CN111803059A - Electrocardiosignal classification method and device based on time domain convolution network - Google Patents
Electrocardiosignal classification method and device based on time domain convolution network Download PDFInfo
- Publication number
- CN111803059A CN111803059A CN202010607140.0A CN202010607140A CN111803059A CN 111803059 A CN111803059 A CN 111803059A CN 202010607140 A CN202010607140 A CN 202010607140A CN 111803059 A CN111803059 A CN 111803059A
- Authority
- CN
- China
- Prior art keywords
- network
- time
- training
- time domain
- feature extraction
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000012549 training Methods 0.000 claims abstract description 74
- 238000000605 extraction Methods 0.000 claims abstract description 40
- 238000013145 classification model Methods 0.000 claims abstract description 20
- 238000013528 artificial neural network Methods 0.000 claims abstract description 18
- 238000012360 testing method Methods 0.000 claims description 18
- 238000004590 computer program Methods 0.000 claims description 11
- 238000011156 evaluation Methods 0.000 claims description 9
- 238000012795 verification Methods 0.000 claims description 8
- 238000010200 validation analysis Methods 0.000 claims description 5
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 230000000747 cardiac effect Effects 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 7
- 238000004364 calculation method Methods 0.000 description 8
- 230000000875 corresponding effect Effects 0.000 description 7
- 238000013527 convolutional neural network Methods 0.000 description 6
- 238000002790 cross-validation Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 125000004122 cyclic group Chemical group 0.000 description 4
- 238000013135 deep learning Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000005070 sampling Methods 0.000 description 4
- 208000024172 Cardiovascular disease Diseases 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 208000017667 Chronic Disease Diseases 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 208000026106 cerebrovascular disease Diseases 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000002526 effect on cardiovascular system Effects 0.000 description 1
- 238000005516 engineering process Methods 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
- 210000002569 neuron Anatomy 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 230000005180 public health Effects 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention relates to the technical field of electrocardiosignal classification, and discloses an electrocardiosignal classification method based on a time domain convolution network, which comprises the following steps: acquiring electrocardiosignals, classifying the electrocardiosignals, and marking category labels to obtain a sample data set; adding a time domain convolution network structure in the one-dimensional convolution neural network to obtain a time domain feature extraction network based on the time domain convolution network; training the time domain feature extraction network by adopting the sample data set to obtain a classification model; and classifying the electrocardiosignals to be classified according to the classification model. The method has the technical effects of extracting the time domain characteristics of the electrocardiosignals, accurately classifying and having high classification efficiency.
Description
Technical Field
The invention relates to the technical field of electrocardiosignal classification, in particular to an electrocardiosignal classification method and device based on a time domain convolution network and a computer storage medium.
Background
At present, cardiovascular diseases are one of the common and frequent major chronic diseases, and the mortality rate is always high, so that the cardiovascular diseases become public health problems worldwide. The ECG signal reflecting cardiovascular and cerebrovascular disease states, namely the daily monitoring, analysis, identification and diagnosis of the electrocardiosignal, has high clinical research and development values. With the development and application of deep learning in the medical field, the work load of a doctor can be greatly reduced by utilizing the deep learning technology to assist the doctor in automatic diagnosis and identification of electrocardiosignals, the work efficiency of the doctor is improved, the cause of disease is efficiently obtained, and a foundation is provided for subsequent treatment.
At present, two methods for classifying electrocardiosignals are mainly used, one method is a traditional method for manually extracting characteristics, the steps of the method are complicated, and the universality of different data sets is poor; the other method which is used more is to train and automatically classify data by using deep learning, and the method is mostly based on one-dimensional convolutional neural network or cyclic neural network structures, such as RNN, LSTM and the like, to perform feature extraction, and finally uses softmax or sigmoid classifier to perform classification. The one-dimensional convolutional neural network has high characteristic extraction speed, but has poor characteristic extraction effect on time for a time domain signal of electrocardiosignals; although the cyclic neural network structure can take the time information characteristics of the electrocardiosignals into consideration, the calculation amount is large, and the recognition efficiency is low.
