CN113712525A - Physiological parameter processing method and device and medical equipment - Google Patents
Physiological parameter processing method and device and medical equipment Download PDFInfo
- Publication number
- CN113712525A CN113712525A CN202010436022.8A CN202010436022A CN113712525A CN 113712525 A CN113712525 A CN 113712525A CN 202010436022 A CN202010436022 A CN 202010436022A CN 113712525 A CN113712525 A CN 113712525A
- Authority
- CN
- China
- Prior art keywords
- heart beat
- heart
- classified
- time interval
- preset time
- 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
- 238000003672 processing method Methods 0.000 title abstract description 13
- 238000013528 artificial neural network Methods 0.000 claims abstract description 39
- 238000001514 detection method Methods 0.000 claims abstract description 34
- 238000000034 method Methods 0.000 claims abstract description 30
- 238000004458 analytical method Methods 0.000 claims abstract description 11
- 230000006870 function Effects 0.000 claims description 46
- 238000012545 processing Methods 0.000 claims description 35
- 238000012549 training Methods 0.000 claims description 27
- 210000002569 neuron Anatomy 0.000 claims description 16
- 238000007781 pre-processing Methods 0.000 claims description 12
- 230000001419 dependent effect Effects 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 3
- 230000033764 rhythmic process Effects 0.000 abstract description 2
- 206010003119 arrhythmia Diseases 0.000 description 7
- 230000006793 arrhythmia Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 230000002861 ventricular Effects 0.000 description 5
- 238000007621 cluster analysis Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 210000003903 pelvic floor Anatomy 0.000 description 2
- 206010015856 Extrasystoles Diseases 0.000 description 1
- 208000000418 Premature Cardiac Complexes Diseases 0.000 description 1
- 206010047281 Ventricular arrhythmia Diseases 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000747 cardiac effect Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000010606 normalization Methods 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
- A61B5/02—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
- A61B5/02—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
- A61B5/0245—Detecting, measuring or recording pulse rate or heart rate by using sensing means generating electric signals, i.e. ECG signals
-
- 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
- A61B5/7246—Details of waveform analysis using correlation, e.g. template matching or determination of similarity
-
- 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
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Cardiology (AREA)
- Molecular Biology (AREA)
- Animal Behavior & Ethology (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Veterinary Medicine (AREA)
- Surgery (AREA)
- Physiology (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Signal Processing (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Psychiatry (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
The application provides a physiological parameter processing method, which comprises the following steps: detecting the preprocessed dynamic electrocardiogram data to obtain heart beats in the preprocessed dynamic electrocardiogram data; using template matching to perform clustering analysis on all heartbeats to obtain a clustering result; and carrying out artifact detection on the clustering result, determining the heart beat to be classified, and classifying the heart beat to be classified by using a neural network. The application also provides a corresponding device and medical equipment. The method and the device can effectively improve the accuracy of heart rhythm classification.
Description
Technical Field
The disclosed embodiments of the present application relate to the technical field of biomedical signal processing, and more particularly, to a physiological parameter processing method, apparatus and medical device.
Background
The arrhythmia detection method is usually electrocardiographic examination, and then the arrhythmia condition is diagnosed artificially according to the electrocardiographic examination.
However, there are challenges with dynamic electrocardiograms in the automated analysis and detection of arrhythmias. The dynamic electrocardiogram has long acquisition time, and a user can carry out daily activities, so that signals are easily influenced by noises such as motion artifacts, baseline drift, power frequency interference and the like, particularly the motion artifacts can interfere with heart beat detection and arrhythmia identification, and the dynamic electrocardiogram has large data volume and long analysis time. In addition, the ventricular arrhythmia waveform has complex and variable shapes and individual difference.
Disclosure of Invention
According to an embodiment of the present application, a method, an apparatus and a medical device for processing physiological parameters are provided to solve the above problems.
According to a first aspect of the present application, an exemplary cardiac rhythm data processing method is disclosed, comprising: detecting the preprocessed dynamic electrocardiogram data to obtain heart beats in the preprocessed dynamic electrocardiogram data; using template matching to perform clustering analysis on all heartbeats to obtain a clustering result; and carrying out artifact detection on the clustering result, determining the heart beat to be classified, and classifying the heart beat to be classified by using a neural network.
In some embodiments, the clustering result comprises a template to which each heart beat belongs and a heart beat number to which the template belongs, wherein an inverse of the heart beat number represents a matching degree of the heart beat.
In some embodiments, the performing artifact detection on the clustering result and determining a heartbeat to be classified includes: acquiring heartbeat data in a preset time interval; determining heart beat matching degree in the preset time interval, wherein the heart beat matching degree in the preset time interval is the sum of the matching degrees of each heart beat in the preset time interval; and if the heart beat matching degree in the preset time interval meets a preset value, determining the heart beats in the preset time interval as heart beats to be classified.
In some embodiments, further comprising: constructing a neural network classifier, wherein the neural network classifier comprises an input layer, a hidden layer and an output layer, input variables of the input layer comprise QRS wave width, a ratio of a current RR interval to a previous RR interval, T wave direction, amplitude difference and width difference of heart beats to be classified and template heart beats, the number of neurons in the hidden layer is 3, and the number of neurons in the output layer is 2; inputting the input variables after fuzzy processing into the input layer to execute network training.
In some embodiments, the fuzzy processing employs a fuzzy membership function comprising at least one of a pi-type function, a trapezoidal function, an S-type function, and a gaussian function.
In some embodiments, the classifying the heartbeat to be classified using a neural network comprises: extracting features of the heart beat to be classified, wherein the extracted features comprise QRS wave width, a ratio of a current RR interval to a previous RR interval, T wave direction, amplitude difference and width difference of the heart beat to be classified and a template heart beat; and carrying out fuzzy processing on the extracted features, and inputting the features into the neural network classifier as input variables to obtain a classification result.
