CN113768514A - Arrhythmia classification method based on convolutional neural network and gated cyclic unit - Google Patents
Arrhythmia classification method based on convolutional neural network and gated cyclic unit Download PDFInfo
- Publication number
- CN113768514A CN113768514A CN202110909940.2A CN202110909940A CN113768514A CN 113768514 A CN113768514 A CN 113768514A CN 202110909940 A CN202110909940 A CN 202110909940A CN 113768514 A CN113768514 A CN 113768514A
- Authority
- CN
- China
- Prior art keywords
- neural network
- convolutional neural
- convolution
- data
- electrocardio
- 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
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 77
- 238000000034 method Methods 0.000 title claims abstract description 52
- 206010003119 arrhythmia Diseases 0.000 title claims abstract description 46
- 230000006793 arrhythmia Effects 0.000 title claims abstract description 46
- 125000004122 cyclic group Chemical group 0.000 title claims abstract description 35
- 238000012549 training Methods 0.000 claims abstract description 26
- 238000012360 testing method Methods 0.000 claims abstract description 8
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 230000006870 function Effects 0.000 claims description 37
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 claims description 35
- 230000004913 activation Effects 0.000 claims description 23
- 238000013528 artificial neural network Methods 0.000 claims description 11
- 238000005070 sampling Methods 0.000 claims description 10
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000002790 cross-validation Methods 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 6
- 230000001629 suppression Effects 0.000 claims description 6
- 238000010977 unit operation Methods 0.000 claims description 6
- 238000012952 Resampling Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 230000008030 elimination Effects 0.000 claims description 3
- 238000003379 elimination reaction Methods 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 abstract description 13
- 230000008034 disappearance Effects 0.000 abstract description 7
- 208000024172 Cardiovascular disease Diseases 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 238000007635 classification algorithm Methods 0.000 description 6
- 230000035945 sensitivity Effects 0.000 description 6
- 206010006578 Bundle-Branch Block Diseases 0.000 description 5
- 206010015856 Extrasystoles Diseases 0.000 description 5
- 208000000418 Premature Cardiac Complexes Diseases 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000013145 classification model Methods 0.000 description 5
- 201000010099 disease Diseases 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 4
- 206010006580 Bundle branch block left Diseases 0.000 description 3
- 230000002763 arrhythmic effect Effects 0.000 description 3
- 230000001746 atrial effect Effects 0.000 description 3
- 230000034994 death Effects 0.000 description 3
- 231100000517 death Toxicity 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 201000001715 left bundle branch hemiblock Diseases 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000033764 rhythmic process Effects 0.000 description 3
- 206010006582 Bundle branch block right Diseases 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000000747 cardiac effect Effects 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 239000012634 fragment Substances 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 201000007916 right bundle branch block Diseases 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 230000002861 ventricular Effects 0.000 description 2
- 206010042434 Sudden death Diseases 0.000 description 1
- 210000004027 cell Anatomy 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000007170 pathology Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
- 230000007704 transition 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
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
- A61B5/352—Detecting R peaks, e.g. for synchronising diagnostic apparatus; Estimating R-R interval
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
- A61B5/363—Detecting tachycardia or bradycardia
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- 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
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Cardiology (AREA)
- Physics & Mathematics (AREA)
- Animal Behavior & Ethology (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Pathology (AREA)
- Artificial Intelligence (AREA)
- Signal Processing (AREA)
- Psychiatry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Power Engineering (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 invention discloses an arrhythmia classification method based on a convolutional neural network and a gated cyclic unit, which is implemented according to the following steps: step 1, selecting electrocardiogram data of an MIT-BIH arrhythmia database as a data set of the invention; step 2, preprocessing the electrocardio data set selected in the step 1; step 3, segmenting the electrocardio data preprocessed in the step 2; constructing a classification network model; step 4, training the network model by using partial beat data segments of the segmented electrocardiogram data in the step 3 as a training set; and 5, inputting the rest electrocardio data segments of the segmented electrocardio data in the step 3 into the trained network model based on the convolutional neural network and the gated circulation unit as a test set, and finally outputting the classification result of the electrocardio signal segments. The problems of disappearance of network gradients and low classification accuracy of partial arrhythmia diseases in the prior art are solved.
Description
Technical Field
The invention belongs to the technical field of medical image processing and signal processing, and relates to an arrhythmia classification method based on a convolutional neural network and a gated cyclic unit.
Background
Cardiovascular diseases are one of the important diseases that are currently seriously threatening human life and health. Cardiovascular diseases (CVDs) are the leading cause of death worldwide as reported by the world health organization in 2020, with more deaths annually from cardiovascular diseases than any other disease. So far, the number of deaths caused by cardiovascular diseases in China reaches 1790.0 ten thousands. Arrhythmia is an important group of cardiovascular diseases, can be singly attacked or can be accompanied with other cardiovascular diseases, even sudden attack causes sudden death, and the arrhythmia diseases are generally diagnosed by using electrocardiogram clinically. Therefore, timely extraction and correct classification of electrocardiogram signal features are important for early discrimination of arrhythmia and diagnosis, treatment and prevention of cardiovascular diseases.
