CN114027846B - Intelligent electrocardiosignal processing method - Google Patents
Intelligent electrocardiosignal processing method Download PDFInfo
- Publication number
- CN114027846B CN114027846B CN202111298084.8A CN202111298084A CN114027846B CN 114027846 B CN114027846 B CN 114027846B CN 202111298084 A CN202111298084 A CN 202111298084A CN 114027846 B CN114027846 B CN 114027846B
- Authority
- CN
- China
- Prior art keywords
- electrocardiosignal
- signal
- representing
- electrocardiosignals
- model
- 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
- 238000003672 processing method Methods 0.000 title claims abstract description 17
- 238000005457 optimization Methods 0.000 claims abstract description 37
- 238000012545 processing Methods 0.000 claims abstract description 32
- 238000004458 analytical method Methods 0.000 claims abstract description 10
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 238000000605 extraction Methods 0.000 claims description 27
- 238000000034 method Methods 0.000 claims description 22
- 230000000747 cardiac effect Effects 0.000 claims description 17
- 239000011159 matrix material Substances 0.000 claims description 15
- 238000000354 decomposition reaction Methods 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 12
- 239000003990 capacitor Substances 0.000 claims description 11
- 230000009467 reduction Effects 0.000 claims description 10
- 238000001228 spectrum Methods 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 9
- 238000001914 filtration Methods 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- 230000008859 change Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—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
- A61B5/72—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
- A61B5/7235—Details of waveform analysis
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/725—Details of waveform analysis using specific filters therefor, e.g. Kalman or adaptive filters
-
- 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
- A61B5/7253—Details of waveform analysis characterised by using transforms
- A61B5/7257—Details of waveform analysis characterised by using transforms using Fourier transforms
-
- 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
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B7/00—Instruments for auscultation
- A61B7/02—Stethoscopes
- A61B7/04—Electric stethoscopes
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Animal Behavior & Ethology (AREA)
- Veterinary Medicine (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- Artificial Intelligence (AREA)
- Signal Processing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Cardiology (AREA)
- Mathematical Physics (AREA)
- Acoustics & Sound (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention relates to the technical field of signal processing, and discloses an intelligent electrocardiosignal processing method, which comprises the following steps: acquiring electrocardiosignals and preprocessing the electrocardiosignals; extracting the characteristics of the preprocessed electrocardiosignals to obtain electrocardiosignal characteristics; taking the electrocardiosignal characteristics as input and the electrocardiosignal classification result as output, constructing a deep stack network electrocardiosignal identification model, and determining a parameter optimization objective function; optimizing and solving the objective function by using a penalty function optimization algorithm to obtain optimized model parameters; according to the optimized model parameters and the electrocardiosignal recognition model, the electrocardiosignal characteristics extracted according to the preprocessing and the characteristics are used as model input, and then the processing and analysis results of the electrocardiosignals can be obtained. The invention realizes the intelligent processing of electrocardiosignals.
Description
Technical Field
The invention relates to the technical field of signal processing, in particular to an intelligent electrocardiosignal processing method.
Background
The existing electrocardiosignal analysis processing generally analyzes the generated electrocardiosignal data, and the current heart rate condition of a patient is difficult to monitor in real time. The change condition of the electrocardiosignals of the patient is very complex, and particularly when the patient faces an emergency, the electrocardiosignals are collected for analysis, so that the precious rescue time of the patient is lost. To this end, the patent proposes a method of monitoring the heart rate of a patient in real time and performing a processing analysis.
The real-time electrocardiosignal processing scene has 3 main requirements, and firstly, the requirement on the running time is higher, so that the real-time electrocardiosignal processing scene needs a quick and low-delay service; secondly, the analysis usually needs to be operated for a long time without interruption, so that the analysis efficiency is required to be as fast as possible; thirdly, the electrocardiosignals need to be analyzed accurately in real time, so that the electrocardiosignals need to have the characteristics of high efficiency and intelligence. In order to meet the requirements, the patent provides an intelligent electrocardiosignal processing method for realizing rapid processing and analysis of electrocardiosignals.
Disclosure of Invention
The invention provides an intelligent electrocardiosignal processing method, aiming at (1) realizing rapid electrocardiosignal analysis and processing; (2) the electrocardiosignal processing efficiency is improved.