Disclosure of Invention
The invention aims to overcome the technical defects, provides an electrocardiosignal classification method and device based on a time domain convolution network and a computer storage medium, and solves the technical problem of central electric signal feature extraction in the prior art.
In order to achieve the technical purpose, the technical scheme of the invention provides an electrocardiosignal classification method based on a time domain convolution network, which comprises the following steps:
acquiring electrocardiosignals, classifying the electrocardiosignals, and marking category labels to obtain a sample data set;
adding a time domain convolution network structure in the one-dimensional convolution neural network to obtain a time domain feature extraction network based on the time domain convolution network;
training the time domain feature extraction network by adopting the sample data set to obtain a classification model;
and classifying the electrocardiosignals to be classified according to the classification model.
The invention also provides an electrocardiosignal classification device based on the time domain convolution network, which comprises a processor and a memory, wherein the memory is stored with a computer program, and when the computer program is executed by the processor, the electrocardiosignal classification method based on the time domain convolution network is realized.
The invention also provides a computer storage medium on which a computer program is stored, which, when executed by a processor, implements the time-domain convolutional-network-based cardiac signal classification method.
Compared with the prior art, the invention has the beneficial effects that: the method is characterized in that a time domain convolution network structure, namely a TCN network layer, is inserted on the basis of the traditional one-dimensional convolution neural network to obtain a time domain feature extraction network, and then a classification model is obtained based on the training of the time domain feature extraction network. Because the TCN network layer is added for extracting the time characteristics, the defect that the traditional one-dimensional convolutional neural network is insensitive to the time characteristics is overcome, the classification precision of the classification model obtained by training is greatly improved, and the detection rate and the accuracy of the focus are improved. Meanwhile, compared with a cyclic neural network, the time domain feature extraction network has smaller computation amount, so that the training speed is greatly improved on the premise that the recognition precision is not lower than that of the cyclic neural network structure. The detection speed of the focus is improved.
Drawings
FIG. 1 is a flowchart of an embodiment of a method for classifying an electrocardiographic signal based on a time domain convolution network according to the present invention;
fig. 2 is a network structure diagram of an embodiment of a time domain feature extraction network provided by the present invention;
FIG. 3 is a schematic structural diagram of an embodiment of a time-domain convolutional network provided in the present invention;
fig. 4 is a schematic structural diagram of an embodiment of a hole convolution part in the time domain convolution network provided by the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. 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
As shown in fig. 1, embodiment 1 of the present invention provides an electrocardiographic signal classification method based on a time domain convolution network, including the following steps:
s1, collecting electrocardiosignals, classifying the electrocardiosignals, and marking category labels to obtain a sample data set;
s2, adding a time domain convolution network structure in the one-dimensional convolution neural network to obtain a time domain feature extraction network based on the time domain convolution network;
s3, training the time domain feature extraction network by adopting the sample data set to obtain a classification model;
and S4, classifying the electrocardiosignals to be classified according to the classification model.
According to the embodiment of the invention, firstly, electrocardiosignals are collected and classified to obtain a sample data set for training a classification model. And then inserting a time domain convolution network structure (TCN network layer) on the basis of the traditional one-dimensional convolution neural network to obtain a time domain feature extraction network. And finally, training the time domain feature extraction network by using the sample data set to obtain a classification model capable of automatically classifying the electrocardiosignals.
The embodiment is improved on the basis of the traditional one-dimensional convolutional neural network structure, combines the time domain characteristics of the ECG signal, adds the TCN network layer to classify the ECG signal, and exerts the advantage that the TCN network layer can effectively extract the time domain signal characteristics; meanwhile, the traditional recurrent neural network for processing the time domain signal has the defect of large calculation amount, and the speed of extracting the time characteristics of the time domain signal by the TCN network layer is higher than that of the recurrent neural network, so that the defect of large calculation amount can be obviously improved by using the TCN network layer, and the training speed and the detection speed are improved.
Specifically, the original electrocardiographic signals collected in this embodiment are 20000 electrocardiographic signals with 8 leads; each data is the same length, 10 seconds, and has a data length of 5000. The collected electrocardiosignals contain 34 categories, and each datum at least contains one category and possibly has multiple categories.