The research based on the automatic identification and classification of the computer-aided electrocardiosignals draws attention, and a plurality of researchers deeply research the problem, but the research is still a challenging task. First, a data preprocessing stage. The lack of effective filtering operations for the original cardiac electrical signal. Because of weak electrocardiosignal, the internal noise components are diverse, and part of the noise is similar to P wave and T wave, and needs to be reasonably decomposed and suppressed. Secondly, the electrocardiosignal is segmented. The data volume can be reduced by dividing the data in fixed time, but the situation that the input fragment is inconsistent with the label of the time is generated, and the loss or redundancy of the electrocardio information exists between one beat and another beat; finally, the existing classification method is comprehensive, but the problem of long-distance dependence on the extracted sequence electrocardio characteristic data cannot be well processed, and the phenomena of large depth of a designed network model, difficult tuning and gradient disappearance also exist. In order to accurately and efficiently classify and identify arrhythmia diseases, the current mainstream methods can be divided into two categories, namely a traditional machine learning method and a deep learning method. In the conventional method, a heartbeat classification method based on a support vector machine is proposed. Some people also propose classification of cardiac beats based on particle swarm optimization and radial basis function neural networks. A Bayesian network heart beat classification method based on a decision threshold is proposed, and the classification precision is optimized. However, the conventional machine learning-based method is very dependent on artificial features, and requires considerable professional knowledge of pathology and signalology, so that the difficulty is increased for designing an algorithm, high-quality features are difficult to extract correspondingly, the classification accuracy is low, and misdiagnosis or missed diagnosis is easily caused.
In recent years, deep learning techniques have shown excellent performance in pattern recognition applications. Therefore, the study of classification of electrocardiograms based on the deep learning related technology becomes a focus of attention of researchers and engineers. Many scholars have done a lot of work on the study of electrocardiographic classification using deep learning techniques. Deep Belief Networks (DBNs) have been proposed to automatically extract features from ECG resampling, the DBN stacking using two types of limited boltzmann (RBM), and adjusting RBM parameters using two algorithms, contrast divergence and continuous contrast divergence. A method based on a one-dimensional convolutional neural network has been proposed to classify 5 types of arrhythmia signals. The 2D-CNN is also used for classifying the ECG arrhythmia, the method does not need to filter the original electrocardiosignal, well converts the time domain characteristic of the electrocardiosignal into the time-frequency domain characteristic for classifying the electrocardiosignal, and the final classification performance is outstanding. Some researchers have proposed a new classification algorithm for patient-specific electrocardiogram based on Recurrent Neural Network (RNN), which can learn the time correlation from the samples of electrocardiogram signals and classify the electrocardiographic beats of different heart rhythms. Although the classification accuracy performance is improved, the problems of complex network model, gradient loss, difficult optimization and the like are inevitably brought.
Disclosure of Invention
The invention aims to provide an arrhythmia classification method based on a convolutional neural network and a gated cyclic unit, which solves the problems of network gradient disappearance and low classification accuracy of partial arrhythmia diseases in the prior art.
The technical scheme adopted by the invention is that the arrhythmia classification method based on the convolutional neural network and the gated cyclic unit is implemented according to the following steps:
step 2, preprocessing the electrocardio data selected in the step 1;
step 3, segmenting the electrocardio data preprocessed in the step 2, and constructing a classification network model;
step 4, training a network model by using the electrocardio data segmented in the step 3;
and 5, inputting the electrocardiogram data segments to be tested into the network model trained in the step 3 and based on the convolutional neural network and the gated circulation unit, and finally outputting the classification result of the electrocardiogram signal segments.
The invention is also characterized in that:
step 2 is specifically carried out as follows:
step 2.1, reading original electrocardio data in the selected database;
step 2.2, performing noise suppression on the original electrocardio data read in the step 2.1 by utilizing wavelet 9-level hierarchical characteristics, and when analyzing the discretized non-stationary electrocardio signals, expressing any signal f (t) by using a multi-resolution analysis formula as follows:
wherein the content of the first and second substances,is the projection of f (t) in scale space, is a smooth approximation of f (t), whereIs a scale function, c0,kIs a scale factor;is the projection of (t) in wavelet space, wherej,k(t) is a wavelet function, selected using Daubechies5(db5) as the wavelet function, dj,kIs a wavelet coefficient, k is a position coefficient,phi and phij,k(t) the calculation formulas are respectively as follows:
φj,k(t)=2j/2φ(2jt-k) (3);
c0,kand dj,kThe expansion coefficient calculation formulas of (a) are respectively as follows:
step 2.3, the soft threshold function is used for suppressing and eliminating the noise, and the mathematical formula is expressed as follows:
wherein, wj,kIs a signal value after scale decomposition, w'j,kIn order to suppress the signal value after noise elimination, j is the order, k is the position coefficient, and the threshold value lambda satisfiesSigma is a noise standard deviation, and N is a signal length;
and 2.4, performing inverse transformation on the signal transformed in the step 2.3 to obtain the electrocardiosignal subjected to noise suppression.
The specific process of segmenting the electrocardiographic data preprocessed in the step 2 in the step 3 is as follows:
step 3.1.1, obtaining the position of the R wave crest and a corresponding label;
step 3.1.2, selecting the front 144 sampling points to the rear 180 sampling points of the R wave crest as a beat;
and 3.1.3, resampling each beat to 250 after segmentation, and using the resampled beat as the input of a subsequent network model.