According to a second aspect of the present application, an exemplary physiological parameter processing device is disclosed, comprising: the preprocessing module is used for acquiring dynamic electrocardiogram data and preprocessing the dynamic electrocardiogram data; the heart beat detection module is used for detecting the preprocessed dynamic electrocardiogram data to obtain heart beats in the preprocessed dynamic electrocardiogram data; the heartbeat clustering module is used for carrying out clustering analysis on all heartbeats by using template matching so as to obtain a clustering result; the artifact detection module is used for carrying out artifact detection on the clustering result and determining heart beats to be classified; and a heartbeat classification module for classifying the heartbeats to be classified using a neural network.
In some embodiments, the clustering result comprises a template to which each heart beat belongs and a heart beat number to which the template belongs, wherein an inverse of the heart beat number represents a matching degree of the heart beat; the artifact detection module is specifically configured to: acquiring heartbeat data in a preset time interval; determining heart beat matching degree in the preset time interval, wherein the heart beat matching degree in the preset time interval is the sum of the matching degrees of each heart beat in the preset time interval; and if the heart beat matching degree in the preset time interval meets a preset value, determining the heart beats in the preset time interval as heart beats to be classified.
According to a third aspect of the present application, an example medical device is disclosed, comprising a processor and a memory connected to the processor, the memory storing instructions that, when executed, cause the processor to perform the method as in the second aspect above.
According to a fourth aspect of the present application, an exemplary non-volatile storage medium is disclosed that stores instructions that, when executed, cause the processor to perform the method as in the second aspect above.
The beneficial effect of this application has: the heart beats in the dynamic electrocardiogram data are obtained by preprocessing and detecting the dynamic electrocardiogram data, then all the heart beats are subjected to clustering analysis by using template matching to obtain a clustering result, then the heart beats to be classified are determined by performing artifact detection on the clustering result, the heart beats to be classified are classified by using a neural network, and the heart rate classification accuracy is improved.
Drawings
The present application will be further described with reference to the accompanying drawings and embodiments, in which:
fig. 1 is a flowchart of a physiological parameter processing method according to an embodiment of the present application.
Fig. 2 is a partial flowchart of a physiological parameter processing method according to an embodiment of the present application.
Fig. 3 is a partial flowchart of a physiological parameter processing method according to an embodiment of the present application.
Fig. 4 is a schematic structural diagram of a physiological parameter processing device according to an embodiment of the present application.
Fig. 5 is a schematic structural diagram of a medical device according to an embodiment of the present application.
Fig. 6 is another schematic structural diagram of the medical device according to the embodiment of the present application.
Detailed Description
In order to make those skilled in the art better understand the technical solutions of the present application, the following detailed description is made with reference to the accompanying drawings and the detailed description.
Fig. 1 is a flowchart of a physiological parameter processing method according to an embodiment of the present application. The method may be performed by a medical device, wherein the physiological parameter comprises heart rate data. The medical equipment can be equipment with a dynamic electrocardiogram acquisition function, such as a holter, a wearable dynamic electrocardiogram recorder and the like, or can be independent data processing equipment, such as a mobile terminal, a computer and the like, and the method comprises the following steps:
step 110: and acquiring dynamic electrocardiogram data and preprocessing the dynamic electrocardiogram data.
The dynamic electrocardiogram data can be acquired by wearing a dynamic electrocardiogram acquisition box, or the dynamic electrocardiogram data can directly use the data in the MIT-BIH arrhythmia standard database. The dynamic electrocardiogram data is preprocessed, and the influence of noise and artifacts on the dynamic electrocardiogram data, such as power frequency noise, baseline drift, electromyographic noise and the like, is removed.
The pre-processing may include filtering, such as using a basic digital filter, or may use a wavelet adaptive thresholding approach.
Step 120: and detecting the preprocessed dynamic electrocardiogram data to obtain the heart beat in the preprocessed dynamic electrocardiogram data.
And detecting the preprocessed dynamic electrocardiogram data to obtain heart beats in the dynamic electrocardiogram data, namely identifying the heart beats in the dynamic electrocardiogram data, wherein the QRS wave is the main component of the electrocardio signal, so that the detection comprises the identification and the positioning of the QRS wave. In some embodiments, methods of QRS wave detection include differential thresholding, wavelet modulus maxima, and the like.
Step 130: using template matching, a clustering analysis is performed on all heartbeats to obtain a clustering result.
Using template matching to perform cluster analysis on all heartbeats, specifically, initializing a plurality of heart beat templates, calculating a similarity index between each heart beat and a certain heart beat template, if the similarity index meets a threshold requirement, namely is greater than or equal to a preset threshold, matching the heart beat and the heart beat template, and if the similarity index does not meet the threshold requirement, newly building a template or updating the template until all heart beats are analyzed.
The clustering result comprises the template to which each heart beat belongs and the heart beat number of the dependent template. The reciprocal of the number of heart beats indicates the degree of matching of the heart beat. For example, assuming that the number of heart beats is n, 1/n is the matching degree of the heart beats, which may be expressed in percentage form in some embodiments.
Generally speaking, a dominant quantity of a dynamic electrocardiogram is sinus heartbeats, that is, a section of heartbeats in a 24-hour electrocardiogram signal is normally sinus heartbeats, so after heart beats are clustered, a similarity index of the heart beats can be obtained, and the heart beats with a large quantity are directly classified as normal heart beats, that is, the heart beat with the largest number of heart beats of the dependent template in the clustering result is directly normal sinus heartbeats, so that the subsequent processing of a part of heartbeats can be reduced, the computation amount is reduced, and the computation efficiency is improved.