Preferably, a time domain convolution network structure is added to the one-dimensional convolution neural network to obtain a time domain feature extraction network based on the time domain convolution network, and the method specifically comprises the following steps:
and adding the time domain convolution network structure between the convolution layer and the full connection layer of the one-dimensional convolution neural network to obtain a time domain feature extraction network based on the time domain convolution network.
After the convolution layer of the one-dimensional convolution neural network, a TCN network layer is added, so that time characteristics are extracted after convolution characteristics of the electrocardiosignals are extracted, training of time domain characteristics is completed, and finally the convolution characteristics and the time domain characteristics are integrated through a full connection layer to obtain a classification result.
Preferably, the one-dimensional convolutional neural network is any one of VGG-16, inclusion Net, Res Net and Dense Ne.
The one-dimensional convolutional neural network adopted in the embodiment is VGG-16, except for VGG-16, other network structures with a feature extraction function, such as Incepration Net, Res Net, Dense Net and the like, are used as main network frames, and a TCN network layer is added to achieve a corresponding effect.
Specifically, fig. 2 shows a specific structure of the VGG-16 network after being inserted into the TCN network layer; the original electrocardiosignal data is 8-lead data with the sampling rate of 500Hz and the time of 10 seconds and the length of 5000, namely a 5000 x 8 matrix, and 1250 x 8 signals obtained by 4 times of average value down-sampling are taken as input, namely the down-sampling factor is 4. Performing 2-time convolution with convolution kernel of 3 on the down-sampled electrocardiosignal in a same fill mode, keeping the data length before and after convolution the same, performing 1-time convolution with convolution kernel of 24 and step length of 2 as down-sampling operation, and reducing overfitting through a dropout layer with loss rate of 0.2. The operations of convolution 3 times and dropout 1 time are repeated 4 times, the number of the extracted features is doubled, namely the number of the features extracted for the first time is 16, and the steps are 32, 64, 128 and 256.
The output at this time is a 36 × 256 matrix, which is input to the TCN network layer, and the TCN network layer is input to obtain a 36 × 34 matrix, and overfitt is reduced by one dropout layer with a loss rate of 0.2. And then, carrying out full connection Dense with the number of the neurons being 34, and calculating probability and loss through a sigmoid loss function to obtain a result Output.
Specifically, the TCN network layer structure is shown in fig. 3, and the TCN network layer includes a hierarchical scaled cause Conv of the cavity convolution portion, a weight normalization norm, a ReLu function layer, a dropout layer, and a residual connection, which are used for further extracting the time characteristic. The cavity convolution part in the TCN has an exponentially-increased cavity convolution structure of 2, specifically, as shown in fig. 4, from bottom to top, the cavities k of each layer are equal to 1, 2, 4, …, and 32, padding indicates the number of element elements filled in each layer, and hidden indicates a hidden element; thus each output element (y) at the last layer0To y35) Are almost all elements (x) in the input0To x35) And thus abstract features in time series can be extracted. And because only convolution operation is involved in the calculation process, the calculation amount is small, and the calculation speed is high.
Preferably, the time domain feature extraction network is trained by using the sample data set to obtain a classification model, which specifically comprises:
dividing the sample data set into a training set and a verification set, and dividing the sample data in the training set into a plurality of batches;
training the time domain feature extraction network by adopting sample data of one batch in the training set to obtain a training model;
and evaluating the training model by adopting a verification set to obtain an evaluation coefficient, judging whether the evaluation coefficient is greater than an evaluation index, if so, outputting the training model as a classification model, and otherwise, turning to the previous step to carry out the training of the next batch.
Specifically, in this embodiment, during training, 20000 original electrocardiographic signals are divided into 16000 training data and 4000 testing data. The training samples participate in forward propagation in the training process, the prediction result of each iteration is fed back in real time, but the training samples do not participate in the correction process of backward propagation on the network weight, so that the training set can show the generalization capability and the accuracy of the model and express whether the training trend of the model is normal or not. And the test sample does not participate in training, and the final performance result of the model is obtained by using the test set after the network training is finished and is used as data for finally evaluating the effect of the training model.