The construction of the classification network model in the step 3 is specifically implemented as follows:
step 3.2.1, building a convolution neural network, wherein the convolution neural network comprises a convolution structure and a full connection layer;
the first layer of convolution structure consists of 16 convolution kernels with the size of 21 multiplied by 1, and the step length is 1; the convolution operation is followed by a batch normalization and a ReLU activation function for operation;
the subsequent convolution structure consists of 8 convolution kernels of size 25 × 1 with step size 1; the convolution operation is followed by a batch normalization and a ReLU activation function for operation;
respectively adding the residual block I and the residual block II with two structures into a convolutional neural network to form a convolutional neural network part based on a convolutional neural network and a gated cyclic unit network model:
step 3.2.2, adding a residual block I after the convolution structure of the convolution neural network built in the step 3.2.1 to form a jump connection structure, wherein the formula is as follows:
xl+1=xl+F(xl+Wl) (7);
wherein x isl+1Is the convolution result of the (l + 1) th convolutional layer, xlAs a result of convolution of the first convolutional layer, wlIs the weight of the first convolutional layer, F (x)l+Wl) Is a residual error part; adding the residual block into a convolution network to form a jump connection structure;
3.2.3, adding the residual block II to the next layer of the two layers of the residual blocks I in the convolutional neural network to form a jump connection structure again, and finally completing the construction of the whole convolutional network;
step 3.2.4, inputting the characteristic information of the step 3.2.3 into a convolution structure, and fusing the electrocardio characteristic information;
step 3.2.5, adding a gated cyclic unit with an output spatial dimension of 32 after the convolution structure consisting of 8 convolution kernels of size 25 × 1 in step 3.2.1; the gated loop unit consists of two gates, a reset gate and an update gate, respectively, formulated as follows:
zt=σ(W*[ht-1,xt]) (8);
rt=σ(W*[ht-1,xt]) (9);
yt=σ(W*ht) (12);
wherein x istRepresenting the characteristic of the input ECG sequence, ztIndicating a gating signal (update gate), rtA reset signal is indicated which is a signal that is reset,representing the current content to be memorized, W representing a weight value, ht-1From the previous layer of key hidden information, h, representing memorytRepresenting the current layer key hidden information of memory, sigma representing SAn igmoid activation function, tanh denotes an activation function, ytRepresenting the output current layer signal characteristic information, representing a matrix multiplication,representing a multiplication of the respective position element with the respective position element.
Step 4 is specifically implemented as follows:
step 4.1, the electrocardio beats processed in the step 3.1 and a label corresponding to each beat are used as a data set; training a model by using 10-fold cross validation, inputting data of a training set into a convolutional neural network and gating cycle unit network-based model to train the model;
and 4.2, training a network model consisting of a convolutional neural network and a gate control cycle unit by using a 10-fold cross validation method.
Step 5 is specifically implemented as follows: firstly, adopting data of a subset remained in each compromise except a training set as electrocardiogram data to be tested; then, inputting each to-be-tested electrocardio beat into a trained network model based on a convolutional neural network and a gated circulation unit, performing 2 times of convolution operation, 2 times of residual module 1 operation, 1 time of residual module 2 operation and gated circulation unit operation, finally performing category prediction on each beat on a test set through softmax, and outputting a final classification result.
The invention has the beneficial effects that: the arrhythmia classification method based on the convolutional neural network and the gated cyclic unit solves the problems of network gradient disappearance and low classification accuracy of partial arrhythmia diseases in the prior art. The short-circuit connection structure is fully utilized in the first-stage convolutional neural network, and a gated cycle unit (GRU) part is arranged next to the first-stage convolutional neural network, so that the parameters of the part are less, the training speed is higher, the electrocardio characteristic information at the last moment can be stored, the electrocardio characteristic information cannot be eliminated along with time, and the subsequent classification performance is guaranteed. Furthermore, the gated loop unit (GRU) has an update gate and a reset gate, and these two gating vectors determine which signal characteristics can be finally used as the output of the gated loop unit. The problems of gradient disappearance caused by large network depth and dependency of the front and rear characteristic information of the electrocardiosignals are solved, the classification performance is improved to be optimal, and the algorithm robustness is also enhanced.
Drawings
FIG. 1 is a schematic flow chart of the arrhythmia classification method based on a convolutional neural network and a gated cyclic unit according to the present invention;
FIG. 2 is a schematic diagram of the structure of the basic residual block in the arrhythmia classification method based on the convolutional neural network and the gated cyclic unit according to the present invention;
FIG. 3 is a schematic diagram of the overall structure of the arrhythmia classification method based on the convolutional neural network and the gated cyclic unit according to the present invention;
FIG. 4 is a schematic structural diagram of a residual block I in the arrhythmia classification method based on a convolutional neural network and a gated cyclic unit according to the present invention;
FIG. 5 is a schematic diagram of a residual block II structure in the arrhythmia classification method based on a convolutional neural network and a gated cyclic unit according to the present invention;
FIG. 6 is a schematic structural diagram of a gated cyclic unit in the arrhythmia classification method based on a convolutional neural network and the gated cyclic unit according to the present invention;
FIG. 7 is a schematic flowchart of network model training based on a convolutional neural network and a gated cycle unit in the arrhythmia classification method based on a convolutional neural network and a gated cycle unit according to the present invention;
FIG. 8 is a schematic diagram of cross validation of 10 folds in the arrhythmia classification method based on convolutional neural network and gated cyclic unit according to the present invention;
FIG. 9 is a graph of the accuracy of the arrhythmia classification method based on the convolutional neural network and the gated cyclic unit of the present invention and the prior art 1-dimensional convolutional neural network method;
FIG. 10 is a graph of loss values for the arrhythmia classification method based on convolutional neural network and gated cyclic unit of the present invention and the prior art 1-dimensional convolutional neural network method.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.