Step 140: and performing artifact detection on the clustering result, and determining the heart beat to be classified so as to classify the heart beat to be classified by using a neural network.
And (3) carrying out artifact detection on the clustering result, not carrying out subsequent classification on the heart beats marked as artifacts, determining the heart beats to be classified by the heart beats not marked as artifacts, and further classifying the heart beats to be classified by using a neural network. Before the classification of the heart beat, the pseudo-error detection is carried out, the influence of interference on the classification of the heart beat is eliminated, the heart beat with poor signal quality is reduced and is misjudged as the ventricular heart beat, and the classification accuracy is indirectly improved.
In the embodiment, the heart beats in the dynamic electrocardiogram data are obtained by preprocessing and detecting the dynamic electrocardiogram data, then the template matching is used for carrying out cluster analysis on all the heart beats to obtain a cluster result, then the pseudo-error detection is carried out on the cluster result to determine the heart beats to be classified, and the neural network is used for classifying the heart beats to be classified, so that the heart rate classification accuracy is improved.
In some embodiments, as shown in fig. 2, in step 140, performing artifact detection on the clustering result and determining the heart beat to be classified includes:
step 141: acquiring heartbeat data in a preset time interval.
The preset time interval may be a time period before and after the current heart beat, for example, a time period of 4 seconds before and after the current heart beat.
Step 142: and determining the heart beat matching degree in a preset time interval.
The matching degree of the heart beats in the preset time interval is the sum of the matching degrees of each heart beat in the preset time interval.
As described above, the reciprocal of the heart beat number in the clustering result represents the matching degree of the heart beats, that is, the reciprocal of the heart beat number in the clustering result represents the matching degree of a single heart beat.
Step 143: and if the heart beat matching degree in the preset time interval meets the preset value, determining the heart beats in the preset time interval as the heart beats to be classified.
If the heart beat matching degree in the preset time interval does not meet the preset value, the heart beat data in the preset time interval is marked as pseudo-error, subsequent classification is not carried out, if the heart beat matching degree in the preset time interval meets the preset value, the heart beat data in the preset time interval is determined as the heart beat to be classified, therefore, the interference of some noises can be eliminated, and some heart beats with poor signal quality are mistakenly judged as ventricular heart beats.
In the embodiment, the heart beat matching degree in the preset time interval is the sum of the matching degrees of all heart beats in the preset time interval, and the heart beat matching degree in the preset time interval meets the preset value, so that the heart beat data in the preset time interval is determined as the heart beats to be classified, interference of some noises can be eliminated, and misjudgment of some heart beats with poor signal quality as ventricular heart beats is reduced.
After the heart beat to be classified is determined, the neural network is used for classifying the heart beat to be classified, and the accuracy rate of detecting the heart beat is improved. As shown in fig. 3, in some embodiments, the method further comprises:
step 150: and constructing a neural network classifier.
The neural network classifier comprises an input layer, a hidden layer and an output layer.
The input variables of the input layer comprise one or more of QRS wave width, the ratio of the current RR interval to the previous RR interval, T wave direction, amplitude difference and width difference between heart beat to be classified and template heart beat, the number of neurons in the hidden layer is 3, and the number of neurons in the output layer is 2.
The QRS wave is generally wider than the normal sinus heartbeat when the ventricular heartbeat occurs, the direction of the T wave is opposite to the main wave of the normal sinus heartbeat, and some premature beats occur. In addition, the difference of the amplitude and the width of the heart beat to be classified and the template heart beat can reflect the morphological difference between the heart beats. The template heart beat may be a normal sinus heart beat, which is detected by the above step 130, using template matching.
The input variables of the input layer are extracted by using time domain features and a time domain manner, and the calculated amount is small, of course, the input variables of the input layer may also be extracted by using frequency domain features and a transform domain manner, for example, the frequency domain feature extraction of FFT, the wavelet transform method extraction of wavelet components, and the calculation of information entropy, etc., which is not limited in this application.
The selection of the number of hidden layer neurons is relatively complex, based on N ═ N (N)in+max(Nout,NClass) 2 where N represents the number of hidden layer neurons, NinRepresenting the number of input variables, NoutRepresenting the number of outputs, NClassIndicating the number of classifications. The neural network in this application accomplishes two classification problems, the number of hidden layer neurons N is approximately 3, based on the number of input variables mentioned above.
Since the neural network of the present application performs a two classification problem, the number of output layer neurons is 2.
Step 160: and inputting the input variable subjected to the fuzzy processing into an input layer to perform network training.
After the neural network classifier is constructed, input variables after fuzzy processing are input into an input layer, network training is executed, and network related parameters are determined.
It should be noted that steps 150 and 160 in the present embodiment and the steps in the above embodiments are not necessarily performed in the numerical order, for example, step 150 may be performed simultaneously with step 130.
In the embodiment, the difference between the heart beat and the template can be better reflected by taking the amplitude difference and the width difference between the heart beat to be classified and the template heart beat as input variables, so that the arrhythmia classification accuracy is further improved.
In some embodiments, the fuzzy processing uses a fuzzy membership function, the fuzzy membership function comprising at least one of a pi-type function, a trapezoidal function, an S-type function, and a gaussian function.
Each input data of the network usually has different physical meanings and different dimensions, and after the input variables are converted into decimal numbers in the range of [0,1], the difference of each component caused by different dimensions can be reduced, so that each component has the same important position. In the embodiment, the fuzzy membership function is adopted to fuzzify the input variable, so that the method has more practical significance by simulating the discrimination behavior of a doctor on the electrocardiogram based on experience, simplifies the differential influence of the input variable on the classification model caused by different dimensions, and further improves the accuracy of the heart rate component. In other embodiments, the fuzzy processing may be implemented by using a normalization method.
