CN111956214A - QRS wave automatic detection method based on U-net end-to-end neural network model - Google Patents
QRS wave automatic detection method based on U-net end-to-end neural network model Download PDFInfo
- Publication number
- CN111956214A CN111956214A CN202010744022.4A CN202010744022A CN111956214A CN 111956214 A CN111956214 A CN 111956214A CN 202010744022 A CN202010744022 A CN 202010744022A CN 111956214 A CN111956214 A CN 111956214A
- Authority
- CN
- China
- Prior art keywords
- convolution
- unit
- neural network
- pooling
- layer
- 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
- 238000003062 neural network model Methods 0.000 title claims abstract description 20
- 238000001514 detection method Methods 0.000 title claims abstract description 10
- 238000012549 training Methods 0.000 claims abstract description 23
- 238000012360 testing method Methods 0.000 claims abstract description 18
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 11
- 230000011218 segmentation Effects 0.000 claims abstract description 6
- 238000007781 pre-processing Methods 0.000 claims abstract description 5
- 238000011176 pooling Methods 0.000 claims description 44
- 238000013528 artificial neural network Methods 0.000 claims description 18
- 230000005284 excitation Effects 0.000 claims description 15
- 238000005070 sampling Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 7
- 230000003321 amplification Effects 0.000 claims description 6
- 238000010586 diagram Methods 0.000 claims description 6
- 238000003199 nucleic acid amplification method Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 3
- 238000000034 method Methods 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 2
- 230000006870 function Effects 0.000 description 21
- 238000000718 qrs complex Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 3
- 238000004195 computer-aided diagnosis Methods 0.000 description 3
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 2
- 230000000747 cardiac effect Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000000877 morphologic effect Effects 0.000 description 2
- 208000024172 Cardiovascular disease Diseases 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000003759 clinical diagnosis Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000002045 lasting effect Effects 0.000 description 1
- 230000004660 morphological change Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000011179 visual inspection Methods 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—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
Abstract
The invention discloses a QRS wave automatic detection method based on a U-net end-to-end neural network model, which comprises the following steps: 1) generating a single-lead electrocardiosignal sample by preprocessing an original signal and combining a tag set; 2) building a semantic segmentation neural network model of the single-lead electrocardiosignal; 3) training parameters of a convolutional neural network; 4) automatically identifying the test set sample; the recognition rate of the training samples exceeds 60%.
Description
Technical Field
The invention relates to the technical field of medical signal processing, in particular to a QRS wave automatic detection method based on a U-net end-to-end neural network model.
Background
With the development of digital technology, computer-aided diagnosis systems have become the most promising clinical diagnosis solution due to their fast and reliable analysis means. Today, by advanced hardware facilities, the electrocardiosignals of a patient, namely the electrocardiogram, can be easily obtained. Physicians can judge the state of a patient by observing the information contained in the electrocardiogram, however the process of manual or visual inspection and inferring these subtle morphological changes in the long continuous electrocardiographic beat is time consuming and prone to error due to fatigue. Therefore, a real-time computer-aided diagnosis system is essential to help doctors monitor the condition of patients in real time, overcoming these limitations of evaluation of electrocardiogram signals.
The computer aided diagnosis system can analyze the information in the electrocardiogram in real time to obtain the effective information in the electrocardiogram. An important problem in the processing of cardiac electrical signals is the detection of the waveform of the ECG signal. The QRS wave is the most prominent wave form in the ECG signal, and contains much useful information inside, so that only after the QRS complex is first determined (including the position and time width of the QRS complex wave), the subsequent heart rate and other information are calculated organically. However, the R wave is known to be the most significant wave form in the re-QRS complex. Therefore, the analysis of the ECG signal is based on the analysis of other waveforms in the ECG signal after the position determination of the R wave is performed first.
The R wave automatic identification system working on calculator hardware is the core of the equipment, and the technical approach is to extract a characteristic vector capable of representing effective information of an electrocardiogram, input the characteristic vector into a semantic segmentation algorithm to obtain the outline of a segmented QRS complex, and further judge the approximate position of an R point. The technical difficulty in the step of extracting the feature vector is the extraction of morphological features, and reasonable feature extraction can directly influence the accuracy and reliability of the result. The morphological characteristics are supplemented with other characteristics on the electrocardiogram to form a characteristic vector, the characteristic vector is input into a semantic segmentation algorithm, and the profile of the QRS complex, namely the position of the QRS complex, is output after processing.