The invention provides an intelligent electrocardiosignal processing method, which comprises the following steps:
s1: acquiring electrocardiosignals and preprocessing the electrocardiosignals;
s2: extracting the characteristics of the preprocessed electrocardiosignals to obtain electrocardiosignal characteristics;
s3: taking the electrocardiosignal characteristics as input and the electrocardiosignal classification result as output, constructing a deep stack network electrocardiosignal identification model, and determining a parameter optimization objective function;
s4: optimizing and solving the objective function by using a penalty function optimization algorithm to obtain optimized model parameters;
s5: and (4) according to the optimized model parameters and the electrocardiosignal identification model obtained in the step (S4), the electrocardiosignal processed according to the step (S1-2) is used as model input, and then the processing and analysis result of the electrocardiosignal can be obtained.
As a further improvement of the method of the invention:
the step S1 is that electrocardiosignals are collected, a high-pass filter is constructed, and the electrocardiosignals are filtered by the high-pass filter, which comprises
Connecting the electrocardio electrode plate with the electrocardio sensor, fixing the electrocardio electrode plate on the skin of a human body, continuously acquiring electrocardiosignals of the human body by the electrocardio electrode plate, and transmitting the electrocardiosignals to the electrocardio sensor;
constructing a high-pass filter in a signal receiving portion of the electrocardiograph sensor, the high-pass filter including a capacitor and a resistor, the capacitor being connected in series with the electrocardiograph signal path and the resistor being connected in parallel with the signal path;
the high-pass filter allows the frequency of signals received by the electrocardio sensor to be higher than the cut-off frequencyThe cut-off frequency ofThe calculation formula of (2) is as follows:
wherein:
r represents a resistance value of a resistor in the high-pass filter;
c denotes the capacitance of the capacitor in the high-pass filter.
In the step S1, performing noise reduction processing on the filtered electrocardiographic signal by using a noise reduction algorithm combined with IMF signal decomposition to obtain a preprocessed electrocardiographic signal, including:
performing EMD signal decomposition on the electrocardiosignals after filtering processing to obtain IMF components of the electrocardiosignals:
1) filtering the processed electrocardiosignalAdding white noise of the same lengthTo obtain a noisy signal:
2) Go through to obtain the signalConnecting all the maximum value points and all the minimum value points by cubic spline curves respectively to obtain signalsThe upper envelope line and the lower envelope line are the mean value curves of the upper envelope line and the lower envelope lineThen, then;
4) Repeating the step 3) 9 times to obtainAs a signalOf a first set of IMF components, whereinThe remaining components of the first set of IMF components;
5) To pairCarrying out steps 2) -4) to obtain k-1 group IMF componentsAnd a residual component(ii) a Will contain a noise signalThe decomposition is of the formula:
the IMF component obtained by decompositionPerforming fast Fourier transform to obtain frequency spectrum of IMF component:
wherein:
if the frequency domain energy of 0-60Hz in the frequency spectrum of the IMF component reaches more than 90% of the energy of the whole frequency domain, judging the IMF component as the IMF component of the electrocardiosignal, otherwise, judging the IMF component as the IMF component of the noise signal, and deleting the IMF component of the noise signal;
performing signal reconstruction processing on all IMF components of the electrocardiosignals according to the following formula to obtain the electrocardiosignals after noise reduction:
wherein:
j represents the number of groups of components of the cardiac signal IMF.
The step S2 of extracting the electrocardiographic envelope characteristic of the preprocessed electrocardiographic signal includes:
to the electrocardiosignals after normalization processingAnd (3) carrying out sectional processing, wherein each 30ms is a section of signal, a section of signal is taken every 20ms, the overlapping part between two adjacent sections of signals is a 10ms signal, and the entropy of each section of electrocardiosignal is as follows:
wherein:
n represents the number of sampled electrocardiosignal data of 30 ms;
taking the entropy of each section of electrocardiosignal as the electrocardio envelope characteristic of each section of electrocardiosignal, and determining the electrocardio envelope characteristic of the electrocardiosignal asWherein p represents the number of segments in the cardiac signal,and representing the electrocardio envelope characteristics of the ith segment of electrocardiosignal.
The step of S2, extracting the electrocardiographic signal features from the electrocardiographic envelope features, includes:
characterizing the cardiac envelopeAs an input of the feature extraction network, a formula for extracting the characteristics of the electrical signal at the center of the electrocardio envelope characteristics by using the feature extraction network is as follows:
wherein:
the serial number of a hidden layer in the feature extraction network is shown, and a V-layer hidden layer is shared in the feature extraction network;
representing the features extracted at layer v in the feature extraction network,representing the features extracted at level v-1 in the feature extraction network, when v equals 1,representing the electrocardiogramEnvelope feature;
representing an activation function in the feature extraction network, wherein the selected activation function is a Sigmoid function;
and taking the features extracted by the V layer as the electrocardiosignal features L.