Preferably, the training of the time domain feature extraction network is performed by using sample data of one batch in the training set to obtain a training model, which specifically comprises:
initializing the time domain feature extraction network;
inputting sample data of a batch into the time domain feature extraction network to obtain a prediction result;
calculating an error between the prediction result and the corresponding class label;
correcting the weight of the time domain feature extraction network through the back propagation of the error to complete one iteration;
and judging whether the set termination condition is met, if so, stopping training and outputting a training model, and if not, performing next iteration.
The training process is based on the BP algorithm: firstly, a training set obtains a prediction result through forward propagation of randomly initialized weights; carrying out error calculation on the prediction result and the category label of the corresponding electrocardiosignal; the correction quantity of the weight by the back propagation is calculated through the error, the network weight is corrected, and the weight is corrected layer by layer from the last layer of the network to the first layer; and the corrected weight and the training data are propagated in the forward direction again, and the steps are repeated. When the iteration is performed once, the weight is gradually corrected to make the final prediction result of the network to the training set approximate to the true value. In the training process, the training set is often divided into a plurality of batches, each batch contains several sample data for training, and such batch size is called batch size. Compared with the mode that all data are sent into the network, the calculation amount of each time is greatly reduced in batches, the network training speed can be accelerated, and the convergence speed of the model is increased.
Preferably, the training model is evaluated by using a validation set to obtain an evaluation coefficient, specifically:
respectively inputting the sample data in the verification set into the training model to obtain a test result, comparing the test result with a corresponding class label, and judging whether the test result is correct;
and calculating the F1 value of the training model according to the correctness of the test result:
wherein F1 is F1 value, P is a1/a2, a1 is the number of samples with correct test result and the test result is the type of abnormal electrocardiogram, a2 is the number of samples with correct test result and R is a1/a, and a is the number of samples with type label as the type of abnormal electrocardiogram.
In the present embodiment, the average value of the F1 values of the electrocardiographic signal prediction results of each category is used as the final evaluation index, and the electrocardiographic abnormality categories referred to in P, R refer to all categories, specifically 34 categories in the present embodiment. Are all made of
Preferably, the time domain feature extraction network is trained by using the sample data set to obtain a classification model, which specifically comprises:
dividing the sample data set into a plurality of sub data sets;
training the time domain feature extraction network by respectively adopting each sub data set to obtain a training model;
and combining a plurality of training models to obtain a classification model.
In this embodiment, the sample data is further divided into multiple parts, a training set and a verification set are divided for each part of sample data, and then the results of the trained multiple model predictions are integrated, so that the generalization capability of the model can be effectively and significantly improved, and the accuracy of the final result is improved. Therefore, a ten-fold cross validation method is adopted during training, the training set and the validation set are divided into 10 equal parts respectively, 9 parts of the training set and the remaining 1 part of the validation set are taken as the training set each time, and the validation data of each time are different. Through such data partitioning, 10 models were trained. And after all the models are trained, obtaining ten outputs from each piece of data of the verification set through ten models, and obtaining the final prediction result by averaging the ten outputs.
The deep learning frame used for training is keras at the rear end of Tensorflow, the batch size of data is 16, the iteration number is 5000, the loss function is binary cross entropy loss, and the optimizer is Adam.
In order to prove the effectiveness and superiority of the embodiment of the invention, two groups of comparison groups are set, one group is a VGG-16 model A with the TCN structural layer removed in figure 1; the other group is a model B in fig. 1, in which the TCN structural layer is removed and replaced with a GRU layer having 34 hidden units, and the model structure proposed in the embodiment of the present invention is a model C. All conditions were the same except for the model structure. The comparison results are shown in tables 1 and 2. TABLE 1, F1 values for 10 cross-validations of the three models, and F1 mean comparison Table
TABLE 2 comparison table of 10 times of cross-validation of three models
Table 1 shows the F1 values and their mean values for each of the three models at 10 cross-validations of the test data, and it can be seen that the model structure proposed by the present invention is clearly superior to the control models a and B. Table 2 shows the time consumed by each iteration of the 10 cross-validation training processes for the three models, and it can be seen that the training speed of the model structure proposed by the present invention is greater than that of the control group B. The 10 cross-validations in tables 1 and 2 are denoted by K1, K2, …, K10, respectively, with the Mean being denoted by Mean, plus the suffix "-F1" for the corresponding F1 value, plus the suffix "-time" for the corresponding iteration time.