Disclosure of Invention
The invention provides a QRS wave automatic detection method based on a U-net end-to-end neural network model to solve the problems.
A QRS wave automatic detection method based on a U-net end-to-end neural network model comprises the following steps:
1) generating single-lead electrocardiosignal sample by preprocessing original signal and combining label set
Reading in single-lead electrocardiosignal data, removing a point with a particularly large wave crest from each signal, taking 50 points from the front and back of each lead electrocardiosignal according to the position of the peak of the R wave at the same moment, marking the points as 1, and marking the rest positions as 0 to form a label set with the same length as the electrocardiosignals. Splicing the electrocardiosignal and the tag set according to a first dimension, and amplifying the original 1 x w-dimensional data set into a 1 x 2 w-dimensional signal + tag form;
processing all single-lead electrocardiosignals in the above mode and splicing the processed electrocardiosignals according to a second dimension to form a data set U, wherein each sample in the U is the electrocardiosignal data of the 1 x 2w dimension;
2) semantic segmentation neural network model for building single-lead electrocardiosignal
The whole network consists of 10 convolutional layers, 2 pooling layers, 2 upsampling layers and 1 LSTM layer and is connected with an input X;
3) training parameters of convolutional neural networks
Initializing parameters of the convolutional neural network, randomly extracting 80% of samples from a sampled data set U as a training set, and taking other unselected samples as a test set; inputting the electrocardiosignal samples in the training set into the initialized neural network, performing iteration by taking a minimized cost function as a target, generating and storing parameters of the U-net neural network;
4) automatic identification of test set samples
Inputting the divided test set samples into a convolutional neural network and operating to obtain a 2-dimensional prediction graph corresponding to the test set samples, performing matrix processing on the prediction graph to obtain final prediction output, comparing the output prediction value with the label of the test set samples to check whether the prediction is correct, and judging the performance of the model through a prediction result y _ pred;
the parameters of the U-net neural network are as follows: inputting X into electrocardiosignal samples, wherein each electrocardiosignal sample is in dimension of 1X W, 1 is the number of leads, and W is the number of points contained in one signal; inputting the electrocardiosignal samples into a U-net neural network in batches, wherein the U-net neural network comprises two convolution pooling units, an LSTM unit and two pooling up-sampling units which are connected in series. The first convolution pooling unit comprises two convolution layers, the number of convolution kernels is 8, the sizes of the convolution kernels are 3, an excitation unit behind the convolution layer unit is a relu function, the size of a pooling kernel of the pooling layer unit is 2, and the pooling step length is 1; the dimension of the characteristic diagram after passing through the first convolution pooling unit is (W/2) × 8; the second convolution pooling unit comprises two convolution layers, the number of convolution kernels is 16, the number of convolution kernels is 3, an excitation unit behind the convolution layer unit is a relu function, a dropout layer is arranged between the convolution layers and the pooling layer, the parameter is 0.5, the pooling kernel size of the pooling layer unit is 2, and the pooling step length is 1; the dimension of the characteristic diagram after passing through the first convolution pooling unit is (W/4) × 16; the second convolution pooling unit is connected with an LSTM unit in series, the stm layer hidden layer is provided with 8 neurons, and the excitation unit is a relu function; then a dropout layer is formed, and the parameter is 0.5; the dimension of the characteristic diagram after passing through the LSTM unit is (W/4) × 8; the first deconvolution unit comprises an upper sampling layer and two convolution layers, wherein the amplification factor of the upper sampling layer is 2, the number of convolution kernels of the two convolution layers is 16, the size of the convolution kernels is 3, and an excitation unit behind the convolution layer unit is a relu function; the dimension of the feature map after the first deconvolution unit is (W/2) × 16; the second deconvolution unit comprises an upper sampling layer and three convolutional layers, the amplification coefficient of the upper sampling layer is 2, the number of convolution kernels of the first two convolutional layers is 8, the size of the convolution kernels is 3, the number of convolution kernels of the third convolutional layer is 16, the size of the convolution kernels is 3, and an excitation unit behind the convolutional layers is a relu function; the dimension of the feature graph after passing through the first deconvolution unit is W x 2; finally, after passing through a convolution unit, the number of convolution kernels is 1, the size of the convolution kernels is 1, and the activation function is a sigmoid function; the final output heatmap dimension is W × 1.
