e., the T wave ending point and starting point are the maximum values after the T wave front, respectively. Otherwise, when the T wave direction is upward, the same is true.
Rule 6: the relationship between the maximum value of the QRS complex and the maximum value of the T complex of the heart beat is satisfied;
when the T wave direction is downward, the minimum value of the T wave group is smaller than the minimum value of the QRS wave group, and the maximum value of the T wave group is smaller than the maximum value of the QRS wave group; the maximum value of the T complex is greater than the maximum value of the QRS complex and the minimum value of the T complex is greater than the minimum value of the QRS complex when the T complex is directed upward.
Rule 7: firstly, repositioning R waves aiming at heart beats with downward main wave direction and R wave peak amplitude smaller than 0: calculating the maximum point of the intermediate region of the Q wave and the S wave to be used as the peak of the R wave;
[1] For heart beats with the main wave downward direction and the amplitude of the R wave top point being more than or equal to 0, the difference h 1 between the R wave top point and the Q wave top point amplitude is calculated, the difference h 2 between the R wave top point and the S wave top point amplitude is calculated, and the width w from the R wave to the QRS wave end point is satisfied, wherein the value h 1/h2 w is less than or equal to the set threshold value 0.01.
[2] Heart beat with downward main wave direction and R wave peak amplitude less than 0:
a) Excluding heart beats with R waves larger than Q waves and S wave amplitudes;
b) The voltage fluctuations in the region from the QRS wave end point to the T wave start point, in region 1 and in region 4 of the T wave group are smaller: calculating the maximum value and the minimum value of the voltage of each region, and then making a difference with the voltage values of the starting point and the ending point of the corresponding region (if the region monotonically increases, the voltage of the starting point of the region is made different from the voltage of the minimum value, and if the region monotonically decreases, the voltage of the ending point of the region is made different from the voltage of the maximum value, and if the region monotonically decreases, the opposite is true), so that the difference is smaller than the set threshold value of 0.1;
[3] for heart beat with main wave direction upward:
a) The difference between the amplitude of the termination point of the QRS and the amplitude of the S wave is smaller than or equal to a set threshold value of 0.1;
b) The voltage fluctuations in the region from the QRS wave end point to the T wave start point, in region 1 and in region 4 of the T wave group are smaller: and calculating the maximum value and the minimum value of the voltage of each region, and then performing difference with the voltage values of the starting point and the ending point of the corresponding region (if the region monotonically increases, the voltage of the starting point of the region is the minimum value, and if the region monotonically decreases, the voltage of the ending point of the region is the difference with the voltage of the maximum value), so that the difference is smaller than the set threshold value of 0.1.
Rule 8: heart beat cosine similarity: the transverse index is the QRS wave group starting point to the T wave group ending point of the superimposed wave, and the longitudinal direction is the alignment of the main wave vertexes of the front heart beat and the superimposed wave (the alignment of the R wave vertexes when the main wave direction is upward, the alignment of the S wave vertexes when the main wave direction is downward);
a) For the main wave direction upward heart beat: the cosine similarity is smaller than the set threshold value of 0.7.
B) Downward heart beat for main wave direction: the cosine similarity is smaller than the set threshold value of 0.2.
Rule 9: for the main wave direction upward heart beat:
a) The R wave amplitude of the current heart beat is larger than that of the superimposed wave.
B) The R-wave amplitude of the current beat is greater than the maximum amplitude of the ST segment.
Step four: and extracting approximate waveform characteristics of the heart beat. According to the point position of the R wave top point of the heart beat positioned in the second step, the QRS wave is approximately expressed by overlapping a few points before and after the point position, so that the most critical morphological characteristics of the QRS wave group of the heart beat are extracted, the complex flow of searching the starting point and the ending point of the QRS wave in the segmentation positioning stage of the electrocardio wave is replaced, the QRS wave morphological characteristics of each heart beat are described by a plurality of parameters such as slope, height, width, proportion and the like, and the characteristic set comprises a slope characteristic group, a height characteristic group and a width characteristic group.