In the embodiment, the neural network classifier is constructed, so that the neural network is used for classifying the heart beats to be classified, and the accuracy of heart beat detection is improved.
The following describes the process of network training. Specifically, in step 160, in the process of performing network training, first, the weight vector from the input layer to the hidden layer and the hidden layer to the output are initializedThe method comprises the steps of layer weight vector, error function e, calculation precision value and maximum training times M, wherein the weight vector from an input layer to a hidden layer and the weight vector from the hidden layer to an output layer are random numbers in (0,1), a normal heart beat and at least one type of ventricular heart beat are selected as training samples, a random method is selected for selection of the training samples, and input marks are X0,X1,…,X5Target output is d0，d1。
Then, each training sample is input for training, and the actual output Y of each layer in the network is calculatedi＝f(WijXij+θi)，WijIs a weight between layers, θiAs an offset value, in the present embodiment, θiIs set to 0.
The weights are then modified according to the actual output of each layer to continue training, specifically, each weight is modified starting from the output node to the hidden layer according to the following formula.
Wij(t+1)＝Wij(t)+ηδjXij+a(Wij(t)-Wij(t-1))
Where η is the learning speed, and a is the inertia coefficient value (0, 1).
When j is an output layer node, the back propagation error is as follows:
δj＝Yi(1-Yi)(dj-Yi)。
and then, after training all the training samples, determining errors, and after training all the training samples, indicating that one training period is finished, wherein the errors are as follows:
and finally, if the error is less than or equal to the initialized error function e or the maximum training is reached, ending the training or starting the next training period until the training is finished, thereby determining the network related parameters.
Classifying the heartbeat to be classified using a neural network based on the network-related parameters determined by the training completion, in some embodiments classifying the heartbeat to be classified using the neural network comprises:
firstly, extracting features of a heart beat to be classified, wherein the extracted features comprise QRS wave width, a ratio of a current RR interval to a previous RR interval, T wave direction, amplitude difference and width difference between the heart beat to be classified and a template heart beat.
The template heart beat may be a normal sinus heart beat, which is detected by the above step 130, using template matching.
And then, carrying out fuzzy processing on the extracted features, and inputting the features serving as input variables into a neural network classifier to obtain a classification result.
In some embodiments, the fuzzy processing uses a fuzzy membership function, the fuzzy membership function comprising at least one of a pi-type function, a trapezoidal function, an S-type function, and a gaussian function.
As shown in fig. 4, which is a schematic structural diagram of a physiological parameter processing apparatus according to an embodiment of the present application, the apparatus 400 includes a preprocessing module 410, a heartbeat detection module 420, a heartbeat clustering module 430, a artifact detection module 440, and a heartbeat classification module 450.
The preprocessing module 410 is used for acquiring the dynamic electrocardiogram data and preprocessing the dynamic electrocardiogram data.
And a heart beat detection module 420 for detecting the preprocessed dynamic electrocardiogram data to obtain heart beats therein.
The heartbeat clustering module 430 is configured to perform cluster analysis on all heartbeats using template matching to obtain a clustering result.
The artifact detection module 440 is configured to perform artifact detection on the clustering result to determine a heart beat to be classified.
The heart beat classification module 450 is used to classify the heart beats to be classified using a neural network.
In the embodiment, the heart beats in the dynamic electrocardiogram data are obtained by preprocessing and detecting the dynamic electrocardiogram data, then the template matching is used for carrying out cluster analysis on all the heart beats to obtain a cluster result, then the pseudo-error detection is carried out on the cluster result to determine the heart beats to be classified, and the neural network is used for classifying the heart beats to be classified, so that the heart rate classification accuracy is improved.
The heart rate data processing device in the present embodiment implements the heart rate data processing method provided in any one of the above embodiments of the present application and any non-conflicting combination. For details of the heart rate data processing method, the description of the above embodiments is given, and the description is omitted here.
In some embodiments, the clustering result comprises a template to which each heart beat belongs and the heart beat number of the dependent template, wherein the reciprocal of the heart beat number represents the matching degree of the heart beats;
the artifact detection module 440 is specifically configured to:
acquiring heartbeat data in a preset time interval;
determining heart beat matching degree in a preset time interval, wherein the heart beat matching degree in the preset time interval is the sum of the matching degrees of each heart beat in the preset time interval;
and if the heart beat matching degree in the preset time interval meets the preset value, determining the heart beats in the preset time interval as the heart beats to be classified.
In some embodiments, heart beat classification module 450 is further to:
constructing a neural network classifier, wherein input variables of an input layer comprise QRS wave width, a ratio of a current RR interval to a previous RR interval, a T wave direction, an amplitude difference and a width difference between a heart beat to be classified and a template heart beat, the number of neurons in an implicit layer is 3, and the number of neurons in an output layer is 2;
and inputting the input variable subjected to the fuzzy processing into an input layer to perform network training.
In some embodiments, the fuzzy processing uses a fuzzy membership function, the fuzzy membership function comprising at least one of a pi-type function, a trapezoidal function, an S-type function, and a gaussian function.
In some embodiments, heart beat classification module 450 is specifically configured to:
extracting features of the heart beat to be classified, wherein the extracted features comprise QRS wave width, the ratio of the current RR interval to the previous RR interval, T wave direction, amplitude difference and width difference between the heart beat to be classified and the template heart beat;
and carrying out fuzzy processing on the extracted features, and inputting the features into the neural network classifier as input variables to obtain a classification result.