The iteration is as follows: and (4) iteratively updating the training parameters once until the loss value and the accuracy of the convolutional neural network are stabilized near a certain value, stopping training and storing the training parameters and the model structure information of the current network.
The invention provides a QRS wave automatic detection method based on a U-net end-to-end neural network model, which comprises the following steps: 1) generating a single-lead electrocardiosignal sample by preprocessing an original signal and combining a tag set; 2) building a semantic segmentation neural network model of the single-lead electrocardiosignal; 3) training parameters of a convolutional neural network; 4) automatically identifying the test set sample; the result shows that the recognition rate of the training sample exceeds 60 percent.
Detailed Description
Embodiment 1 QRS wave automatic detection method based on U-net end-to-end neural network model
A specific example source is icbeb2019, chinese physiological signal challenge race, which database contains 2,000 single lead electrocardiogram records collected from patients with cardiovascular disease (CVD), each record lasting 10 s. Downloading data on an icbeb.2019.org website, and making labels according to the marked positions of R points during downloading, only classifying points belonging to QRS waves and points not belonging to QRS waves, wherein the correspondence between the labels of the two categories and the categories in a data set is shown in table 1. In this example, this is done by a software system (windows linux) operating on a computer and the Matlab and python software environments known in the industry.
The detailed steps of this example are as follows:
generating single-lead electrocardiosignal samples
Reading single lead electrocardiosignal data of a data set downloaded from an icbeb.2019.org website by python, firstly removing ultrahigh peak signal interference from an original signal, then respectively marking 50 points in front of and behind the same time as the R wave peak position as 1, marking the rest points as 0, so that the marked 0 and 1 matrixes are labels of the signal, carrying out the same operation on the used signal, then splicing the processed signals according to a second dimension, and finally obtaining a data set containing 2000 x 5000 x 1 dimensional data, wherein 2000 refers to the number of signal strips, namely the number of samples, and 5000 refers to the number of points contained in each sample. Each sample is 5000X 1 of 1 single lead cardiac signal data X as input to the u-net neural network.
Two, building single-lead u-net neural network model
The input of the neural network model is an electrocardiosignal sample X, wherein X is a (5000X 1) -dimensional sample of the electrocardiosignal output by the preprocessing part, 1 is the number of the leads of the used electrocardiosignals, namely the number of input channels, and 5000 is the number of points of each signal. Each 5000 x 1 dimensional data is transmitted into a u-net neural network. The u-net network comprises 2 convolution pooling units, 1 LSTM unit and 2 deconvolution units, wherein each convolution pooling unit consists of two convolution layers and one pooling layer; the number of the convolution kernels of the two convolution layers of the first convolution pooling unit is 8, the sizes of the convolution kernels are 3, the excitation unit behind the convolution layer unit is a relu function, the size of the pooling kernel of the pooling layer unit is 2, and the pooling step length is 1; the dimension of the feature map after passing through the first convolution pooling unit is (5000/2) × 8; the number of convolution kernels of two convolution layers of the second convolution pooling unit is 16, the number of the convolution kernels is 3, the excitation unit behind the convolution layer unit is a relu function, a dropout layer is arranged between the convolution layer and the pooling layer, the parameter is 0.5, the pooling kernel size of the pooling layer unit is 2, and the pooling step length is 1; the dimension of the feature map after the first convolution pooling unit is (5000/4) × 16; the hidden layer of the LSTM layer is provided with 8 neurons, and the excitation unit is a relu function; then a dropout layer is formed, and the parameter is 0.5; the feature map dimension after passing through the LSTM unit is (5000/4) × 8; the amplification factor of a sampling layer on the first deconvolution unit is 2, the number of convolution kernels of two layers of convolution layers is 16, the size of the convolution kernels is 3, and an excitation unit behind the convolution layer unit is a relu function; the dimension of the feature map after the first deconvolution unit is (5000/2) × 16; the amplification factor of the sampling layer on the second deconvolution unit is 2, the number of convolution kernels of the first two layers of convolution layers is 8, the size of the convolution kernels is 3, the number of convolution kernels of the third layer of convolution layers is 16, the size of the convolution kernels is 3, and the excitation unit behind the convolution layers is a relu function; the dimension of the feature map after passing through the first deconvolution unit is 5000 x 2; finally, after passing through a convolution unit, the number of convolution kernels is 1, the size of the convolution kernels is 1, and the activation function is a sigmoid function; the final output heatmap is a one-dimensional vector with dimensions 5000 x 1, where the value of each point represents the predicted label for each point of the original signal. The model was built using a keras open source framework and python language. The parameters of the u-net network can be seen in table 2.