(1) Slope characteristic group
Sum of R-wave vertex of current heart beat and slope of the previous 7 points (sum 1);
sum of R-wave peak and slope of the previous 7 points of the superimposed wave (sum 2);
sum of slopes of the R-wave vertex of the current beat and 7 points later (sum 3);
Sum of R-wave peak of superimposed wave and slope of 7 points later (sum 4);
difference, ratio and percentage of difference between sum1 and sum 2;
difference, ratio and percentage of difference between sum3 and sum 4;
(2) Height feature group
The sum of absolute values of differences (sum 5) is correspondingly made at 15 points (the peak point of the R wave and 7 points on the front and back) of the R wave of the current heart beat and the superimposed wave;
Sum of absolute values of 15 points of R wave of the current heart beat (sum 6);
sum of absolute values of 201 points of the current heart beat (sum 7);
sum of absolute values of 15 points of the R wave of the superimposed wave (sum 8);
sum of 201 points absolute values of the superimposed wave (sum 9);
Ratio R1 of sum6 to sum 7;
ratio R2 of sum8 to sum 9;
difference, ratio and percentage of difference between R1 and R2;
The sum of absolute difference values (sum 10) is correspondingly made for each 201 points of the current heart beat and the superimposed wave;
Ratio of sum10 to sum 9;
Ratio of sum5 to sum 8;
(3) Width feature group
R wavefront: drawing 7 transverse lines by 7 points in front of the R wave vertex of the current heart beat, respectively calculating the sum of the differences between 7 transverse indexes of the intersection point of the current heart beat and the R wave vertex index of the current heart beat as R wave front width (w 1) of the superimposed wave, and taking the sum of the differences between 7 transverse indexes of the current heart beat and the intersection point index of the superimposed wave as R wave front width difference characteristic (w 2);
R wave: similarly, 7 transverse lines are stippled by 7 points behind the R wave vertex of the current heart beat to obtain the R wave width (w 3) and the R wave width difference value characteristic (w 4) of the superimposed wave;
Step five: and on the basis of detecting the special ventricular premature beat for the first time based on medical knowledge in the step three, detecting the ventricular premature beat continuously based on a LightGBM model of an approximate waveform. And (3) extracting the slope, height and width characteristic groups of the non-ventricular premature beat detected in the step (III), and inputting the characteristic groups into a LightGBM model for training by combining corresponding cardiac beat labels to obtain a ventricular premature beat detection model.
Step six: and (5) evaluating a model. The medical knowledge and trained ventricular premature beat recognition model were applied to sample data of 60399 test sets, of which 1537 ventricular premature beat data were compared with the label of the test set sample, and the performance of the model was evaluated using recall and precision. The calculation formulas of the precision and the recall rate are as follows:
TP (True Positive): predicting to be positive, and the actual value to be positive;
FP (False Positive): predicted positive, but the actual value negative;
TN (True Negative): predicted negative, actual value negative;
FN (False Negative): the prediction is negative, but the actual value is positive.
The test results are shown in Table 1, and the calculation formulas of the precision and the recall rate can be obtained: the precision of the model is 95.8%, and the recall rate of the model is 96.6%.
TABLE 1 test results
Claims (4)
1. A single-lead electrocardiograph ventricular premature beat identification method aiming at dry electrode acquisition, which is characterized in that: the method comprises the following steps:
Step one: acquiring and preprocessing single-lead electrocardiosignals, acquiring i training sets and j single-lead ECG samples of a test set with fixed length of m seconds by adopting a dry electrode according to a preset sampling rate, and carrying out filtering denoising, myoelectricity removing, power frequency interference removing and baseline drift removing treatment on each sample;
Step two: the method comprises the steps of (1) positioning 9 lines of an ECG heart beat waveform, namely firstly, from each ECG sample preprocessed in the step one, positioning the R wave position with most obvious heart beat positioning characteristics and high detection accuracy; secondly, taking the detected R wave position as the center, taking a fixed window with a certain number of sampling points forwards and backwards as the basis for segmentation, representing a heart beat QRS complex represented by the R wave, and co-segmenting to obtain n heart beats; thirdly, weighting the 2 nd to n-1 th single heart beats to obtain a superposition wave corresponding to the n-2 heart beats of one sample ECG signal; finally, the positions of a QRS wave starting point, a Q wave vertex, an R wave vertex, an S wave vertex, a QRS wave ending point, a T wave starting point, a T wave vertex, a T wave ending point and a main wave are carried out on all single heart beats and superimposed waves of each sample, and the positions of 9 sub-wave key points are positioned;