The invention trains one-dimensional signal data by using a TCN structure, and because the TCN has the cavity convolution layers which are overlapped layer by layer, the sensitivity field of the TCN on the characteristic extraction of a time sequence is greatly increased, and the time sequence characteristics which are difficult to extract by common convolution are fully extracted.
Example 2
The electrocardiosignal classification device based on the time domain convolution network provided by the embodiment of the invention is used for realizing the electrocardiosignal classification method based on the time domain convolution network, so that the electrocardiosignal classification device based on the time domain convolution network has the technical effect, and the electrocardiosignal classification device based on the time domain convolution network also has the technical effect, and is not repeated herein.
Example 3
Embodiment 3 of the present invention provides a computer storage medium having a computer program stored thereon, where the computer program, when executed by a processor, implements the time-domain convolutional-network-based electrocardiographic signal classification method provided in embodiment 1.
The computer storage medium provided by the embodiment of the invention is used for realizing the electrocardiosignal classification method based on the time domain convolution network, so that the electrocardiosignal classification method based on the time domain convolution network has the technical effects, and the computer storage medium also has the technical effects, and is not repeated herein.
The above-described embodiments of the present invention should not be construed as limiting the scope of the present invention. Any other corresponding changes and modifications made according to the technical idea of the present invention should be included in the protection scope of the claims of the present invention.
Claims (9)
1. An electrocardiosignal classification method based on a time domain convolution network is characterized by comprising the following steps:
acquiring electrocardiosignals, classifying the electrocardiosignals, and marking category labels to obtain a sample data set;
adding a time domain convolution network structure in the one-dimensional convolution neural network to obtain a time domain feature extraction network based on the time domain convolution network;
training the time domain feature extraction network by adopting the sample data set to obtain a classification model;
and classifying the electrocardiosignals to be classified according to the classification model.
2. The method for classifying electrocardiosignals based on the time-domain convolution network as claimed in claim 1, wherein a time-domain convolution network structure is added to the one-dimensional convolution neural network to obtain a time-domain feature extraction network based on the time-domain convolution network, and the method specifically comprises the following steps:
and adding the time domain convolution network structure between the convolution layer and the full connection layer of the one-dimensional convolution neural network to obtain a time domain feature extraction network based on the time domain convolution network.
3. The method for classifying electrocardiosignals based on the time-domain convolution network according to claim 1, wherein the one-dimensional convolution neural network is any one of VGG-16, Incepton Net, Res Net and Dense Ne.
4. The method for classifying electrocardiosignals based on a time-domain convolution network according to claim 1, wherein the time-domain feature extraction network is trained by using the sample data set to obtain a classification model, which specifically comprises:
dividing the sample data set into a training set and a verification set, and dividing the sample data in the training set into a plurality of batches;
training the time domain feature extraction network by adopting sample data of one batch in the training set to obtain a training model;
and evaluating the training model by adopting a verification set to obtain an evaluation coefficient, judging whether the evaluation coefficient is greater than an evaluation index, if so, outputting the training model as a classification model, and otherwise, turning to the previous step to carry out the training of the next batch.
5. The method for classifying electrocardiosignals based on the time-domain convolution network according to claim 4, wherein the time-domain feature extraction network is trained by using sample data of one batch in the training set to obtain a training model, which specifically comprises:
initializing the time domain feature extraction network;
inputting sample data of a batch into the time domain feature extraction network to obtain a prediction result;
calculating an error between the prediction result and the corresponding class label;
correcting the weight of the time domain feature extraction network through the back propagation of the error to complete one iteration;
and judging whether the set termination condition is met, if so, stopping training and outputting a training model, and if not, performing next iteration.