Thirdly, training parameters of the convolutional neural network model
Firstly, initializing training parameters of the neural network model, dividing the sampled signals into training set samples and testing set samples, and enabling a divided data set U to be shown in a table 3. And inputting the electrocardiosignals of the single leads in the training set into the initialized u-net neural network model, wherein a cross entropy function is used as a cost function in the u-net neural network. Since the partition is nearly binary, Keras uses a binary _ crosssentropy function, in the neural network an object Model is instantiated by a constructed functional Model, and in the Model function a parameter loss is set to 'binary _ crosssentropy'. Performing iteration by using an Adam optimizer and taking a minimized cost function as a target, and optimizing by setting a parameter optimizer in a model. complex function as 'Adam', wherein the learning rate is automatic, so as to generate the deep neural network and store a file my _ model. hd5 which is a suffix of hd 5; wherein the training parameters are updated once per iteration. And stopping training and storing the training parameters and model structure information of the current network until the loss value and the accuracy of the deep neural network are stabilized near a certain value. The neural network trains 10 rounds in total.
Fourthly, automatically identifying the sample of the test set
And inputting all the divided test set samples into the stored u-net neural network model.hd5, operating the u-net neural network to obtain 5000 x 1 dimensional predicted value vector output y _ pred corresponding to the test set samples, comparing the output predicted value with the label of the test set samples to check whether the classification is correct, counting the number n of samples with the same corresponding position values of the y _ pred and the y _ label, and dividing the n by the total number of the test set samples to obtain the final accuracy.
Fifthly, outputting the classification result
The results are shown in table 3.
Claims (3)
1. A QRS wave automatic detection method based on a U-net end-to-end neural network model comprises the following steps:
1) generating single-lead electrocardiosignal sample by preprocessing original signal and combining label set
Reading in single-lead electrocardiosignal data, removing a point with a particularly large wave crest from each signal, taking 50 points from the front to the back of each lead electrocardiosignal according to the position of the peak of the R wave at the same moment, marking the points as 1, and marking the rest positions as 0 to form a label set with the same length as the electrocardiosignals; splicing the electrocardiosignal and the tag set according to a first dimension, and amplifying the original 1 x w-dimensional data set into a 1 x 2 w-dimensional signal + tag form;
processing all single-lead electrocardiosignals in the above mode and splicing the processed electrocardiosignals according to a second dimension to form a data set U, wherein each sample in the U is the electrocardiosignal data of the 1 x 2w dimension;
2) semantic segmentation neural network model for building single-lead electrocardiosignal
The whole network consists of 10 convolutional layers, 2 pooling layers, 2 upsampling layers and 1 LSTM layer and is connected with an input X;
3) training parameters of convolutional neural networks
Initializing parameters of the convolutional neural network, randomly extracting 80% of samples from a sampled data set U as a training set, and taking other unselected samples as a test set; inputting the electrocardiosignal samples in the training set into the initialized neural network, performing iteration by taking a minimized cost function as a target, generating and storing parameters of the U-net neural network;
4) automatic identification of test set samples
Inputting the divided test set samples into a convolutional neural network and operating to obtain a 2-dimensional prediction graph corresponding to the test set samples, performing matrix processing on the prediction graph to obtain final prediction output, comparing the output prediction value with the label of the test set samples to check whether the prediction is correct, and judging the performance of the model according to the prediction result y _ pred.