In the step S3, the electrocardiosignal features are used as input, the electrocardiosignal classification result is used as output, and a deep stack network electrocardiosignal identification model is constructed, including:
taking the electrocardiosignal characteristics as a training set of a deep stack network electrocardiosignal identification model, and labeling a signal classification label corresponding to the electrocardiosignal characteristics;
inputting the electrocardiosignal characteristics L into a deep stack network electrocardiosignal identification model, and outputting the prediction probability of the identification result of different electrocardiosignal characteristics by the deep stack network electrocardiosignal identification model:
wherein:
z represents the category number of the electrocardiosignal categories predicted by the deep stack network electrocardiosignal recognition model;
l represents the electrocardiosignal characteristics;
The parameter optimization objective function of the deep stack network electrocardiosignal identification model constructed in the step S3 comprises the following steps:
the parameter optimization objective function of the built deep stack network electrocardiosignal identification model is as follows:
wherein:
z represents the number of categories identifying the category of the cardiac signal;
representing samples of an electrocardiosignalIdentifying results in the deep stack network electrocardiosignal identification model;
the constraint conditions of the constructed parameter optimization objective function are as follows:
wherein:
In the step S4, performing optimization solution on the parameter optimization objective function by using a penalty function optimization algorithm, including:
converting the parameter optimization objective function as follows:
wherein:
,representing samples of an electrocardiographic signalThe probability parameter of (a) is determined,representing samples of an electrocardiographic signalA weight vector of a jth category output matrix;
calculating a penalty term for the objective function:
constructing a penalty function of the parameter optimization objective function:
wherein:
solving the penalty function whenAnd the value of H is the optimized model parameter value of the deep stack network electrocardiosignal identification model.
In the step S5, according to the optimized model parameters and the electrocardiosignal recognition model obtained in the step S4, the electrocardiosignal processed in the step S1-2 is input as a model to obtain an electrocardiosignal recognition result, which includes:
taking the optimized model parameters as model parameters of the deep stack network electrocardiosignal recognition model to obtain the optimized deep stack network electrocardiosignal recognition model;
and (4) processing the electrocardiosignals according to the electrocardiosignal processing method of the steps S1-S2 to obtain electrocardiosignal characteristics, inputting the electrocardiosignal characteristics as a model of a deep stack network electrocardiosignal identification model, and outputting an electrocardiosignal identification result with the maximum prediction probability by the deep stack network electrocardiosignal identification model.
Compared with the prior art, the invention provides an intelligent electrocardiosignal processing method, which has the following advantages:
firstly, the scheme provides an electrocardiosignal preprocessing flow, a high-pass filter is constructed in a signal receiving part in an electrocardio sensor, the high-pass filter comprises a capacitor and a resistor, the capacitor is connected with an electrocardiosignal path in series, and the resistor is connected with the signal path in parallel, so that the high-pass filter allows the electrocardiosignal receiving frequency of the electrocardio sensor to be higher than a cut-off frequencyThe received signal frequencies are all higher than the cut-off frequency, so that part of low-frequency noise signals are filtered; meanwhile, the electrocardiosignal is subjected to noise reduction by utilizing a noise reduction algorithm combined with IMF signal decomposition, and the center frequency of an IMF component generated by decomposition is divided into two when an EMD method is used for processing Gaussian white noiseThe multiplied speed is reduced, the frequency range is increased at double speed, Gaussian white noise is added into the original signal, the original signal can be decomposed into a series of signals with different scales according to binary characteristics, elimination of noise signals with various scales can be realized, most of energy of the electrocardiosignal is in the frequency domain range of 0-50Hz, if the frequency domain energy of 0-60Hz in the IMF frequency spectrum after FFT processing does not reach more than 90% of the energy of the whole frequency domain, the IMF component can be regarded as the IMF component of the noise signal, and the noise reduction processing of the electrocardiosignal is realized by deleting the IMF component.
