CN114886436B - Premature beat recognition method based on improved convolutional neural network - Google Patents
Premature beat recognition method based on improved convolutional neural network Download PDFInfo
- Publication number
- CN114886436B CN114886436B CN202210507194.9A CN202210507194A CN114886436B CN 114886436 B CN114886436 B CN 114886436B CN 202210507194 A CN202210507194 A CN 202210507194A CN 114886436 B CN114886436 B CN 114886436B
- Authority
- CN
- China
- Prior art keywords
- output
- input
- residual
- jump
- convolutional
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
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]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
- A61B5/364—Detecting abnormal ECG interval, e.g. extrasystoles, ectopic heartbeats
-
- 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
-
- 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
- G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Cardiology (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Pathology (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Signal Processing (AREA)
- Computational Linguistics (AREA)
- Veterinary Medicine (AREA)
- Public Health (AREA)
- Computing Systems (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Psychiatry (AREA)
- Physiology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
本发明公开一种基于改进卷积神经网络的早搏识别方法，提出基于改进卷积神经网络的早搏识别模型。先利用训练样本对到构建的室性早搏识别模型进行训练，再将预测样本送入到训练好的室性早搏识别模型中，实现待分类的心电图的分类。本发明从模型的深度和宽度两个方面对原始ResNet34模型进行改进，再将SimAM注意力机制融入到改进了的ResNet模型中，以此模型来提升早搏识别的准确率。
The present invention discloses a premature beat recognition method based on an improved convolutional neural network, and proposes a premature beat recognition model based on an improved convolutional neural network. First, a constructed ventricular premature beat recognition model is trained using training samples, and then the predicted samples are sent to the trained ventricular premature beat recognition model to achieve the classification of the electrocardiogram to be classified. The present invention improves the original ResNet34 model from the depth and width of the model, and then integrates the SimAM attention mechanism into the improved ResNet model, so as to improve the accuracy of premature beat recognition by using this model.
Description
Technical Field
The invention relates to the technical field of heart rhythm premature beat recognition, in particular to a premature beat recognition method based on an improved convolutional neural network.
Background
Arrhythmia is a clinically common cardiovascular disease, with premature beat being the most common arrhythmia. By introducing a computer-aided premature beat analysis technology, doctors can perform effective medical intervention as early as possible, plays a very important role in preventing and treating diseases, and can solve the problem that medical resources in remote mountain areas are scarce and cannot enjoy basic medical services to a certain extent.
Traditional premature beat recognition methods, such as traditional neural networks, support vector machines, rule reasoning and the like, all require that the Electrocardiogram (ECG) features be manually extracted in advance and then classified. The accuracy of feature extraction is a key factor affecting the performance of traditional premature beat classification methods. However, features of small amplitude in an electrocardiogram, such as P-wave, T-wave, QRS-wave boundary points, etc., are susceptible to high frequency noise. The extraction accuracy of the features is low, so that the manually extracted features cannot accurately reflect the pathological information of the ECG (electrocardiogram) so as to influence the subsequent classification performance.
Disclosure of Invention
The invention aims to solve the problem that the existing premature beat recognition method cannot accurately reflect pathological information of ECG (electrocardiogram) by adopting a manual feature extraction mode so as to influence subsequent classification performance, and provides the premature beat recognition method based on an improved convolutional neural network.
In order to solve the problems, the invention is realized by the following technical scheme:
The premature beat identification method based on the improved convolutional neural network comprises the following steps:
Step 1, constructing a premature beat recognition model;
The premature beat recognition model consists of 1 convolution layer, 1 maximum pooling layer, 4 residual unit blocks and 1 global average pooling layer; the input of the convolution layer is used as the input of the premature beat identification model, the output of the convolution layer is connected with the input of the maximum pooling layer, the output of the maximum pooling layer is connected with the input of the first residual error unit block, the output of the first residual error unit block is connected with the input of the second residual error unit block, the output of the second residual error unit block is connected with the input of the third residual error unit block, the output of the third residual error unit block is connected with the input of the fourth residual error unit block, the output of the fourth residual error unit block is connected with the input of the global average pooling layer, and the output of the global average pooling layer is used as the output of the premature beat identification model;
step 2, preprocessing the electrocardiogram marked with the classification as a training sample, and training the premature beat recognition model constructed in the step1 by using the training sample to obtain a trained premature beat recognition model;
And step 3, preprocessing the electrocardiogram to be classified to obtain a prediction sample, and sending the prediction sample into the premature beat recognition model trained in the step 2 to realize classification of the electrocardiogram to be classified.