2. The method for automatically detecting QRS waves based on the U-net end-to-end neural network model as claimed in claim 1, wherein: the parameters of the U-net neural network are as follows: inputting X into electrocardiosignal samples, wherein each electrocardiosignal sample is in dimension of 1X W, 1 is the number of leads, and W is the number of points contained in one signal; inputting the electrocardiosignal samples into a U-net neural network in batches, wherein the U-net neural network comprises two convolution pooling units, an LSTM unit and two pooling up-sampling units which are connected in series; the first convolution pooling unit comprises two convolution layers, the number of convolution kernels is 8, the sizes of the convolution kernels are 3, an excitation unit behind the convolution layer unit is a relu function, the size of a pooling kernel of the pooling layer unit is 2, and the pooling step length is 1; the dimension of the characteristic diagram after passing through the first convolution pooling unit is (W/2) × 8; the second convolution pooling unit comprises two convolution layers, the number of convolution kernels is 16, the number of convolution kernels is 3, an excitation unit behind the convolution layer unit is a relu function, a dropout layer is arranged between the convolution layers and the pooling layer, the parameter is 0.5, the pooling kernel size of the pooling layer unit is 2, and the pooling step length is 1; the dimension of the characteristic diagram after passing through the first convolution pooling unit is (W/4) × 16; the second convolution pooling unit is connected with an LSTM unit in series, the stm layer hidden layer is provided with 8 neurons, and the excitation unit is a relu function; then a dropout layer is formed, and the parameter is 0.5; the dimension of the characteristic diagram after passing through the LSTM unit is (W/4) × 8; the first deconvolution unit comprises an upper sampling layer and two convolution layers, wherein the amplification factor of the upper sampling layer is 2, the number of convolution kernels of the two convolution layers is 16, the size of the convolution kernels is 3, and an excitation unit behind the convolution layer unit is a relu function; the dimension of the feature map after the first deconvolution unit is (W/2) × 16; the second deconvolution unit comprises an upper sampling layer and three convolutional layers, the amplification coefficient of the upper sampling layer is 2, the number of convolution kernels of the first two convolutional layers is 8, the size of the convolution kernels is 3, the number of convolution kernels of the third convolutional layer is 16, the size of the convolution kernels is 3, and an excitation unit behind the convolutional layers is a relu function; the dimension of the feature graph after passing through the first deconvolution unit is W x 2; finally, after passing through a convolution unit, the number of convolution kernels is 1, the size of the convolution kernels is 1, and the activation function is a sigmoid function; the final output heatmap dimension is W × 1.
3. The method for automatically detecting QRS waves based on the U-net end-to-end neural network model as claimed in claim 2, wherein: the iteration is as follows: and (4) iteratively updating the training parameters once until the loss value and the accuracy of the convolutional neural network are stabilized near a certain value, stopping training and storing the training parameters and the model structure information of the current network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010744022.4A CN111956214A (en) | 2020-07-29 | 2020-07-29 | QRS wave automatic detection method based on U-net end-to-end neural network model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010744022.4A CN111956214A (en) | 2020-07-29 | 2020-07-29 | QRS wave automatic detection method based on U-net end-to-end neural network model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111956214A true CN111956214A (en) | 2020-11-20 |
Family
ID=73363415
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010744022.4A Pending CN111956214A (en) | 2020-07-29 | 2020-07-29 | QRS wave automatic detection method based on U-net end-to-end neural network model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111956214A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113197583A (en) * | 2021-05-11 | 2021-08-03 | 广元市中心医院 | Electrocardiogram waveform segmentation method based on time-frequency analysis and recurrent neural network |
| CN113349791A (en) * | 2021-05-31 | 2021-09-07 | 平安科技（深圳）有限公司 | Abnormal electrocardiosignal detection method, device, equipment and medium |
| CN113762483A (en) * | 2021-09-16 | 2021-12-07 | 华中科技大学 | 1D U-net neural network processor for electrocardiosignal segmentation |
| CN113951893A (en) * | 2021-12-02 | 2022-01-21 | 清华大学 | Multi-lead electrocardiosignal characteristic point extraction method combining deep learning and electrophysiological knowledge |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170112401A1 (en) * | 2015-10-27 | 2017-04-27 | CardioLogs Technologies | Automatic method to delineate or categorize an electrocardiogram |