Meanwhile, according to the scheme, the electrocardiosignal features are extracted by using the feature extraction network, the electrocardiosignal features are used as input, the electrocardiosignal recognition result is used as output, a deep stack network electrocardiosignal recognition model is built, a deep stack network electrocardiosignal recognition model optimization strategy is provided, and a target function is optimized by building parameters of the deep stack network electrocardiosignal recognition model:
wherein: y represents the number of training samples and,representing the ith sample in the training sample set; z represents the number of categories identifying the category of the cardiac signal;representing the attenuation coefficient of the weight vector,it is set to 0.8;representing samples of an electrocardiographic signalIdentifying results in the deep stack network electrocardiosignal identification model;representing samples of an electrocardiographic signalTrue signal classification tags of (1);representing samples of an electrocardiosignalA weight vector of a jth category output matrix; the constraint conditions of the constructed parameter optimization objective function are as follows:
wherein:a weight vector representing the output matrix of the jth category in the softmax classifier; and converting the parameter optimization objective function as follows:
wherein:,representing samples of an electrocardiographic signalThe probability parameter of (a) is,representing samples of an electrocardiographic signalA weight vector of a jth category output matrix; calculating a penalty term for the objective function:
constructing a penalty function of the parameter optimization objective function:
wherein:a penalty factor representing a penalty function, set to 20; solving the penalty function whenAnd the value of H is the optimized model parameter value of the deep stack network electrocardiosignal identification model. According to the scheme, the parameter optimization objective function of the deep stack network electrocardiosignal recognition model with the constraint condition is constructed, and the parameter optimization objective function with the constraint condition is converted into the unconstrained objective function by using a penalty function method, so that the parameter optimization value in the objective function is calculated by using a function solving method, the parameter optimization process is simplified, and the training efficiency of the deep stack network electrocardiosignal recognition model is improved.
Drawings
Fig. 1 is a schematic flow chart of an intelligent electrocardiographic signal processing method according to an embodiment of the present invention;
the implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Example 1:
s1: collecting electrocardiosignals and preprocessing the electrocardiosignals.
The step S1 is that electrocardiosignals are collected, a high-pass filter is constructed, and the electrocardiosignals are filtered by the high-pass filter, which comprises
Connecting the electrocardio electrode plate with the electrocardio sensor, fixing the electrocardio electrode plate on the skin of a human body, continuously acquiring electrocardiosignals of the human body by the electrocardio electrode plate, and transmitting the electrocardiosignals to the electrocardio sensor;
constructing a high-pass filter in a signal receiving part in the electrocardio sensor, wherein the high-pass filter comprises a capacitor and a resistor, the capacitor is connected with an electrocardio signal path in series, and the resistor is connected with the signal path in parallel;
the high-pass filter allows the frequency of signals received by the electrocardio sensor to be higher than the cut-off frequencySaid cut-off frequency of the cardiac signalThe calculation formula of (2) is as follows:
wherein:
r represents a resistance value of a resistor in the high-pass filter;
c denotes the capacitance of the capacitor in the high-pass filter.
In the step S1, performing noise reduction processing on the filtered electrocardiographic signal by using a noise reduction algorithm combined with IMF signal decomposition to obtain a preprocessed electrocardiographic signal, including:
performing EMD signal decomposition on the electrocardiosignals after filtering processing to obtain IMF components of the electrocardiosignals:
1) filtering the processed electrocardiosignalAdding white noise of the same lengthObtaining a noisy signal:
2) Go through to obtain the signalConnecting all the maximum value points and all the minimum value points by cubic spline curves respectively to obtain signalsThe upper envelope line and the lower envelope line are the mean value curves of the upper envelope line and the lower envelope lineThen, then;
4) Repeating the step 3) 9 times to obtainAs a signalOf a first set of IMF components, whereinThe remaining components of the first set of IMF components;
5) To pairCarrying out steps 2) -4) to obtain k-1 group IMF componentsAnd a residual component(ii) a Will contain a noisy signalThe decomposition is of the formula:
the IMF component obtained by decompositionPerforming fast Fourier transform to obtain frequency spectrum of IMF component:
wherein:
if the frequency domain energy of 0-60Hz in the frequency spectrum of the IMF component reaches more than 90% of the energy of the whole frequency domain, judging the IMF component as the IMF component of the electrocardiosignal, otherwise, judging the IMF component as the IMF component of the noise signal, and deleting the IMF component of the noise signal;
performing signal reconstruction processing on all IMF components of the electrocardiosignals according to the following formula to obtain the electrocardiosignals after noise reduction:
wherein:
j represents the number of groups of components of the cardiac signal IMF.
S2: and extracting the characteristics of the preprocessed electrocardiosignals to obtain the electrocardiosignal characteristics.
The step of S2, extracting the ecg envelope characteristics of the preprocessed ecg signal, includes:
to the electrocardio after normalization processingSignalAnd (3) carrying out sectional processing, wherein each 30ms is a section of signal, a section of signal is taken every 20ms, the overlapping part between two adjacent sections of signals is a 10ms signal, and the entropy of each section of electrocardiosignal is as follows:
wherein:
n represents the number of sampled electrocardiosignal data of 30 ms;
taking the entropy of each section of electrocardiosignal as the electrocardio envelope characteristic of each section of electrocardiosignal, and determining the electrocardio envelope characteristic of the electrocardiosignal asWherein p represents the number of segments in the cardiac signal,and representing the electrocardio envelope characteristics of the ith segment of electrocardiosignal.