6. The method for classifying electrocardiosignals based on the time-domain convolutional network as claimed in claim 4, wherein the training model is evaluated by using a validation set to obtain an evaluation coefficient, specifically:
respectively inputting the sample data in the verification set into the training model to obtain a test result, comparing the test result with a corresponding class label, and judging whether the test result is correct;
and calculating the F1 value of the training model according to the correctness of the test result:
wherein F1 is F1 value, P is a1/a2, a1 is the number of samples with correct test result and the test result is the type of abnormal electrocardiogram, a2 is the number of samples with correct test result and R is a1/a, and a is the number of samples with type label as the type of abnormal electrocardiogram.
7. The method for classifying electrocardiosignals based on a time-domain convolution network according to claim 1, wherein the time-domain feature extraction network is trained by using the sample data set to obtain a classification model, which specifically comprises:
dividing the sample data set into a plurality of sub data sets;
training the time domain feature extraction network by respectively adopting each sub data set to obtain a training model;
and combining a plurality of training models to obtain a classification model.
8. An apparatus for classifying an electrocardiographic signal based on a time-domain convolutional network, comprising a processor and a memory, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements the method for classifying an electrocardiographic signal based on a time-domain convolutional network according to any one of claims 1 to 7.
9. A computer storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the time-domain convolutional network-based classification of cardiac signals according to any of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010607140.0A CN111803059A (en) | 2020-06-30 | 2020-06-30 | Electrocardiosignal classification method and device based on time domain convolution network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010607140.0A CN111803059A (en) | 2020-06-30 | 2020-06-30 | Electrocardiosignal classification method and device based on time domain convolution network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111803059A true CN111803059A (en) | 2020-10-23 |
Family
ID=72855563
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010607140.0A Pending CN111803059A (en) | 2020-06-30 | 2020-06-30 | Electrocardiosignal classification method and device based on time domain convolution network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111803059A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112793797A (en) * | 2021-02-03 | 2021-05-14 | 东航技术应用研发中心有限公司 | Early warning method and system for landing bounce of airplane |
| CN113095241A (en) * | 2021-04-16 | 2021-07-09 | 武汉理工大学 | Target detection method based on CARLA simulator |
| CN113491523A (en) * | 2021-07-30 | 2021-10-12 | 济南汇医融工科技有限公司 | Electrocardiosignal characteristic point detection method and system |
| CN113761805A (en) * | 2021-09-14 | 2021-12-07 | 东华理工大学 | Controllable source electromagnetic data denoising method, system, terminal and readable storage medium based on time domain convolution network |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109805898A (en) * | 2019-03-22 | 2019-05-28 | 中国科学院重庆绿色智能技术研究院 | Critical illness Mortality Prediction method based on attention mechanism timing convolutional network algorithm |
| CN110596506A (en) * | 2019-10-30 | 2019-12-20 | 福州大学 | Converter fault diagnosis method based on time convolution network |
| CN110720894A (en) * | 2019-11-08 | 2020-01-24 | 广州视源电子科技股份有限公司 | Atrial flutter detection method, device, equipment and storage medium |
| US20200074274A1 (en) * | 2018-08-28 | 2020-03-05 | Beijing Jingdong Shangke Information Technology Co., Ltd. | System and method for multi-horizon time series forecasting with dynamic temporal context learning |
| WO2020049571A1 (en) * | 2018-09-06 | 2020-03-12 | Technion Research & Development Foundation Limited | Predicting ventricular fibrillation |
| 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 |
| CN111012336A (en) * | 2019-12-06 | 2020-04-17 | 重庆邮电大学 | Parallel convolutional network motor imagery electroencephalogram classification method based on spatio-temporal feature fusion |
-
2020
- 2020-06-30 CN CN202010607140.0A patent/CN111803059A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200074274A1 (en) * | 2018-08-28 | 2020-03-05 | Beijing Jingdong Shangke Information Technology Co., Ltd. | System and method for multi-horizon time series forecasting with dynamic temporal context learning |