Fig. 5 is a schematic structural diagram of a medical apparatus according to an embodiment of the present application. The medical equipment can be equipment with a dynamic electrocardiogram acquisition function, such as a holter, a wearable dynamic electrocardiogram recorder and the like. The medical device 500 includes a processor 510, a dynamic electrocardiograph collection cartridge 520 and a memory 530, the dynamic electrocardiograph collection cartridge 520 and the memory 530 being coupled to the processor 510.
The dynamic electrocardiogram collecting box 520 is used for collecting dynamic electrocardiogram data.
The memory 530 is used for storing the electrocardiogram dynamic data and the neural network related parameters in the above embodiments. Memory 530 may include read-only memory and/or random access memory, etc., and provides instructions and data to processor 510. A portion of the memory 530 may also include non-volatile random access memory (NVRAM).
The memory 530 stores instructions that, when executed, enable the processor 510 to implement the heart rate data processing method provided by any one of the above embodiments of the present application, and any non-conflicting combinations thereof, via the dynamic ecg collection box 520. For details of the heart rate data processing method, the description of the above embodiments is given, and the description is omitted here.
In particular, processor 510 is configured to:
detecting the preprocessed dynamic electrocardiogram data to obtain heart beats in the preprocessed dynamic electrocardiogram data;
using template matching to perform clustering analysis on all heartbeats to obtain a clustering result;
and performing artifact detection on the clustering result, and determining the heart beat to be classified so as to classify the heart beat to be classified by using a neural network.
In some embodiments, the clustering result includes a template to which each heart beat belongs and a heart beat number of the dependent template, wherein an inverse of the heart beat number represents a matching degree of the heart beats. Processor 510 is specifically configured to:
acquiring heartbeat data in a preset time interval;
determining heart beat matching degree in a preset time interval, wherein the heart beat matching degree in the preset time interval is the sum of the matching degrees of each heart beat in the preset time interval;
and if the heart beat matching degree in the preset time interval meets the preset value, determining the heart beats in the preset time interval as the heart beats to be classified.
In some embodiments, processor 510 is further specifically configured to:
constructing a neural network classifier, wherein input variables of an input layer comprise QRS wave width, a ratio of a current RR interval to a previous RR interval, a T wave direction, an amplitude difference and a width difference between a heart beat to be classified and a template heart beat, the number of neurons in an implicit layer is 3, and the number of neurons in an output layer is 2;
and inputting the input variable subjected to the fuzzy processing into an input layer to perform network training.
In some embodiments, the fuzzy processing uses a fuzzy membership function, the fuzzy membership function comprising at least one of a pi-type function, a trapezoidal function, an S-type function, and a gaussian function.
In some embodiments, processor 510 is specifically configured to:
extracting features of the heart beat to be classified, wherein the extracted features comprise one or more of QRS wave width, the ratio of a current RR interval to a previous RR interval, T wave direction, amplitude difference and width difference of the heart beat to be classified and the template heart beat;
and carrying out fuzzy processing on the extracted features, and inputting the features into the neural network classifier as input variables to obtain a classification result.
As shown in fig. 6, which is another schematic structural diagram of the medical device according to the embodiment of the present application, the medical device 600 may be a separate data processing device, such as a mobile terminal, a computer, etc. like a mobile phone, and is connected to a dynamic electrocardiograph collection box, such as the dynamic electrocardiograph collection box 520 of the above embodiment, or may not be connected to the dynamic electrocardiograph collection box and use data in the MIT-BIH arrhythmia standard database. The medical device 600 is used to process heart rate data.
The medical device 600 includes a memory 610, a processor 620, and a communication circuit 630. The memory 610 is coupled to the processor 620.
The communication circuit 630 is used for transmitting and receiving data, and is an interface for the terminal device 300 to communicate with an external device.
The processor 620 executes the parameter setting method of the pelvic floor training apparatus of the above-described embodiment of the present application through the communication circuit 630. In particular, upon execution of the instructions in the memory 610, the processor 620 is configured to:
detecting the preprocessed dynamic electrocardiogram data to obtain heart beats in the preprocessed dynamic electrocardiogram data;
using template matching to perform clustering analysis on all heartbeats to obtain a clustering result;
and performing artifact detection on the clustering result, and determining the heart beat to be classified so as to classify the heart beat to be classified by using a neural network.
In some embodiments, the clustering result includes a template to which each heart beat belongs and a heart beat number of the dependent template, wherein an inverse of the heart beat number represents a matching degree of the heart beats. Processor 620 is specifically configured to:
acquiring heartbeat data in a preset time interval;
determining heart beat matching degree in a preset time interval, wherein the heart beat matching degree in the preset time interval is the sum of the matching degrees of each heart beat in the preset time interval;
and if the heart beat matching degree in the preset time interval meets the preset value, determining the heart beats in the preset time interval as the heart beats to be classified.
In some embodiments, processor 620 is further specifically configured to:
constructing a neural network classifier, wherein input variables of an input layer comprise QRS wave width, a ratio of a current RR interval to a previous RR interval, a T wave direction, an amplitude difference and a width difference between a heart beat to be classified and a template heart beat, the number of neurons in an implicit layer is 3, and the number of neurons in an output layer is 2;
and inputting the input variable subjected to the fuzzy processing into an input layer to perform network training.
In some embodiments, the fuzzy processing uses a fuzzy membership function, the fuzzy membership function comprising at least one of a pi-type function, a trapezoidal function, an S-type function, and a gaussian function.
In some embodiments, processor 620 is specifically configured to:
extracting features of the heart beat to be classified, wherein the extracted features comprise one or more of QRS wave width, the ratio of a current RR interval to a previous RR interval, T wave direction, amplitude difference and width difference of the heart beat to be classified and the template heart beat;
and carrying out fuzzy processing on the extracted features, and inputting the features into the neural network classifier as input variables to obtain a classification result.