The arrhythmia classification method based on the convolutional neural network and the gated cyclic unit is implemented according to the following steps as shown in fig. 1:
Step 2, preprocessing the electrocardio data selected in the step 1; preprocessing the electrocardio data of the database, including reading the electrocardio data, filtering, and removing unstable data from head to tail;
step 2 is specifically carried out as follows:
step 2.1, reading original electrocardio data in the selected database;
step 2.2, performing noise suppression by using the wavelet 9-level grading characteristic; when analyzing the discretization non-stationary electrocardiosignal, the arbitrary signal f (t) can be expressed by a multi-resolution analysis formula as follows:
wherein the content of the first and second substances,is the projection of f (t) in scale space, is a smooth approximation of f (t), whereIs a scale function, c0,kAre approximations or scale coefficients;is the projection of f (t) in wavelet space, is a supplement to the details of f (t), where phij,k(t) isWavelet function, selected from Daubechies5(db5) as wavelet function, dj,kDetail or wavelet coefficients. k is a position coefficient.Phi and phij,k(t) the calculation formulas are respectively as follows:
φj,k(t)=2j/2φ(2jt-k) (3);
c0,kand dj,kThe expansion coefficient calculation formulas of (a) are respectively as follows:
dj,k=<f(t),φj,k(t)>=∫f(t)φj,k(t)d(t) (5);
the noise is then suppressed using a soft threshold function, whose mathematical formula is expressed as follows:
wherein, wj,kIs a signal value after scale decomposition, w'j,kIn order to suppress the signal value after noise elimination, j is the order, k is the position coefficient, and the threshold value lambda satisfiesSigma is a noise standard deviation, and N is a signal length;
and finally, inversely transforming the transformed signal back to obtain the electrocardiosignal after noise suppression.
And 2.3, adopting zero-mean subtraction for the baseline drift situation of the electrocardiosignals, namely subtracting the mean value of each dimension from the filtered electrocardio data of each dimension to inhibit the electrocardio data of the baseline drift position.
Step 3, segmenting the electrocardiogram data preprocessed in the step 2, and then resampling the electrocardiogram segments to 250, wherein a data set consists of the electrocardiogram data segments; constructing a classification network model, and firstly constructing a basic convolutional neural network for analyzing the characteristics of the electrocardio segments;
in the step 3, the segmentation of the electrocardiographic data preprocessed in the step 2 is specifically implemented as follows:
3.1.1, acquiring the position of the R wave crest and a corresponding label;
3.1.2, selecting the front 144 sampling points to the rear 180 sampling points of the R wave crest as a beat;
3.1.3, resampling to 250 for each beat after segmentation, and using the resampled sample as the input of a subsequent network model.
As shown in fig. 3, the construction of the classification network model in step 3 is specifically implemented as follows:
step 3.2.1, building a basic convolutional neural network; the convolutional neural network is similar to the VGG network in structure and comprises a convolutional structure and a full connection layer:
the first layer of convolution structure consists of 16 convolution kernels of size 21 x 1 with a step size of 1. The convolution operation is followed by a batch normalization and a ReLU activation function for operation;
the subsequent convolution structure consists of 8 convolution kernels of size 25 × 1 with a step size of 1. The convolution operation is followed by a batch normalization and a ReLU activation function for operation;
the full connection layer is used for performing full connection on the neurons of the upper layer to realize nonlinear combination of characteristics, and is used for classifying 5 types of arrhythmia diseases, so that the parameter setting is 5;
the output values of the multi-classification can be converted into relative probabilities by using an activation function softmax, and the cumulative sum of the values is 1;
adding the residual blocks with two structures into a convolutional neural network to form a convolutional neural network part based on a convolutional neural network and a gated cyclic unit network model, and specifically comprising the following steps:
determining a structural unit of the residual block I; the structural unit is used for acquiring the characteristic information of the electrocardio data and better positioning the wave crest characteristic to obtain the characteristic data of the electrocardio wave band;
determining a structural unit of the residual block II, wherein the structural unit is used for selecting the element characteristics with the highest activation degree to form the whole convolutional neural network part;
step 3.2.2, adding a residual block I after the convolution layer of the convolutional neural network to form a jump connection structure; as shown in fig. 2, the basic residual structure is constructed as a graph. The formula is as follows:
xl+1=xl+F(xl+Wl) (7);
wherein x isl+1Is the convolution result of the (l + 1) th convolutional layer, xlAs a result of convolution of the first convolutional layer, wlIs the weight of the first convolutional layer, F (x)l+Wl) Is a residual error part; adding the residual block into a convolution network to form a jump connection structure; a residual block I is added after a convolution layer of the convolution neural network to form a jump connection structure, so that the characteristic information of the electrocardiogram data is not lost, the peak characteristic can be better positioned, the method is equivalent to a 'soft down sampling' mode, and smooth transition between elements is fully utilized; the structure of the residual block I in the network model is shown in fig. 4, and it can be seen from the structure diagram that a layer of convolution operation is adopted on the main line to be added with another branch, and finally, the ReLU activation is performed, so that some negative weight values are well filtered, and the following characteristic information redundancy is reduced.
And 3.2.3, adding the residual block II to the next layer of the two layers of the residual blocks I in the convolutional neural network, and forming a jump connection structure again to complete the construction of the convolutional neural network part based on the convolutional neural network and the gated cyclic unit network model. The module is used for extracting input feature information, mainly adopts a structure similar to hard down sampling, can select features with high activation degree and provides good feature information for a subsequent gating circulation unit;
the residual block II is shown in fig. 5, and the specific process is as follows:
first, the residual block II is determined. The residual block II is used for extracting input feature information, mainly adopts a structure similar to hard down sampling, can select features with high activation degree, and is beneficial to removing redundant information and subsequent classification.
Then, information is fused using a jump connection. Followed by the use of the ReLU activation function to improve the expressive power of the feature.
The invention forms the convolution neural network part of the network model by two residual error networks, and fully utilizes the characteristics of the two structures to improve effective sequence characteristic information for the positioning and extraction of the characteristics and the subsequent input as a gating circulation network. As shown in fig. 3, two residual structures of the design are represented in the figure by a residual block I and a residual block II.
And 3.2.4, inputting the characteristic information of the step 3.2.3 into a convolution structure, and further fusing the electrocardio characteristic information.