S3: and (3) taking the electrocardiosignal characteristics as input and the electrocardiosignal classification result as output, constructing a deep stack network electrocardiosignal identification model, and determining a parameter optimization objective function.
In the step S3, the electrocardiosignal features are used as input, the electrocardiosignal classification result is used as output, and a deep stack network electrocardiosignal identification model is constructed, including:
taking the electrocardiosignal characteristics as a training set of a deep stack network electrocardiosignal identification model, and labeling a signal classification label corresponding to the electrocardiosignal characteristics;
inputting the electrocardiosignal characteristics L into a deep stack network electrocardiosignal identification model, and outputting the prediction probability of the identification result of different electrocardiosignal characteristics by the deep stack network electrocardiosignal identification model:
wherein:
z represents the category number of the electrocardiosignal categories predicted by the deep stack network electrocardiosignal recognition model;
l represents the electrocardiosignal characteristics;
The parameter optimization objective function of the deep stack network electrocardiosignal identification model constructed in the step S3 comprises the following steps:
the parameter optimization objective function of the built deep stack network electrocardiosignal identification model is as follows:
wherein:
z represents the number of categories identifying the category of the cardiac signal;
representing samples of an electrocardiographic signalIdentifying results in the deep stack network electrocardiosignal identification model;
representing samples of an electrocardiographic signalA weight vector of a jth category output matrix;
the constraint conditions of the constructed parameter optimization objective function are as follows:
wherein:
S4: and (5) optimizing and solving the objective function by using a penalty function optimization algorithm to obtain optimized model parameters.
Converting the parameter optimization objective function as follows:
wherein:
,representing samples of an electrocardiographic signalThe probability parameter of (a) is,representing samples of an electrocardiosignalA weight vector of a jth category output matrix;
calculating a penalty term for the objective function:
constructing a penalty function of the parameter optimization objective function:
wherein:
solving the penalty function whenAnd the value of H is the optimized model parameter value of the deep stack network electrocardiosignal identification model.
In the step S5, according to the optimized model parameters and the electrocardiosignal recognition model obtained in the step S4, the electrocardiosignal processed in the step S1-2 is input as a model to obtain an electrocardiosignal recognition result, which includes:
taking the optimized model parameters as model parameters of the deep stack network electrocardiosignal recognition model to obtain the optimized deep stack network electrocardiosignal recognition model;
and (4) processing the electrocardiosignals according to the electrocardiosignal processing method of the steps S1-S2 to obtain electrocardiosignal characteristics, inputting the electrocardiosignal characteristics as a model of a deep stack network electrocardiosignal identification model, and outputting an electrocardiosignal identification result with the maximum prediction probability by the deep stack network electrocardiosignal identification model.
Example 2:
this example is substantially the same as example 1, except that:
s2: and extracting the characteristics of the preprocessed electrocardiosignals to obtain the electrocardiosignal characteristics.
The step of S2, extracting the electrocardiographic signal feature in the electrocardiographic envelope feature, includes:
characterizing the cardiac envelopeAs an input of the feature extraction network, a formula for extracting the characteristics of the electrical signal at the center of the electrocardio envelope characteristics by using the feature extraction network is as follows:
wherein:
the serial number of a hidden layer in the feature extraction network is shown, and a V-layer hidden layer is shared in the feature extraction network;
representing the features extracted at layer v in the feature extraction network,representing the features extracted at level v-1 in the feature extraction network, when v equals 1,representing features of an envelope of the electrocardiogram;
representing activation functions in a feature extraction network, selected activation functionsIs a Sigmoid function;
and taking the features extracted by the V layer as the electrocardiosignal features L.