Step three: detecting special ventricular premature beats based on medical knowledge, classifying each beat of a sample according to the background of the field and the ventricular premature beats, determining 9 rules from 9 positioning points of the second step to detect the special PVC beats, judging whether the beat signals accord with the 9 rules, if so, judging that the beat is the ventricular premature beats, and if not, continuing the fourth step;
Step four: extracting approximate waveform features of the heart beat, namely, according to the top point position of the R wave of the heart beat positioned in the second step, approximately expressing the QRS waveform by overlapping a few points before and after the top point position of the R wave of the heart beat point by point, further extracting the morphological features of the QRS wave group of the heart beat, replacing the complex flow of searching the starting point and the ending point of the QRS wave in the segmentation positioning stage of the heart wave, and describing the morphological features of the QRS wave of each heart beat by using a plurality of parameters including slope, height, width and proportion, wherein the features comprise a slope feature group, a height feature group and a width feature group;
Step five: on the basis of detecting special ventricular premature beat for the first time based on medical knowledge in the step three, continuing to detect the ventricular premature beat based on a LightGBM model of an approximate waveform, extracting slope, height and width characteristic groups of the ventricular premature beat detected in the step three according to the step four, combining corresponding cardiac beat labels, and inputting the characteristic groups into a LightGBM model for training to obtain a ventricular premature beat detection model;
Step six: model evaluation, namely applying medical knowledge and a trained ventricular premature beat recognition model to sample data of a test set, comparing a final output result with a label of the sample of the test set, and evaluating the performance of the model by using recall rate and precision;
the 9 rules in the third step are respectively as follows:
Rule 1: the product of the difference delta h 1 between the amplitude of the heart beat R wave and the amplitude of the QRS wave starting point, and the difference delta h 2,Δh1 between the amplitude of the T wave and the amplitude of the QRS wave starting point, and the difference delta h 2 are smaller than 0;
Rule 2: the heart beat slope 1 is larger than the slope 4, the slope 2 is larger than the slope 4, and the slope 4 is larger than the slope 3;
The slope 1 and 2 is calculated by: if the main wave is on, calculating the slopes of the R wave, the Q wave and the S wave as slopes 1 and 2 respectively; if the main wave is downward and the R wave amplitude value is more than or equal to 0, calculating the slopes of the S wave, the R wave and the QRS wave termination points to be respectively 1 and 2; if the main wave is downward and the R wave amplitude value is smaller than 0, calculating the slopes of the starting point and the ending point of the S wave and the QRS wave to be respectively 1 and 2;
slope 3 calculation mode: slope of the T wave apex and T wave onset;
slope 4 calculation mode: slope of T wave apex and T wave termination point;
Rule 3: the amplitude of the main wave peak of the heart beat is simultaneously larger than or smaller than the amplitude of the start and the end of the QRS wave, and the amplitude of the T wave peak is also simultaneously larger than or smaller than the amplitude of the start and the end of the T wave;
Rule 4: the multiplication product of the main wave amplitude value of the heart beat and the amplitude value of the T wave is smaller than 0;
rule 5: the monotonicity of the front and the back of the heart beat T wave is satisfied;
When the T wave direction is downward, dividing the area from the starting point of the T wave to the vertex of the T wave into two sub-areas: the method comprises the steps of taking a voltage value of a T wave starting point as a maximum value max1 of the subarea 1, calculating a maximum value max2 of the subarea 2, and determining a voltage value of the T wave starting point; the area from the T-wave apex to the T-wave termination point is subdivided into two sub-areas: the voltage value of the T wave termination point is taken as the maximum value max4 of the subarea 4, and then the maximum value max3 of the subarea 3 is calculated; because the direction of the T wave is downward, the left part is monotonously decreased, and the right part is monotonously increased, at the moment, max1> =max2 and max4> =max3 need to be satisfied, namely, the ending point and the starting point of the T wave are respectively the maximum value of the T wave front and the rear of the T wave, otherwise, when the direction of the T wave is upward, the same applies;
rule 6: the relationship between the maximum value of the QRS complex and the maximum value of the T complex of the heart beat is satisfied;