| CN109864714A (en) * | 2019-04-04 | 2019-06-11 | 北京邮电大学 | A kind of ECG Signal Analysis method based on deep learning |
| CN110141245A (en) * | 2019-06-20 | 2019-08-20 | 鲁东大学 | A kind of Characteristics of electrocardiogram vector extracting method |
| CN110353665A (en) * | 2019-08-14 | 2019-10-22 | 苏州大学 | ECG signal signature waveform detection method is singly led based on improved one-dimensional U-net |
| CN111110228A (en) * | 2020-01-17 | 2020-05-08 | 武汉中旗生物医疗电子有限公司 | Electrocardiosignal R wave detection method and device |
-
2020
- 2020-07-29 CN CN202010744022.4A patent/CN111956214A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170112401A1 (en) * | 2015-10-27 | 2017-04-27 | CardioLogs Technologies | Automatic method to delineate or categorize an electrocardiogram |
| CN109864714A (en) * | 2019-04-04 | 2019-06-11 | 北京邮电大学 | A kind of ECG Signal Analysis method based on deep learning |
| CN110141245A (en) * | 2019-06-20 | 2019-08-20 | 鲁东大学 | A kind of Characteristics of electrocardiogram vector extracting method |
| CN110353665A (en) * | 2019-08-14 | 2019-10-22 | 苏州大学 | ECG signal signature waveform detection method is singly led based on improved one-dimensional U-net |
| CN111110228A (en) * | 2020-01-17 | 2020-05-08 | 武汉中旗生物医疗电子有限公司 | Electrocardiosignal R wave detection method and device |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113197583A (en) * | 2021-05-11 | 2021-08-03 | 广元市中心医院 | Electrocardiogram waveform segmentation method based on time-frequency analysis and recurrent neural network |
| CN113349791A (en) * | 2021-05-31 | 2021-09-07 | 平安科技（深圳）有限公司 | Abnormal electrocardiosignal detection method, device, equipment and medium |
| CN113762483A (en) * | 2021-09-16 | 2021-12-07 | 华中科技大学 | 1D U-net neural network processor for electrocardiosignal segmentation |
| CN113951893A (en) * | 2021-12-02 | 2022-01-21 | 清华大学 | Multi-lead electrocardiosignal characteristic point extraction method combining deep learning and electrophysiological knowledge |
| CN113951893B (en) * | 2021-12-02 | 2023-02-03 | 清华大学 | Multi-lead electrocardiosignal characteristic point extraction method combining deep learning and electrophysiological knowledge |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111160139B (en) | Electrocardiosignal processing method and device and terminal equipment | |
| CN110141219B (en) | Lead fusion deep neural network-based myocardial infarction automatic detection method | |
| CN110141220B (en) | Myocardial infarction automatic detection system based on multi-mode fusion neural network | |
| CN108875575B (en) | Automatic arrhythmia analysis method based on multi-channel signal fusion deep neural network | |
| CN111956214A (en) | QRS wave automatic detection method based on U-net end-to-end neural network model | |
| EP3692901A1 (en) | Automatic recognition and classification method for electrocardiogram heartbeat based on artificial intelligence | |
| CN107832737B (en) | Artificial intelligence-based electrocardiogram interference identification method | |
| CN111956212B (en) | Inter-group atrial fibrillation recognition method based on frequency domain filtering-multi-mode deep neural network | |
| CN109544518B (en) | Method and system applied to bone maturity assessment | |
| US7174205B2 (en) | Cardiac diagnostic system and method | |
| CN108720831B (en) | Automatic arrhythmia analysis method based on lead deep neural network | |
| EP2224346B1 (en) | Ecg signal analysis tool | |
| CN108937912B (en) | Automatic arrhythmia analysis method based on deep neural network | |
| CN112294341B (en) | Sleep electroencephalogram spindle wave identification method and system based on light convolutional neural network | |
| CN108090509B (en) | Data length self-adaptive electrocardiogram classification method | |
| Ahmed et al. | An investigative study on motifs extracted features on real time big-data signals | |
| CN108766557B (en) | Automatic arrhythmia analysis method based on channel signal fusion neural network | |
| CN110638430B (en) | Method for building cascade neural network ECG signal arrhythmia classification model | |
| CN111626114A (en) | Electrocardiosignal arrhythmia classification system based on convolutional neural network | |
| CN112426160A (en) | Electrocardiosignal type identification method and device | |
| Zhang et al. | An ECG heartbeat classification method based on deep convolutional neural network | |
| CN111759304B (en) | Electrocardiogram abnormity identification method and device, computer equipment and storage medium | |
| CN110897634A (en) | Electrocardiosignal generation method based on generation countermeasure network | |
| CN115221926A (en) | Heart beat signal classification method based on CNN-GRU network model | |
| CN115530788A (en) | Arrhythmia classification method based on self-attention mechanism |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201120 |