It should be noted that the above-mentioned numbers of the embodiments of the present invention are merely for description, and do not represent the merits of the embodiments. And the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, apparatus, article, or method. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, apparatus, article, or method that includes the element.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present invention or portions thereof contributing to the prior art may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) as described above and includes several instructions for enabling a terminal device (which may be a mobile phone, a computer, a server, or a network device) to execute the method according to the embodiments of the present invention.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (7)
1. An intelligent electrocardiosignal processing method is characterized by comprising the following steps:
s1: acquiring electrocardiosignals and preprocessing the electrocardiosignals;
s2: extracting the features of the preprocessed electrocardiosignals to obtain the electrocardiosignal features, and extracting the electrocardio envelope features of the preprocessed electrocardiosignals, wherein the extracting comprises the following steps:
to the electrocardiosignals after normalization processingAnd (3) carrying out sectional processing, wherein each 30ms is a section of signal, a section of signal is taken every 20ms, the overlapping part between two adjacent sections of signals is a 10ms signal, and the entropy of each section of electrocardiosignal is as follows:
wherein:
n represents the number of sampled electrocardiosignal data of 30 ms;
taking the entropy of each section of electrocardiosignal as the electrocardio envelope characteristic of each section of electrocardiosignal, and determining the electrocardio envelope characteristic of the electrocardiosignal asWherein p represents the number of segments in the cardiac signal,representing the electrocardio envelope characteristics of the ith segment of electrocardiosignal;
extracting electrocardiosignal characteristics in the electrocardio envelope characteristics, comprising:
characterizing the cardiac envelopeAs an input of the feature extraction network, a formula for extracting the characteristics of the electrical signal at the center of the electrocardio envelope characteristics by using the feature extraction network is as follows:
wherein:
representing the features extracted at layer v in the feature extraction network,representing the features extracted at level v-1 in the feature extraction network, when v equals 1,representing features of an envelope of the electrocardiogram;
taking the features extracted from the V layer as electrocardiosignal features L;
s3: taking the electrocardiosignal characteristics as input and the electrocardiosignal classification result as output, constructing a deep stack network electrocardiosignal identification model, and determining a parameter optimization objective function;
s4: optimizing and solving the objective function by using a penalty function optimization algorithm to obtain optimized model parameters;
s5: and (4) according to the optimized model parameters and the electrocardiosignal identification model obtained in the step (S4), the electrocardiosignal processed according to the step (S1-2) is used as model input, and then the processing and analysis result of the electrocardiosignal can be obtained.
2. The method as claimed in claim 1, wherein said step S1 is a step of collecting electrocardiosignals, constructing a high-pass filter, and performing filtering process on the electrocardiosignals by using the high-pass filter, comprising
Connecting the electrocardio electrode plate with the electrocardio sensor, fixing the electrocardio electrode plate on the skin of a human body, continuously acquiring electrocardiosignals of the human body by the electrocardio electrode plate, and transmitting the electrocardiosignals to the electrocardio sensor;
a signal receiving part in the electrocardio sensor is provided with a high-pass filter, the high-pass filter comprises a capacitor and a resistor, the capacitor is connected with an electrocardio signal path in series, and the resistor is connected with the signal path in parallel;
the high-pass filter allows the frequency of signals received by the electrocardio sensor to be higher than the cut-off frequencyThe cut-off frequency ofThe calculation formula of (2) is as follows:
wherein:
r represents a resistance value of a resistor in the high-pass filter;
c denotes the capacitance of the capacitor in the high-pass filter.
3. The intelligent electrocardiosignal processing method according to claim 2, wherein in the step S1, the denoising algorithm combined with IMF signal decomposition is used for denoising the filtered electrocardiosignal to obtain the preprocessed electrocardiosignal, and the method comprises:
performing EMD signal decomposition on the electrocardiosignals after filtering processing to obtain IMF components of the electrocardiosignals:
1) filtering the processed electrocardiosignalAdding white noise of the same lengthTo obtain a noisy signal:
2) Go through to obtain the signalConnecting all the maximum value points and all the minimum value points by cubic spline curves respectively to obtain signalsUpper envelope and lower envelope ofThe mean curve of the upper and lower envelope curves isThen, then;
4) Repeating the step 3) 9 times to obtainAs a signalOf a first set of IMF components, whereinThe remaining components of the first set of IMF components;
5) To pairCarrying out steps 2) -4) to obtain k-1 group IMF componentsAnd a residual component(ii) a Will contain a noise signalThe decomposition is of the formula:
the IMF component obtained by decompositionPerforming fast Fourier transform to obtain frequency spectrum of IMF component:
wherein:
if the frequency domain energy of 0-60Hz in the frequency spectrum of the IMF component reaches more than 90% of the energy of the whole frequency domain, judging the IMF component as the IMF component of the electrocardiosignal, otherwise, judging the IMF component as the IMF component of the noise signal, and deleting the IMF component of the noise signal;
performing signal reconstruction processing on all IMF components of the electrocardiosignals according to the following formula to obtain the electrocardiosignals after noise reduction:
wherein:
j represents the number of groups of components of the cardiac signal IMF.