When the T wave direction is downward, the minimum value of the T wave group is smaller than the minimum value of the QRS wave group, and the maximum value of the T wave group is smaller than the maximum value of the QRS wave group; the maximum value of the T wave group is larger than the maximum value of the QRS wave group and the minimum value of the T wave group is larger than the minimum value of the QRS wave group when the T wave direction is upward,
Rule 7: firstly, repositioning R waves aiming at heart beats with downward main wave direction and R wave peak amplitude smaller than 0: calculating the maximum point of the intermediate region of the Q wave and the S wave to be used as the peak of the R wave;
[1] For heart beats with the main wave direction downward and the amplitude value of the R wave top point being more than or equal to 0, calculating the difference h 1 between the R wave top point and the Q wave top point amplitude value, the difference h 2 between the R wave top point and the S wave top point amplitude value, the width w from the R wave to the QRS wave end point, meeting the condition that the value of h 1/h2 w is less than or equal to the set threshold value TH 1,
[2] Heart beat with downward main wave direction and R wave peak amplitude less than 0:
a) Excluding heart beats with R waves larger than Q waves and S wave amplitudes;
b) The voltage fluctuations in the region from the QRS wave end point to the T wave start point, in region 1 and in region 4 of the T wave group are smaller: calculating the maximum value and the minimum value of the voltage of each region, then making a difference with the voltage values of the starting point and the ending point of the corresponding region, if the region monotonically increases, making a difference with the voltage of the minimum value, and if the region monotonically decreases, on the contrary, satisfying the difference being smaller than the set threshold value TH 2;
[3] for heart beat with main wave direction upward:
a) The difference between the amplitude of the termination point of the QRS and the amplitude of the S wave is smaller than or equal to a set threshold value TH 3;
b) The voltage fluctuations in the region from the QRS wave end point to the T wave start point, in region 1 and in region 4 of the T wave group are smaller: calculating the maximum value and the minimum value of the voltage of each region, then performing difference with the voltage values of the starting point and the ending point of the corresponding region, if the region monotonically increases, performing difference with the voltage of the starting point of the region and the minimum value, and if the region monotonically decreases inversely, satisfying the condition that the difference is smaller than the set threshold TH 4;
Rule 8: heart beat cosine similarity: the transverse index is from the QRS wave group starting point to the T wave group ending point of the superimposed wave, the longitudinal direction is that the main wave vertexes of the current heart beat and the superimposed wave are aligned, when the main wave direction is upward, the R wave vertexes are aligned, and when the main wave direction is downward, the S wave vertexes are aligned;
a) For the main wave direction upward heart beat: satisfying that the cosine similarity is smaller than a set threshold value TH 5;
b) Downward heart beat for main wave direction: satisfying that the cosine similarity is smaller than a set threshold value TH 6;
Rule 9: for the main wave direction upward heart beat:
a) The R wave amplitude of the current heart beat is larger than that of the superimposed wave;
b) The R-wave amplitude of the current beat is greater than the maximum amplitude of the ST segment.
2. The method for identifying ventricular premature beats with respect to single lead electrocardiographic acquisition by dry electrode according to claim 1, characterized in that: the slope characteristic group in the fourth step is as follows:
Sum of R-wave vertex of current heart beat and slope of the previous 7 points (sum 1);
sum of R-wave peak and slope of the previous 7 points of the superimposed wave (sum 2);
sum of slopes of the R-wave vertex of the current beat and 7 points later (sum 3);
Sum of R-wave peak of superimposed wave and slope of 7 points later (sum 4);
difference, ratio and percentage of difference between sum1 and sum 2;
Difference, ratio, percent difference between sum3 and sum 4.
3. The method for identifying ventricular premature beats with respect to single lead electrocardiographic acquisition by dry electrode according to claim 1, characterized in that: the height feature group in the fourth step is as follows:
15 points of R wave of the current heart beat and the superimposed wave are correspondingly calculated as the sum of absolute values of differences (sum 5); the 15 points of the R wave are the R wave peak and 7 points in front and back;
Sum of absolute values of 15 points of R wave of the current heart beat (sum 6);
sum of absolute values of 201 points of the current heart beat (sum 7);
sum of absolute values of 15 points of the R wave of the superimposed wave (sum 8);
sum of 201 points absolute values of the superimposed wave (sum 9);
Ratio R1 of sum6 to sum 7;
ratio R2 of sum8 to sum 9;
difference, ratio and percentage of difference between R1 and R2;
The sum of absolute difference values (sum 10) is correspondingly made for each 201 points of the current heart beat and the superimposed wave;
Ratio of sum10 to sum 9;
ratio of sum5 to sum 8.