It will be apparent to those skilled in the art that many modifications and variations can be made in the devices and methods while maintaining the teachings of the present application. Accordingly, the above disclosure should be considered limited only by the scope of the following claims.
Claims (10)
1. A method of processing a physiological parameter, comprising:
detecting the preprocessed dynamic electrocardiogram data to obtain heart beats in the preprocessed dynamic electrocardiogram data;
using template matching to perform clustering analysis on all heartbeats to obtain a clustering result;
and carrying out artifact detection on the clustering result, determining the heart beat to be classified, and classifying the heart beat to be classified by using a neural network.
2. A method as claimed in claim 1, wherein the clustering result includes a template to which each heart beat is dependent and a number of heart beats dependent from the template, wherein the reciprocal of the number of heart beats represents the degree of matching of the heart beats.
3. The method as claimed in claim 2, wherein said performing artifact detection on said clustering result, determining heart beats to be classified, comprises:
acquiring heartbeat data in a preset time interval;
determining heart beat matching degree in the preset time interval, wherein the heart beat matching degree in the preset time interval is the sum of the matching degrees of each heart beat in the preset time interval;
and if the heart beat matching degree in the preset time interval meets a preset value, determining the heart beats in the preset time interval as heart beats to be classified.
4. The method as recited in claim 1, further comprising:
constructing a neural network classifier, wherein the neural network classifier comprises an input layer, a hidden layer and an output layer, and input variables of the input layer comprise one or more of the following: QRS wave width, the ratio of the current RR interval to the previous RR interval, T wave direction, amplitude difference and width difference between heart beat to be classified and template heart beat; the number of hidden layer neurons is 3, and the number of output layer neurons is 2;
inputting the input variables after fuzzy processing into the input layer to execute network training.
5. The method as recited in claim 4,
the fuzzy processing adopts a fuzzy membership function, and the fuzzy membership function at least comprises one of a pi-type function, a trapezoidal function, an S-type function and a Gaussian function.
6. The method as recited in claim 4, wherein the classifying the heart beat to be classified using a neural network comprises:
performing feature extraction on the heart beat to be classified, wherein the extracted features comprise one or more of the following: QRS wave width, the ratio of the current RR interval to the previous RR interval, T wave direction, amplitude difference and width difference between heart beat to be classified and template heart beat;
and carrying out fuzzy processing on the extracted features, and inputting the features into the neural network classifier as input variables to obtain a classification result.
7. A physiological parameter processing apparatus, comprising:
the preprocessing module is used for acquiring dynamic electrocardiogram data and preprocessing the dynamic electrocardiogram data;
the heart beat detection module is used for detecting the preprocessed dynamic electrocardiogram data to obtain heart beats in the preprocessed dynamic electrocardiogram data;
the heartbeat clustering module is used for carrying out clustering analysis on all heartbeats by using template matching so as to obtain a clustering result;
the artifact detection module is used for carrying out artifact detection on the clustering result and determining heart beats to be classified; and
and the heart beat classification module is used for classifying the heart beats to be classified by using a neural network.
8. A physiological parameter processing apparatus as claimed in claim 7, wherein the clustering result includes a template to which each heart beat is dependent and a heart beat number to which the template is dependent, wherein an inverse of the heart beat number indicates a matching degree of the heart beats;
the artifact detection module is specifically configured to:
acquiring heartbeat data in a preset time interval;
determining heart beat matching degree in the preset time interval, wherein the heart beat matching degree in the preset time interval is the sum of the matching degrees of each heart beat in the preset time interval;
and if the heart beat matching degree in the preset time interval meets a preset value, determining the heart beats in the preset time interval as heart beats to be classified.
9. A medical device comprising a processor and a memory coupled to the processor, the memory storing instructions that, when executed, cause the processor to perform the method of any of claims 1-6.
10. A non-volatile storage medium having stored thereon instructions that, when executed, cause the processor to perform the method of any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010436022.8A CN113712525B (en) | 2020-05-21 | Physiological parameter processing method and device and medical equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010436022.8A CN113712525B (en) | 2020-05-21 | Physiological parameter processing method and device and medical equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113712525A true CN113712525A (en) | 2021-11-30 |
| CN113712525B CN113712525B (en) | 2024-07-12 |