At step 3.2.5, the gated loop elements are shown in FIG. 6. Taking the output of the step 3.2.4 as the input of the gating circulation unit, the structure can selectively retain the key electrocardiogram information at the last moment and eliminate some redundant information, whereinThe current electrocardio characteristics can be well memorized. The formula is as follows:
zt=σ(W*[ht-1,xt]) (8);
rt=σ(W*[ht-1,xt]) (9);
yt=σ(W*ht) (12);
wherein x istRepresenting the characteristic of the input ECG sequence, ztIndicating a gating signal (update gate), rtA reset signal is indicated which is a signal that is reset,representing the current content to be memorized, W representing a weight value, ht-1From the previous layer of key hidden information, h, representing memorytRepresenting the key hidden information of the current layer of memory, sigma representing Sigmoid activation function, tanh representing an activation function, ytRepresenting the output current layer signal characteristic information, representing a matrix multiplication,representing a multiplication of the respective position element with the respective position element.
Step 3.2.6, the full connection layer is used for carrying out full connection on the neurons of the previous layer to realize nonlinear combination of characteristics; and finally outputting 5 category prediction probability values by using a softmax activation function.
Step 4, training a network model by using the electrocardio data segmented in the step 3;
as shown in fig. 7, step 4 is specifically performed as follows:
step 4.1, the electrocardio beats processed in the step 3 and the corresponding labels are used as a data set; aiming at the problem of unbalance of various electrocardio sample data in an electrocardio data set, a 10-fold cross validation method is used for training and validating the network performance. The data set is divided into 10 parts, 9 parts of the data set are extracted as a training set, and the training set is input into a designed network model based on a convolutional neural network and a gated loop unit. Just as in the flow shown in fig. 8, firstly, segmented data is required to be used as a data set; dividing the data set into 10 subsets, wherein the distributed data amount of each subset is approximately equal; 9 subsets of these were used as input data for the network model.
Step 4.2, training the network model based on the convolutional neural network and the gated loop unit by using the training set data divided by 10-fold cross validation:
firstly, inputting 9 training subsets of a first compromise into a network model, wherein each beat in training data corresponds to a label; wherein, the size of each beat input into the network model is (250, 1), and the input beat needs to be subjected to front and back two-layer convolution operation; 2 residual blocks I and 1 residual block II; a gated cycle cell structure; finally, 5 classifications are carried out on the beats through a softmax activation function, and the output 5 classification probability value is the classification result predicted by the network;
for the measurement of model loss, it is the difference between the predicted value of the model and the true value of a specific sample, and when facing the task of multi-classification or multi-label, the two measurement values are used for evaluation, one is the cross entropy loss function (sparse _ cross _ entropy) and the other is the sparse _ cross entropy loss function (sparse _ cross _ entropy). The latter is selected to measure the model loss degree because the sparse cross entropy loss function is used for a target result which is an integer value, a matrix does not need to be stored, and the use of a memory space can be reduced; the loss function is formulated as follows:
wherein n is the number of samples, m is the number of categories,to predict the output value, yi,jIs an actual value;
parameters generated in the calculation process of each operation are calculated through a back propagation algorithm to reduce the loss function value, and the parameters generated in the back propagation calculation are the weight of the network and are marked as W.
Secondly, inputting the other 9-fold electrocardiogram data into the network model, repeating the operation of the previous step, continuously recording the loss value of each fold, and finally updating the corresponding weight W after the loss value of each fold is kept stable; and after the data of all the folds are trained, obtaining the training result of each fold, and obtaining the trained model based on the convolutional neural network and the gated cyclic unit network.
And 5, inputting the electrocardiogram data segments to be tested into the trained network model based on the convolutional neural network and the gated circulation unit, and finally outputting the classification result of the electrocardiogram signal segments. Step 5 is specifically implemented as follows: firstly, adopting a subset remained in each compromise except a training set as electrocardiogram data to be tested; and then, inputting the electrocardio beats to be tested in each compromise into a trained network model, performing 2 times of convolution operation, 2 times of residual block I operation, 1 time of residual block II operation and gate control cycle unit operation, finally performing class prediction on each beat in the test set through a softmax activation function, and outputting a final classification result.
After the classification result of the electrocardio beats on each test set is obtained, a confusion matrix is used for displaying which part of the classification model is confused during prediction, and the method can be used for evaluating the quality of a classification algorithm. As shown in Table 1, it can be seen from the table that the test set is 9945, the data on the diagonal is correctly classified for each class, and the data on the off-diagonal is incorrectly classified, which is very small. The number of misclassifications for normal beats, left bundle branch block beats, right bundle branch block beats, atrial premature beats, ventricular premature beats is quite small, 35, 24, 11, 5, 4 respectively.
TABLE 1
Inputting the preprocessed electrocardiogram data segment to be tested into a trained network model based on a convolutional neural network and a gating cycle unit;
and sequentially carrying out 2 times of convolution operation, 2 times of residual block I operation, 1 time of residual block II operation and gate control circulation unit operation on the basis of the convolution neural network and the gate control circulation unit model, finally giving a probability score to each beat by using a softmax activation function, and outputting the classification result and the disease category.
The confusion matrix is used for representing the actual category and the obtained result after classification, and displaying which part of the classification model is confused when the classification model is used for prediction, so that the confusion matrix can be used for evaluating the quality of a classification algorithm;
the evaluation classification algorithm also requires accuracy, sensitivity, precision and specificity as measures.
Table 2 shows the performance index results of the present invention. As can be seen from the table, the classification accuracy of the 5 classes is close to 100%; sensitivity is 90.87% only for the left bundle branch block heartbeat, other categories have appreciable results; in terms of accuracy, the results of normal beats and atrial premature beats reach 99.54% and 99.50%, with almost no recognition errors; the specificity results show that there are few cases of 5 types of misclassifications. Further analysis of the table revealed that the average accuracy of classifying normal beats, left bundle branch block beats, right bundle branch block beats, atrial premature beats and ventricular premature beats was 99.69%, the average sensitivity was 97.53%, the average accuracy was 97.51%, and the average specificity was 99.63%. From these index results, the classification performance of arrhythmia classification method based on convolutional neural network and gated cyclic unit is outstanding.