4. The method for identifying ventricular premature beats with respect to single lead electrocardiographic acquisition by dry electrode according to claim 1, characterized in that: the width feature group in the fourth step is as follows:
R wavefront: drawing 7 transverse lines by 7 points in front of the R wave vertex of the current heart beat, respectively calculating the sum of the differences between 7 transverse indexes of the intersection point of the current heart beat and the R wave vertex index of the current heart beat as R wave front width (w 1) of the superimposed wave, and taking the sum of the differences between 7 transverse indexes of the current heart beat and the intersection point index of the superimposed wave as R wave front width difference characteristic (w 2);
R wave: and similarly, 7 transverse lines are stippled by 7 points behind the R wave vertex of the current heart beat to obtain the R wave width (w 3) and the R wave width difference value characteristic (w 4) of the superimposed wave.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211122514.5A CN115486854B (en) | 2022-09-15 | 2022-09-15 | Single-lead electrocardiograph ventricular premature beat identification method for dry electrode acquisition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211122514.5A CN115486854B (en) | 2022-09-15 | 2022-09-15 | Single-lead electrocardiograph ventricular premature beat identification method for dry electrode acquisition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115486854A CN115486854A (en) | 2022-12-20 |
| CN115486854B true CN115486854B (en) | 2024-04-30 |
Family
ID=84467629
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211122514.5A Active CN115486854B (en) | 2022-09-15 | 2022-09-15 | Single-lead electrocardiograph ventricular premature beat identification method for dry electrode acquisition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115486854B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116570295B (en) * | 2023-07-14 | 2024-04-30 | 浙江好络维医疗技术有限公司 | Electrocardiogram low-voltage T-wave end point positioning method |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20120116213A (en) * | 2011-04-12 | 2012-10-22 | 부산대학교 산학협력단 | System for discriminating premature ventricular contraction, method for discriminating premature ventricular contraction and the recording media storing the program performing the said method |
| CN106063704A (en) * | 2016-05-27 | 2016-11-02 | 浙江好络维医疗技术有限公司 | A kind of QRS wave starting and terminal point localization method based on regularization least square recurrence learning |
| CN108937915A (en) * | 2018-07-24 | 2018-12-07 | 杭州质子科技有限公司 | The recognition methods of premature beat in a kind of dynamic ECG |
| CN111166325A (en) * | 2020-03-05 | 2020-05-19 | 齐鲁工业大学 | Electrocardiosignal QRS complex wave detection method and system based on IPCMM algorithm |
| CN111493858A (en) * | 2020-03-16 | 2020-08-07 | 浙江好络维医疗技术有限公司 | Single-lead specific main wave identification and positioning method based on cluster analysis |
| CN114052741A (en) * | 2021-11-08 | 2022-02-18 | 西安工业大学 | Electrocardiogram recognition method based on multi-scale autoregressive model |
| WO2022052300A1 (en) * | 2020-09-08 | 2022-03-17 | 浙江大学山东工业技术研究院 | Electrocardio parameter calculation method based on deep learning |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6944495B2 (en) * | 2000-11-10 | 2005-09-13 | C.R. Bard, Inc. | Methods for processing electrocardiac signals having superimposed complexes |
| US8909332B2 (en) * | 2010-01-26 | 2014-12-09 | Stmicroelectronics S.R.L. | Method and device for estimating morphological features of heart beats |
-
2022
- 2022-09-15 CN CN202211122514.5A patent/CN115486854B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20120116213A (en) * | 2011-04-12 | 2012-10-22 | 부산대학교 산학협력단 | System for discriminating premature ventricular contraction, method for discriminating premature ventricular contraction and the recording media storing the program performing the said method |
| CN106063704A (en) * | 2016-05-27 | 2016-11-02 | 浙江好络维医疗技术有限公司 | A kind of QRS wave starting and terminal point localization method based on regularization least square recurrence learning |