Family
ID=
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115778402A (en) * | 2022-12-02 | 2023-03-14 | 深圳华清心仪医疗电子有限公司 | Method and system for identifying artifact of dynamic electrocardiosignal |
Citations (25)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4336810A (en) * | 1980-09-30 | 1982-06-29 | Del Mar Avionics | Method and apparatus for arrhythmia analysis of ECG recordings |
| WO2008007236A2 (en) * | 2006-06-07 | 2008-01-17 | Koninklijke Philips Electronics N.V. | Atrial fibrillation detection |
| CN102835954A (en) * | 2012-09-07 | 2012-12-26 | 深圳市邦健电子有限公司 | Method for generating heart beating waveform template and treating ventricular premature beat detection and device |
| CN103720468A (en) * | 2013-12-05 | 2014-04-16 | 深圳先进技术研究院 | Artifact identification method and device applied to dynamic electrocardiogram data |
| CN104545887A (en) * | 2014-12-24 | 2015-04-29 | 深圳先进技术研究院 | Method and device for identifying artifact electrocardiograph waveforms |
| CN104586383A (en) * | 2015-01-12 | 2015-05-06 | 北京世纪今科医疗器械有限公司 | Electrocardiogram waveform classification method and device |
| CN105559771A (en) * | 2014-10-09 | 2016-05-11 | 中国移动通信集团公司 | Electrocardiosignal quality detection method and apparatus |
| CN105975385A (en) * | 2016-04-28 | 2016-09-28 | 浪潮(北京)电子信息产业有限公司 | Fuzzy neural network-based virtual machine energy consumption prediction method and system |
| US20160296762A1 (en) * | 2013-10-03 | 2016-10-13 | University Of Washington Through Its Center For Commercialization | Apparatuses and methods for classification of electrocardiogram signals during cardiopulmonary resuscitation |
| CN106108889A (en) * | 2016-07-20 | 2016-11-16 | 杨平 | Electrocardiogram classification method based on degree of depth learning algorithm |
| CN106923820A (en) * | 2017-03-10 | 2017-07-07 | 深圳竹信科技有限公司 | A kind of electrocardiosignal artifact identification method and electrocardiosignal artifact identifying device |
| CN107358196A (en) * | 2017-07-12 | 2017-11-17 | 北京卫嘉高科信息技术有限公司 | A kind of sorting technique of heart beat type, device and electrocardiogram equipment |
| CN107785066A (en) * | 2016-08-26 | 2018-03-09 | 深圳市理邦精密仪器股份有限公司 | Change the method, apparatus and system of heart beat type |
| CN107951485A (en) * | 2017-11-27 | 2018-04-24 | 乐普（北京）医疗器械股份有限公司 | Ambulatory ECG analysis method and apparatus based on artificial intelligence self study |
| CN108257347A (en) * | 2018-01-10 | 2018-07-06 | 安徽大学 | A kind of flame image sequence sorting technique and device using convolutional neural networks |
| CN109303561A (en) * | 2018-11-01 | 2019-02-05 | 杭州质子科技有限公司 | It is a kind of to clap the recognition methods clapped with the abnormal heart based on the artifact heart of misclassification and supervised learning |
| CN109411041A (en) * | 2018-02-24 | 2019-03-01 | 深圳市凯沃尔电子有限公司 | Ecg information processing method and electro cardio signal workstation system |
| CN109645979A (en) * | 2017-10-10 | 2019-04-19 | 深圳市理邦精密仪器股份有限公司 | Ambulatory ecg signal artifact identification method and device |
| CN109846472A (en) * | 2019-01-30 | 2019-06-07 | 郑州大学 | Beat classification method based on BiLSTM-Attention deep neural network |
| CN110693488A (en) * | 2019-10-22 | 2020-01-17 | 清华大学 | Electrocardiosignal processing system, electrocardiosignal processing method, electronic equipment and readable storage medium |
| CN110693483A (en) * | 2019-09-02 | 2020-01-17 | 乐普智芯(天津)医疗器械有限公司 | Method for automatically analyzing dynamic electrocardiogram |
| CN110929748A (en) * | 2019-10-12 | 2020-03-27 | 杭州电子科技大学 | Motion blur image feature matching method based on deep learning |
| CN111134658A (en) * | 2020-01-17 | 2020-05-12 | 乐普（北京）医疗器械股份有限公司 | Method and device for detecting RonT type heart beat in electrocardiosignal |
| CN111150387A (en) * | 2020-01-15 | 2020-05-15 | 深圳市邦健科技有限公司 | Electrocardiogram template matching method and device, computer equipment and storage medium |
| CN111150386A (en) * | 2019-12-30 | 2020-05-15 | 深圳市瑞康宏业科技开发有限公司 | Method, device and storage medium for electrocardio analysis |
Patent Citations (25)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4336810A (en) * | 1980-09-30 | 1982-06-29 | Del Mar Avionics | Method and apparatus for arrhythmia analysis of ECG recordings |
| WO2008007236A2 (en) * | 2006-06-07 | 2008-01-17 | Koninklijke Philips Electronics N.V. | Atrial fibrillation detection |
| CN102835954A (en) * | 2012-09-07 | 2012-12-26 | 深圳市邦健电子有限公司 | Method for generating heart beating waveform template and treating ventricular premature beat detection and device |
| US20160296762A1 (en) * | 2013-10-03 | 2016-10-13 | University Of Washington Through Its Center For Commercialization | Apparatuses and methods for classification of electrocardiogram signals during cardiopulmonary resuscitation |
| CN103720468A (en) * | 2013-12-05 | 2014-04-16 | 深圳先进技术研究院 | Artifact identification method and device applied to dynamic electrocardiogram data |
| CN105559771A (en) * | 2014-10-09 | 2016-05-11 | 中国移动通信集团公司 | Electrocardiosignal quality detection method and apparatus |
| CN104545887A (en) * | 2014-12-24 | 2015-04-29 | 深圳先进技术研究院 | Method and device for identifying artifact electrocardiograph waveforms |
| CN104586383A (en) * | 2015-01-12 | 2015-05-06 | 北京世纪今科医疗器械有限公司 | Electrocardiogram waveform classification method and device |
| CN105975385A (en) * | 2016-04-28 | 2016-09-28 | 浪潮(北京)电子信息产业有限公司 | Fuzzy neural network-based virtual machine energy consumption prediction method and system |
| CN106108889A (en) * | 2016-07-20 | 2016-11-16 | 杨平 | Electrocardiogram classification method based on degree of depth learning algorithm |