TABLE 2
Table 3 is a comparison result chart of the network model based on the convolutional neural network and the gated cyclic unit of the present invention and the existing 1-dimensional convolutional neural network. In the figure, the second row shows the classification average accuracy and the average loss value of the existing similar arrhythmia classification method based on the 1-dimensional convolutional neural network, and the third row shows the classification average accuracy and the average loss value of the arrhythmia classification method based on the convolutional neural network and the gated cyclic unit.
TABLE 3
FIGS. 9 and 10 are an accuracy curve and a loss curve for the classification of ECG arrhythmias using the present invention and prior art 1-dimensional convolutional neural network method, the solid line representing the experimental data results of the present invention. It is obvious from fig. 9 that, under the same number of iterations, learning rate and batch, the accuracy curve of the method starts to be above 0.7, and gradually stabilizes to be near 0.99 from the back. Meanwhile, the convergence speed of the precision curve of the proposed model is faster than that of a 1-dimensional convolutional neural network model, and the final classification precision is far higher than that of the existing 1D-CNN model. It can be seen from fig. 10 that, at the same learning rate and batch, although the loss curve of the method of the present invention is decreased after rising at the beginning, the subsequent loss value gradually decreases and fluctuates to about 0.04. Meanwhile, the convergence speed of the loss curve of the model is higher than that of the 1D-CNN model, and the final average loss value is lower than that of the 1-dimensional convolution neural model.
Step 5 is specifically implemented as follows: firstly, adopting data of a subset remained in each compromise except a training set as electrocardiogram data to be tested; then, inputting each to-be-tested electrocardio beat into a trained network model based on a convolutional neural network and a gate control cycle unit, performing 2 times of convolution operation, 2 times of residual module 1 operation, 1 time of residual module 2 operation and gate control cycle unit operation, finally performing class prediction on each beat on a test set through softmax, and outputting a final classification result.
Inputting the preprocessed electrocardiogram data segment to be tested into a trained network model based on a convolutional neural network and a gating cycle unit;
the model of the invention is adopted to carry out convolution operation for 2 times, residual error module 1 operation for 2 times, residual error module 2 operation for 1 time and gate control circulation unit operation in sequence, and finally, probability score is given to each beat by utilizing softmax activation function, and the classification result and the disease category are output.
Inputting the electrocardiogram fragments to be classified into the trained network model to obtain a classification result, and then further comprising:
the confusion matrix is used for representing the actual category and the obtained result after classification, and displaying which part of the classification model is confused when the classification model is used for prediction, so that the confusion matrix can be used for evaluating the quality of a classification algorithm;
the evaluation classification algorithm also requires accuracy, sensitivity, precision and specificity as measures. The ratio of the number of correctly classified samples to the total number of test samples is expressed for Accuracy (Accuracy). The formula is as follows:
sensitivity (Sensitivity) indicates the proportion of all samples that are actually positive, correctly classified as arrhythmic. The formula is as follows:
the Precision (Precision) represents the proportion of all samples predicted to be arrhythmic that are correctly classified. The formula is as follows:
the Specificity represents the proportion of samples predicted to be correct for a normal rhythm among all samples actually for a normal rhythm. The formula is as follows:
in each formula, TP is true positive and represents the correct classification into 4 kinds of arrhythmia diseases; TN is true negative, indicating correct classification as normal; FP was false positive, indicating incorrect classification as 4 arrhythmic diseases; FN is false negative, indicating incorrect classification as normal.
The arrhythmia classification method based on the convolutional neural network and the gated cyclic unit solves the problems of network gradient disappearance and low classification accuracy of partial arrhythmia diseases in the prior art. The convolutional neural network in the network model adopts a residual error structure with short-circuit connection, and is followed by a gating cycle unit (GRU) part, so that the parameters of the part are less, the training speed is higher, the electrocardio characteristic information at the last moment can be stored, the electrocardio characteristic information cannot be eliminated along with time, and the subsequent classification performance is guaranteed. Furthermore, the gated loop unit (GRU) has an update gate and a reset gate, and these two gating vectors determine which signal characteristics can be finally used as the output of the gated loop unit. The problems of gradient disappearance caused by large network depth and dependency of the front and rear characteristic information of the electrocardiosignals are solved, the classification performance is improved to be optimal, and the algorithm robustness is also enhanced.
Claims (6)
1. The arrhythmia classification method based on the convolutional neural network and the gated cyclic unit is characterized by comprising the following steps:
step 1, selecting electrocardiogram data of an MIT-BIH arrhythmia database;
step 2, preprocessing the electrocardio data selected in the step 1;
step 3, segmenting the electrocardio data preprocessed in the step 2, and constructing a classification network model;
step 4, training a network model by using the electrocardio data segmented in the step 3;
and 5, inputting the electrocardiogram data segments to be tested into the network model trained in the step 3 and based on the convolutional neural network and the gated circulation unit, and finally outputting the classification result of the electrocardiogram signal segments.