4. The intelligent electrocardiosignal processing method according to claim 1, wherein the step S3 is implemented by constructing a deep stack network electrocardiosignal recognition model by taking electrocardiosignal characteristics as input and electrocardiosignal classification results as output, and comprises the steps of:
taking the electrocardiosignal characteristics as a training set of a deep stack network electrocardiosignal identification model, and labeling a signal classification label corresponding to the electrocardiosignal characteristics;
inputting the electrocardiosignal characteristics L into a deep stack network electrocardiosignal identification model, and outputting the prediction probability of the identification result of different electrocardiosignal characteristics by the deep stack network electrocardiosignal identification model:
wherein:
z represents the category number of the electrocardiosignal categories predicted by the deep stack network electrocardiosignal recognition model;
l represents the electrocardiosignal characteristics;
representing deep stacked network coresThe electric signal identifies model parameters of the model;
5. The intelligent cardiac signal processing method according to claim 1, wherein the step S3 of constructing a parameter optimization objective function of the deep stack network cardiac signal identification model includes:
the parameter optimization objective function of the built deep stack network electrocardiosignal identification model is as follows:
wherein:
y represents the number of training samples,representing the electrocardiosignal characteristics of the ith sample in the training sample set;
z represents the number of categories identifying the category of the cardiac signal;
representing samples of an electrocardiographic signalIdentifying results in the deep stack network electrocardiosignal identification model;
representing samples of an electrocardiographic signalA weight vector of a jth category output matrix;
the constraint conditions of the constructed parameter optimization objective function are as follows:
wherein:
6. The intelligent cardiac signal processing method as set forth in claim 5, wherein the step S4 of performing optimization solution on the parameter optimization objective function by using a penalty function optimization algorithm includes:
and converting the parameter optimization objective function as follows:
wherein:
,representing samples of an electrocardiographic signalThe probability parameter of (a) is,representing samples of an electrocardiographic signalA weight vector of a jth category output matrix;
calculating a penalty term for the objective function:
constructing a penalty function of the parameter optimization objective function:
wherein:
7. The intelligent method as claimed in claim 6, wherein the step S5, based on the optimized model parameters and the ecg signal recognition model obtained in step S4, obtains the recognition result of the ecg signal by inputting the ecg signal processed in step S1-2 as a model, and comprises:
taking the optimized model parameters as model parameters of the deep stack network electrocardiosignal recognition model to obtain the optimized deep stack network electrocardiosignal recognition model;
and (4) processing the electrocardiosignals according to the electrocardiosignal processing method of the steps S1-S2 to obtain electrocardiosignal characteristics, inputting the electrocardiosignal characteristics as a model of a deep stack network electrocardiosignal identification model, and outputting an electrocardiosignal identification result with the maximum prediction probability by the deep stack network electrocardiosignal identification model.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111298084.8A CN114027846B (en) | 2021-11-04 | 2021-11-04 | Intelligent electrocardiosignal processing method |
PCT/CN2021/135218 WO2023077592A1 (en) | 2021-11-04 | 2021-12-03 | Intelligent electrocardiosignal processing method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111298084.8A CN114027846B (en) | 2021-11-04 | 2021-11-04 | Intelligent electrocardiosignal processing method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114027846A CN114027846A (en) | 2022-02-11 |
CN114027846B true CN114027846B (en) | 2022-08-26 |
Family
ID=80136164
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111298084.8A Active CN114027846B (en) | 2021-11-04 | 2021-11-04 | Intelligent electrocardiosignal processing method |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN114027846B (en) |
WO (1) | WO2023077592A1 (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117814805A (en) * | 2024-03-05 | 2024-04-05 | 自贡市第一人民医院 | Intelligent processing method for data of clinical care equipment |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2007053975A1 (en) * | 2005-11-09 | 2007-05-18 | Iris Zhao | A mobile terminal, a medical sevice system and an ecg monitoring method for monitoring |
CN107007279A (en) * | 2017-03-17 | 2017-08-04 | 浙江大学 | A kind of noninvasive intracardiac exciting independent positioning method of exception based on stacking-type self-encoding encoder |
CN110801221A (en) * | 2019-12-09 | 2020-02-18 | 中山大学 | Sleep apnea fragment detection method and device based on unsupervised feature learning |
CN110974213A (en) * | 2019-12-20 | 2020-04-10 | 哈尔滨理工大学 | Electrocardiosignal identification method based on deep stack network |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104523266B (en) * | 2015-01-07 | 2017-04-05 | 河北大学 | A kind of electrocardiosignal automatic classification method |