| CN108937915A (en) * | 2018-07-24 | 2018-12-07 | 杭州质子科技有限公司 | The recognition methods of premature beat in a kind of dynamic ECG |
| CN111166325A (en) * | 2020-03-05 | 2020-05-19 | 齐鲁工业大学 | Electrocardiosignal QRS complex wave detection method and system based on IPCMM algorithm |
| CN111493858A (en) * | 2020-03-16 | 2020-08-07 | 浙江好络维医疗技术有限公司 | Single-lead specific main wave identification and positioning method based on cluster analysis |
| WO2022052300A1 (en) * | 2020-09-08 | 2022-03-17 | 浙江大学山东工业技术研究院 | Electrocardio parameter calculation method based on deep learning |
| CN114052741A (en) * | 2021-11-08 | 2022-02-18 | 西安工业大学 | Electrocardiogram recognition method based on multi-scale autoregressive model |
Non-Patent Citations (3)
| Title |
|---|
| An ECG Processing and Analysis Technique Based on Neural Network;Youkubjon Talatov等;2020 International Conference on Information Science and Communications Technologies (ICISCT);20200119;第1-4页 * |
| 基于小波变换和BP神经网络的室性早搏(PVC)识别;熊平;徐敏;雷英俊;;中国医学物理学杂志;20100315(第02期);第68-71页 * |
| 基于深度学习的ECG心拍数据分类设计;张俊飞;毕志升;王静;吴小玲;;自动化与仪器仪表;20191225(第12期);第77-81页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115486854A (en) | 2022-12-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kathirvel et al. | An efficient R-peak detection based on new nonlinear transformation and first-order Gaussian differentiator | |
| de Chazal et al. | Automatic classification of ECG beats using waveform shape and heart beat interval features | |
| Sasikala et al. | Extraction of P wave and T wave in Electrocardiogram using Wavelet Transform | |
| Andreao et al. | ST-segment analysis using hidden Markov Model beat segmentation: application to ischemia detection | |
| CN115486855B (en) | Electrocardiogram and heart beat classification method based on QRS complex indefinite cycle superposition | |
| CN113499079B (en) | Atrial fibrillation detection method in electrocardiogram | |
| Manriquez et al. | An algorithm for QRS onset and offset detection in single lead electrocardiogram records | |
| CN113057648A (en) | ECG signal classification method based on composite LSTM structure | |
| CN112932498B (en) | T waveform state classification system with generalization capability based on deep learning | |
| CN115486854B (en) | Single-lead electrocardiograph ventricular premature beat identification method for dry electrode acquisition | |
| CN105877739A (en) | Clinical examination method of electrocardio intelligent analyzing system | |
| CN110090016B (en) | Method and system for positioning R wave position and R wave automatic detection method using LSTM neural network | |
| Kar et al. | A technical review on statistical feature extraction of ECG signal | |
| Senapati et al. | Cardiac arrhythmia classification of ECG signal using morphology and heart beat rate | |
| CN111528833B (en) | Rapid identification and processing method and system for electrocardiosignals | |
| Bakhshipour et al. | Myocardial ischemia detection with ECG analysis, using Wavelet Transform and Support Vector Machines | |
| CN108888263B (en) | R wave detection method based on geometric form group characteristics | |
| CN112438735B (en) | Electrocardiogram P wave detection method, analysis device and storage medium | |
| Khandait et al. | Efficient ECG abnormalities recognition using neuro-fuzzy approach | |
| Sahoo et al. | Adaptive thresholding based EMD for delineation of QRS complex in ECG signal analysis | |
| CN115177268A (en) | Atrial fibrillation heart beat detection method, device and storage medium | |
| CN111543977B (en) | Multi-cascade artificial intelligence vagina discharge method based on 12-lead resting electrocardiogram | |
| Mondelo et al. | A channel-dependent algorithm for heart beats detection in ECG recordings | |
| Sheikh et al. | Cardiac disorder diagnosis based on ECG segments analysis and classification | |
| Bhaskar et al. | Probabilistic Neural Network for the Automatic Detection of QRS-complexes in ECG using Slope |
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 |