| CN107785066A (en) * | 2016-08-26 | 2018-03-09 | 深圳市理邦精密仪器股份有限公司 | Change the method, apparatus and system of heart beat type |
| CN106923820A (en) * | 2017-03-10 | 2017-07-07 | 深圳竹信科技有限公司 | A kind of electrocardiosignal artifact identification method and electrocardiosignal artifact identifying device |
| CN107358196A (en) * | 2017-07-12 | 2017-11-17 | 北京卫嘉高科信息技术有限公司 | A kind of sorting technique of heart beat type, device and electrocardiogram equipment |
| CN109645979A (en) * | 2017-10-10 | 2019-04-19 | 深圳市理邦精密仪器股份有限公司 | Ambulatory ecg signal artifact identification method and device |
| CN107951485A (en) * | 2017-11-27 | 2018-04-24 | 乐普（北京）医疗器械股份有限公司 | Ambulatory ECG analysis method and apparatus based on artificial intelligence self study |
| CN108257347A (en) * | 2018-01-10 | 2018-07-06 | 安徽大学 | A kind of flame image sequence sorting technique and device using convolutional neural networks |
| CN109411041A (en) * | 2018-02-24 | 2019-03-01 | 深圳市凯沃尔电子有限公司 | Ecg information processing method and electro cardio signal workstation system |
| CN109303561A (en) * | 2018-11-01 | 2019-02-05 | 杭州质子科技有限公司 | It is a kind of to clap the recognition methods clapped with the abnormal heart based on the artifact heart of misclassification and supervised learning |
| CN109846472A (en) * | 2019-01-30 | 2019-06-07 | 郑州大学 | Beat classification method based on BiLSTM-Attention deep neural network |
| CN110693483A (en) * | 2019-09-02 | 2020-01-17 | 乐普智芯(天津)医疗器械有限公司 | Method for automatically analyzing dynamic electrocardiogram |
| CN110929748A (en) * | 2019-10-12 | 2020-03-27 | 杭州电子科技大学 | Motion blur image feature matching method based on deep learning |
| CN110693488A (en) * | 2019-10-22 | 2020-01-17 | 清华大学 | Electrocardiosignal processing system, electrocardiosignal processing method, electronic equipment and readable storage medium |
| CN111150386A (en) * | 2019-12-30 | 2020-05-15 | 深圳市瑞康宏业科技开发有限公司 | Method, device and storage medium for electrocardio analysis |
| CN111150387A (en) * | 2020-01-15 | 2020-05-15 | 深圳市邦健科技有限公司 | Electrocardiogram template matching method and device, computer equipment and storage medium |
| CN111134658A (en) * | 2020-01-17 | 2020-05-12 | 乐普（北京）医疗器械股份有限公司 | Method and device for detecting RonT type heart beat in electrocardiosignal |
Non-Patent Citations (4)
| Title |
|---|
| 樊小毛: "动态心电心律失常智能检测方法及高性能计算研究", 中国优秀博士学位论文全文数据库, 15 January 2019 (2019-01-15), pages 1 - 116 * |
| 涂岳文;陈杭;付秀泉;李顶立;黄超;汤亚伟;叶树明;: "基于心搏聚类的Holter运动伪差段快速识别算法", 浙江大学学报(工学版), vol. 46, no. 6, 15 June 2012 (2012-06-15), pages 1148 - 1156 * |
| 阔永红: "动态起搏心电图分析方法的研究", 中国优秀博士学位论文全文数据库, 15 November 2006 (2006-11-15), pages 1 - 115 * |
| 阔永红: "动态起搏心电图自动分析", 国外医学.生物医学工程分册, vol. 20, no. 04, 18 August 1997 (1997-08-18), pages 193 - 198 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115778402A (en) * | 2022-12-02 | 2023-03-14 | 深圳华清心仪医疗电子有限公司 | Method and system for identifying artifact of dynamic electrocardiosignal |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109998525B (en) | Arrhythmia automatic classification method based on discriminant deep belief network | |
| CN108714026B (en) | Fine-grained electrocardiosignal classification method based on deep convolutional neural network and online decision fusion | |
| CN107122788B (en) | Identity recognition method and device based on electrocardiosignals | |
| Meintjes et al. | Fundamental heart sound classification using the continuous wavelet transform and convolutional neural networks | |
| Gupta et al. | Efficient R-peak detection in electrocardiogram signal based on features extracted using Hilbert transform and Burg method | |
| Diker et al. | A novel application based on spectrogram and convolutional neural network for ECG classification | |
| CN106725420A (en) | VPB recognition methods and VPB identifying system | |
| CN109948396B (en) | Heart beat classification method, heart beat classification device and electronic equipment | |
| CN110751131B (en) | Arrhythmia detection device | |
| CN110558975B (en) | Electrocardiosignal classification method and system | |
| CN112869716B (en) | Pulse feature identification system and method based on two-channel convolutional neural network | |
| CN112426160A (en) | Electrocardiosignal type identification method and device | |
| CN113057648A (en) | ECG signal classification method based on composite LSTM structure | |
| CN112869717B (en) | Pulse feature recognition and classification system and method based on BL-CNN | |
| CN111460951A (en) | Electrocardiosignal automatic analysis method based on deep learning | |
| CN110693488A (en) | Electrocardiosignal processing system, electrocardiosignal processing method, electronic equipment and readable storage medium | |
| CN115281688A (en) | Cardiac hypertrophy multi-label detection system based on multi-mode deep learning | |
| CN116361688A (en) | Multi-mode feature fusion model construction method for automatic classification of electrocardiographic rhythms | |
| CN115470832A (en) | Electrocardiosignal data processing method based on block chain | |
| CN113069124A (en) | Electrocardio monitoring method based on CNN-ET model | |
| CN113712525B (en) | Physiological parameter processing method and device and medical equipment | |
| CN113712525A (en) | Physiological parameter processing method and device and medical equipment | |
| RU2461877C1 (en) | Method for neural network analysis of cardial condition | |
| CN112257573B (en) | ECG identity recognition method based on t-SNE and Adaboost | |
| CN111839502B (en) | Electrocardiogram data anomaly detection method, device, equipment and storage medium |
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 |