2. The method for classifying arrhythmia based on convolutional neural network and gated cyclic unit as claimed in claim 1, wherein the step 2 is implemented as follows:
step 2.1, reading original electrocardio data in the selected database;
step 2.2, performing noise suppression on the original electrocardio data read in the step 2.1 by utilizing wavelet 9-level hierarchical characteristics, and when analyzing the discretized non-stationary electrocardio signals, expressing any signal f (t) by using a multi-resolution analysis formula as follows:
wherein the content of the first and second substances,is a projection of (t) in scale space and is a smooth approximation of (f), (t), whereIs a scale function, c0,kIs a scale factor;is the projection of (t) in wavelet space, wherej,k(t) is a wavelet function, selected using Daubechies5 as the wavelet function, dj,kIs a wavelet coefficient, k is a position coefficient,phi and phij,k(t) the calculation formulas are respectively as follows:
c0,kand dj,kThe expansion coefficient calculation formulas of (a) are respectively as follows:
step 2.3, the soft threshold function is used for suppressing and eliminating the noise, and the mathematical formula is expressed as follows:
wherein, wj,kIs a signal value after scale decomposition, w'j,kIn order to suppress the signal value after noise elimination, j is the order, k is the position coefficient, and the threshold value lambda satisfiesSigma is a noise standard deviation, and N is a signal length;
and 2.4, performing inverse transformation on the signal transformed in the step 2.3 to obtain the electrocardiosignal subjected to noise suppression.
3. The arrhythmia classification method based on the convolutional neural network and the gated cyclic unit as claimed in claim 1, wherein the specific process of segmenting the electrocardiographic data preprocessed in step 2 in step 3 is as follows:
step 3.1.1, obtaining the position of the R wave crest and a corresponding label;
step 3.1.2, selecting the front 144 sampling points to the rear 180 sampling points of the R wave crest as a beat;
and 3.1.3, resampling each beat to 250 after segmentation, and using the resampled beat as the input of a subsequent network model.
4. The arrhythmia classification method based on the convolutional neural network and the gated cyclic unit as claimed in claim 2, wherein the step 3 of constructing the classification network model is implemented as follows:
step 3.2.1, building a convolution neural network, wherein the convolution neural network comprises a convolution structure and a full connection layer;
the first layer of convolution structure consists of 16 convolution kernels with the size of 21 multiplied by 1, and the step length is 1; the convolution operation is followed by a batch normalization and a ReLU activation function for operation;
the subsequent convolution structure consists of 8 convolution kernels of size 25 × 1 with step size 1; the convolution operation is followed by a batch normalization and a ReLU activation function for operation;
respectively adding the residual block I and the residual block II with two structures into a convolutional neural network to form a convolutional neural network part based on a convolutional neural network and a gated cyclic unit network model:
step 3.2.2, adding a residual block I after the convolution structure of the convolution neural network built in the step 3.2.1 to form a jump connection structure, wherein the formula is as follows:
xl+1=xl+F(xl+Wl) (7);
wherein x isl+1Is the convolution result of the (l + 1) th convolutional layer, xlAs a result of convolution of the first convolutional layer, wlIs the weight of the first convolutional layer, F (x)l+Wl) Is a residual error part; adding the residual block into a convolution network to form a jump connection structure;
3.2.3, adding the residual block II to the next layer of the two layers of the residual blocks I in the convolutional neural network to form a jump connection structure again, and finally completing the construction of the whole convolutional network;
step 3.2.4, inputting the characteristic information of the step 3.2.3 into a convolution structure, and fusing the electrocardio characteristic information;
step 3.2.5, adding a gated cyclic unit with an output spatial dimension of 32 after the convolution structure consisting of 8 convolution kernels of size 25 × 1 in step 3.2.1; the gated loop unit consists of two gates, a reset gate and an update gate, respectively, formulated as follows:
zt=σ(W*[ht-1,xt]) (8);
rt=σ(W*[ht-1,xt]) (9);
yt=σ(W*ht) (12);
wherein x istRepresenting the characteristic of the input ECG sequence, ztRepresents a gating signal, rtA reset signal is indicated which is a signal that is reset,representing the current content to be memorized, W representing a weight value, ht-1From the previous layer of key hidden information, h, representing memorytRepresenting the key hidden information of the current layer of memory, sigma representing Sigmoid activation function, tanh representing an activation function, ytRepresenting the output current layer signal characteristic information, representing a matrix multiplication,representing a multiplication of the respective position element with the respective position element.
5. The method for classifying arrhythmia based on convolutional neural network and gated cyclic unit as claimed in claim 1, wherein the step 4 is implemented as follows:
step 4.1, the electrocardio beats processed in the step 3.1 and a label corresponding to each beat are used as a data set; training a model by using 10-fold cross validation, inputting data of a training set into a convolutional neural network and gating cycle unit network-based model to train the model;
and 4.2, training a network model consisting of a convolutional neural network and a gate control cycle unit by using a 10-fold cross validation method.