CN105117736B (en) * | 2015-08-18 | 2018-04-17 | 西安电子科技大学 | Classification of Polarimetric SAR Image method based on sparse depth heap stack network |
US10685429B2 (en) * | 2017-02-22 | 2020-06-16 | Siemens Healthcare Gmbh | Denoising medical images by learning sparse image representations with a deep unfolding approach |
CN108470156B (en) * | 2018-03-06 | 2021-09-28 | 南京邮电大学 | Heart sound signal classification and identification method |
CN108670291A (en) * | 2018-06-01 | 2018-10-19 | 四川长虹电器股份有限公司 | The heart sound kind identification method of improved MFCC is combined based on EMD |
CN109602416A (en) * | 2019-01-09 | 2019-04-12 | 山东省计算中心(国家超级计算济南中心) | A kind of method of ECG signal joint baseline correction and noise reduction |
CN111832537B (en) * | 2020-07-27 | 2023-04-25 | 深圳竹信科技有限公司 | Abnormal electrocardiosignal identification method and abnormal electrocardiosignal identification device |
-
2021
- 2021-11-04 CN CN202111298084.8A patent/CN114027846B/en active Active
- 2021-12-03 WO PCT/CN2021/135218 patent/WO2023077592A1/en unknown
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2007053975A1 (en) * | 2005-11-09 | 2007-05-18 | Iris Zhao | A mobile terminal, a medical sevice system and an ecg monitoring method for monitoring |
CN107007279A (en) * | 2017-03-17 | 2017-08-04 | 浙江大学 | A kind of noninvasive intracardiac exciting independent positioning method of exception based on stacking-type self-encoding encoder |
CN110801221A (en) * | 2019-12-09 | 2020-02-18 | 中山大学 | Sleep apnea fragment detection method and device based on unsupervised feature learning |
CN110974213A (en) * | 2019-12-20 | 2020-04-10 | 哈尔滨理工大学 | Electrocardiosignal identification method based on deep stack network |
Non-Patent Citations (1)
Title |
---|
基于香农信息熵的心电R波检测算法;苏嘉豪等;《电子世界》;20170423(第08期);第59-64页 * |
Also Published As
Publication number | Publication date |
---|---|
WO2023077592A1 (en) | 2023-05-11 |
CN114027846A (en) | 2022-02-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107844755B (en) | Electroencephalogram characteristic extraction and classification method combining DAE and CNN | |
Khafaga et al. | Optimization of Electrocardiogram Classification Using Dipper Throated Algorithm and Differential Evolution. | |
CN111134662B (en) | Electrocardio abnormal signal identification method and device based on transfer learning and confidence degree selection | |
CN107495959A (en) | A kind of electrocardiosignal sorting technique based on one-dimensional convolutional neural networks | |
CN111476339B (en) | Rolling bearing fault feature extraction method, intelligent diagnosis method and system | |
CN111387975B (en) | Electroencephalogram signal identification method based on machine learning | |
CN107468260A (en) | A kind of brain electricity analytical device and analysis method for judging ANIMAL PSYCHE state | |
Kora et al. | Heart disease detection using hybrid of bacterial foraging and particle swarm optimization | |
US20200054262A1 (en) | Method for real time analyzing stress using deep neural network algorithm | |
Esfahanian et al. | On contour-based classification of dolphin whistles by type | |
CN114027846B (en) | Intelligent electrocardiosignal processing method | |
Renjith et al. | Speech based emotion recognition in Tamil and Telugu using LPCC and hurst parameters—A comparitive study using KNN and ANN classifiers | |
Awati et al. | Smart heart disease detection using particle swarm optimization and support vector machine | |
US20230225663A1 (en) | Method for predicting multi-type electrocardiogram heart rhythms based on graph convolution | |
CN112580486A (en) | Human behavior classification method based on radar micro-Doppler signal separation | |
CN116211320A (en) | Pattern recognition method of motor imagery brain-computer interface based on ensemble learning | |
CN108647584A (en) | Cardiac arrhythmia method for identifying and classifying based on rarefaction representation and neural network | |
Ajitkumar Singh et al. | Heart abnormality classification using PCG and ECG recordings | |
CN113870903A (en) | Pathological voice recognition method, device, equipment and storage medium | |
CN113421546A (en) | Cross-tested multi-mode based speech synthesis method and related equipment | |
CN116211322A (en) | Depression recognition method and system based on machine learning electroencephalogram signals | |
Prajitha et al. | Variance approximation and probabilistic decomposition noise removal framework for arrhythmia detection and classification on internet of medical things environment | |
Yildirim et al. | Heartbeat type classification with optimized feature vectors | |
CN113940638B (en) | Pulse wave signal identification and classification method based on frequency domain dual-feature fusion | |
CN114091529A (en) | Electroencephalogram emotion recognition method based on generation countermeasure network data enhancement |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
PE01 | Entry into force of the registration of the contract for pledge of patent right | ||
PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: An Intelligent ECG Signal Processing Method Effective date of registration: 20231205 Granted publication date: 20220826 Pledgee: China Construction Bank Corporation Shaoyang Jianshe Road Sub-branch Pledgor: HUNAN VENT MEDICAL TECHNOLOGY Co.,Ltd. Registration number: Y2023980069588 |