| WO2020049571A1 (en) * | 2018-09-06 | 2020-03-12 | Technion Research & Development Foundation Limited | Predicting ventricular fibrillation |
| CN109805898A (en) * | 2019-03-22 | 2019-05-28 | 中国科学院重庆绿色智能技术研究院 | Critical illness Mortality Prediction method based on attention mechanism timing convolutional network algorithm |
| CN110596506A (en) * | 2019-10-30 | 2019-12-20 | 福州大学 | Converter fault diagnosis method based on time convolution network |
| CN110720894A (en) * | 2019-11-08 | 2020-01-24 | 广州视源电子科技股份有限公司 | Atrial flutter detection method, device, equipment and 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 |
| CN111012336A (en) * | 2019-12-06 | 2020-04-17 | 重庆邮电大学 | Parallel convolutional network motor imagery electroencephalogram classification method based on spatio-temporal feature fusion |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112793797A (en) * | 2021-02-03 | 2021-05-14 | 东航技术应用研发中心有限公司 | Early warning method and system for landing bounce of airplane |
| CN113095241A (en) * | 2021-04-16 | 2021-07-09 | 武汉理工大学 | Target detection method based on CARLA simulator |
| CN113491523A (en) * | 2021-07-30 | 2021-10-12 | 济南汇医融工科技有限公司 | Electrocardiosignal characteristic point detection method and system |
| CN113761805A (en) * | 2021-09-14 | 2021-12-07 | 东华理工大学 | Controllable source electromagnetic data denoising method, system, terminal and readable storage medium based on time domain convolution network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111160139B (en) | Electrocardiosignal processing method and device and terminal equipment | |
| CN109948647B (en) | Electrocardiogram classification method and system based on depth residual error network | |
| CN111803059A (en) | Electrocardiosignal classification method and device based on time domain convolution network | |
| CN109998525B (en) | Arrhythmia automatic classification method based on discriminant deep belief network | |
| CN109620152B (en) | MutifacolLoss-densenert-based electrocardiosignal classification method | |
| US11324455B2 (en) | Artificial intelligence-based interference recognition method for electrocardiogram | |
| Li et al. | SLC-GAN: An automated myocardial infarction detection model based on generative adversarial networks and convolutional neural networks with single-lead electrocardiogram synthesis | |
| EP3692901A1 (en) | Automatic recognition and classification method for electrocardiogram heartbeat based on artificial intelligence | |
| CN111275093B (en) | Heart beat classification method and device for labeling electrocardiosignals with multiple labels | |
| CN111110228B (en) | Electrocardiosignal R wave detection method and device | |
| CN111990989A (en) | Electrocardiosignal identification method based on generation countermeasure and convolution cyclic network | |
| CN107837082A (en) | Electrocardiogram automatic analysis method and device based on artificial intelligence self study | |
| CN111626114B (en) | Electrocardiosignal arrhythmia classification system based on convolutional neural network | |
| CN110522444B (en) | Electrocardiosignal identification and classification method based on Kernel-CNN | |
| CN111345816B (en) | Multi-lead QRS complex detection method, device, equipment and storage medium | |
| CN111000553A (en) | Intelligent classification method for electrocardiogram data based on voting ensemble learning | |
| CN113274031B (en) | Arrhythmia classification method based on depth convolution residual error network | |
| CN113229825A (en) | Deep neural network-based multi-label multi-lead electrocardiogram classification method | |
| CN116092680B (en) | Abdominal aortic aneurysm early prediction method and system based on random forest algorithm | |
| CN112932433A (en) | Arrhythmia automatic classification method of electrocardiogram | |
| Meng et al. | Biosignal Classification Based on Multi-Feature Multi-Dimensional WaveNet-LSTM Models. | |
| CN113593698B (en) | Traditional Chinese medicine syndrome type identification method based on graph attention network | |
| CN112686091B (en) | Two-step arrhythmia classification method based on deep neural network | |
| CN113768514A (en) | Arrhythmia classification method based on convolutional neural network and gated cyclic unit | |
| CN114255865A (en) | Diagnosis and treatment project prediction method based on recurrent neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201023 |
|
| RJ01 | Rejection of invention patent application after publication |