6. The method for classifying arrhythmia based on convolutional neural network and gated cyclic unit as claimed in claim 5, wherein the step 5 is implemented as follows: firstly, adopting data of a subset remained in each compromise except a training set as electrocardiogram data to be tested; then, inputting each to-be-tested electrocardio beat into a trained network model based on a convolutional neural network and a gated circulation unit, performing 2 times of convolution operation, 2 times of residual module 1 operation, 1 time of residual module 2 operation and gated circulation unit operation, finally performing category prediction on each beat on a test set through softmax, and outputting a final classification result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110909940.2A CN113768514B (en) | 2021-08-09 | 2021-08-09 | Arrhythmia classification method based on convolutional neural network and gating circulation unit |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110909940.2A CN113768514B (en) | 2021-08-09 | 2021-08-09 | Arrhythmia classification method based on convolutional neural network and gating circulation unit |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113768514A true CN113768514A (en) | 2021-12-10 |
CN113768514B CN113768514B (en) | 2024-03-22 |
Family
ID=78837082
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110909940.2A Active CN113768514B (en) | 2021-08-09 | 2021-08-09 | Arrhythmia classification method based on convolutional neural network and gating circulation unit |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113768514B (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105748063A (en) * | 2016-04-25 | 2016-07-13 | 山东大学齐鲁医院 | Intelligent arrhythmia diagnosis method based on multiple-lead and convolutional neural network |
KR20190141326A (en) * | 2018-06-14 | 2019-12-24 | 한국과학기술원 | Method and Apparatus for ECG Arrhythmia Classification using a Deep Convolutional Neural Network |
WO2020047750A1 (en) * | 2018-09-04 | 2020-03-12 | 深圳先进技术研究院 | Arrhythmia detection method and apparatus, electronic device, and computer storage medium |
CN110890155A (en) * | 2019-11-25 | 2020-03-17 | 中国科学技术大学 | Multi-class arrhythmia detection method based on lead attention mechanism |
KR20200071183A (en) * | 2018-12-10 | 2020-06-19 | 순천향대학교 산학협력단 | Arrhythmia classification system and method using deep neural network |
US20200260980A1 (en) * | 2017-11-27 | 2020-08-20 | Lepu Medical Technology (Bejing) Co., Ltd. | Method and device for self-learning dynamic electrocardiography analysis employing artificial intelligence |
CN111626114A (en) * | 2020-04-20 | 2020-09-04 | 哈尔滨工业大学 | Electrocardiosignal arrhythmia classification system based on convolutional neural network |
CN111657926A (en) * | 2020-07-08 | 2020-09-15 | 中国科学技术大学 | Arrhythmia classification method based on multi-lead information fusion |
-
2021
- 2021-08-09 CN CN202110909940.2A patent/CN113768514B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105748063A (en) * | 2016-04-25 | 2016-07-13 | 山东大学齐鲁医院 | Intelligent arrhythmia diagnosis method based on multiple-lead and convolutional neural network |
US20200260980A1 (en) * | 2017-11-27 | 2020-08-20 | Lepu Medical Technology (Bejing) Co., Ltd. | Method and device for self-learning dynamic electrocardiography analysis employing artificial intelligence |
KR20190141326A (en) * | 2018-06-14 | 2019-12-24 | 한국과학기술원 | Method and Apparatus for ECG Arrhythmia Classification using a Deep Convolutional Neural Network |
WO2020047750A1 (en) * | 2018-09-04 | 2020-03-12 | 深圳先进技术研究院 | Arrhythmia detection method and apparatus, electronic device, and computer storage medium |
KR20200071183A (en) * | 2018-12-10 | 2020-06-19 | 순천향대학교 산학협력단 | Arrhythmia classification system and method using deep neural network |
CN110890155A (en) * | 2019-11-25 | 2020-03-17 | 中国科学技术大学 | Multi-class arrhythmia detection method based on lead attention mechanism |
CN111626114A (en) * | 2020-04-20 | 2020-09-04 | 哈尔滨工业大学 | Electrocardiosignal arrhythmia classification system based on convolutional neural network |
CN111657926A (en) * | 2020-07-08 | 2020-09-15 | 中国科学技术大学 | Arrhythmia classification method based on multi-lead information fusion |
Non-Patent Citations (2)
Title |
---|
燕婷: "基于深度学习的不平衡心电样本分类方法研究", 《中国优秀硕士学位论文全文数据库医药卫生科技辑》 * |
燕婷;王于超;金中一;: "基于CNN与GRU降噪自编码的心电信号分类方法", 软件导刊, no. 04 * |
Also Published As
Publication number | Publication date |
---|---|
CN113768514B (en) | 2024-03-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110890155B (en) | Multi-class arrhythmia detection method based on lead attention mechanism | |
CN111160139B (en) | Electrocardiosignal processing method and device and terminal equipment | |
CN109998525B (en) | Arrhythmia automatic classification method based on discriminant deep belief network | |
CN113274031B (en) | Arrhythmia classification method based on depth convolution residual error network | |
CN110534192B (en) | Deep learning-based lung nodule benign and malignant recognition method | |
CN110522444B (en) | Electrocardiosignal identification and classification method based on Kernel-CNN | |
CN111192245A (en) | Brain tumor segmentation network and method based on U-Net network | |
Petrozziello et al. | Deep learning for continuous electronic fetal monitoring in labor | |
Luo et al. | Multi-classification of arrhythmias using a HCRNet on imbalanced ECG datasets | |
CN113288163A (en) | Multi-feature fusion electrocardiosignal classification model modeling method based on attention mechanism | |
CN111329469A (en) | Arrhythmia prediction method | |
WO2022166158A1 (en) | System for performing long-term hazard prediction on hemodialysis complications on basis of convolutional survival network | |
CN113080993B (en) | Automatic arrhythmia analysis method based on improved multi-base classifier integration | |
CN112294342A (en) | Sleep staging method based on deep residual Mask-CCNN | |
CN115281688A (en) | Cardiac hypertrophy multi-label detection system based on multi-mode deep learning | |
CN116503673B (en) | Arrhythmia identification and detection method and system based on electrocardiogram | |
US20230225663A1 (en) | Method for predicting multi-type electrocardiogram heart rhythms based on graph convolution | |
CN111803059A (en) | Electrocardiosignal classification method and device based on time domain convolution network | |
CN112932433A (en) | Arrhythmia automatic classification method of electrocardiogram | |
CN112957054A (en) | 12-lead electrocardiosignal classification method based on channel attention grouping residual error network | |
CN113768515A (en) | Electrocardiosignal classification method based on deep convolutional neural network | |
CN115221926A (en) | Heart beat signal classification method based on CNN-GRU network model | |
Xie et al. | Intelligent analysis of premature ventricular contraction based on features and random forest | |
CN115177262A (en) | Heart sound and electrocardiogram combined diagnosis device and system based on deep learning | |
Wu et al. | A deep neural network ensemble classifier with focal loss for automatic arrhythmia classification |
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 |