In the above scheme, the first residual error unit block is respectively composed of 2 direct jump residual error modules; the input of the first direct jump residual error module is used as the input of the first residual error unit block, the output of the first direct jump residual error module is connected with the input of the second direct jump residual error module, and the output of the second direct jump residual error module is used as the output of the first residual error unit block; the second residual unit block consists of 1 convolution jump residual module and 1 direct jump residual module respectively; the input of the first convolution jump residual error module is used as the input of the second residual error unit block, the output of the first convolution jump residual error module is connected with the input of the third direct jump residual error module, and the output of the third direct jump residual error module is used as the output of the second residual error unit block; the third residual error unit block consists of 1 convolution jump residual error module and 2 direct jump residual error modules respectively; the input of the second convolution jump residual error module is used as the input of the third residual error unit block, the output of the second convolution jump residual error module is connected with the input of the fourth direct jump residual error module, the output of the fourth direct jump residual error module is connected with the input of the fifth direct jump residual error module, and the output of the fifth direct jump residual error module is used as the output of the third residual error unit block; the fourth residual error unit block is respectively composed of 1 convolution jump residual error module and 1 direct jump residual error module; the input of the third convolution jump residual error module is used as the input of the fourth residual error unit block, the output of the third convolution jump residual error module is connected with the input of the sixth direct jump residual error module, and the output of the sixth direct jump residual error module is used as the output of the fourth residual error unit block.
In the scheme, the direct jump residual error module consists of 2 convolution layers, 1 SimAM attention mechanism layers and 1 direct jump branch; the input of the first convolution layer and the direct jump branch are used as the input of the direct jump residual error module, the output of the first convolution layer is connected with the input of the second convolution layer, the output of the second convolution layer is connected with the input of the first SimAM attention mechanism layer, and the output of the first SimAM attention mechanism layer and the output of the direct jump branch are added to be used as the output of the direct jump residual error module.
In the scheme, the convolution jump residual error module consists of 2 convolution layers, 1 SimAM attention mechanism layers and 1 convolution jump branch; the input of the third convolution layer and the convolution jump branch are used as the input of the convolution jump residual error module, the output of the third convolution layer is connected with the input of the fourth convolution layer, the output of the fourth convolution layer is connected with the input of the second SimAM attention mechanism layer, and the output of the second SimAM attention mechanism layer and the output of the convolution jump branch are added to be used as the output of the convolution jump residual error module.
Compared with the prior art, the invention improves the original ResNet model from two aspects of depth and width of the model, and then blends the SimAM attention mechanism into the improved ResNet model, so that the accuracy of premature beat identification is improved by the model.
Drawings
Fig. 1 is a schematic structural diagram of the premature beat recognition model NEWRESNET.
Fig. 2 is a schematic diagram of the structure of the SimAM attentive mechanism layer.
Detailed Description
The present invention will be further described in detail with reference to specific examples in order to make the objects, technical solutions and advantages of the present invention more apparent.
A premature beat identification method based on an improved convolutional neural network, comprising the steps of:
And 1, constructing a premature beat recognition model.
The deep neural network is an end-to-end learning model which can automatically learn signal features from original input data, and avoids the manual feature extraction of the traditional method. Considering that the deep learning network layer number is too deep, the network model can generate serious overfitting, so the invention reduces the layer number of ResNet and the number of characteristic diagrams of each convolution module. In addition, considering SimAM (Parameter-Free Attention Module) is a simple and very effective attention mechanism without additional parameters, which can infer the weight value of 3D from the current neuron, the problems that the existing attention mechanism can only refine the characteristics in one dimension of a channel or a space and the space which changes simultaneously in the channel and the space lacks flexibility are solved, and the like, and therefore, the invention integrates the SimAM attention mechanism into an improved ResNet model.
Based on the above analysis, the premature beat recognition model NEWRESNET constructed by the present invention improves on the original ResNet model from the depth and width of the model, and then fuses the SimAM attention mechanism into the improved ResNet model. As shown in fig. 1, the premature beat recognition model NEWRESNET consists of 1 convolutional layer (conv), 1 max pooling layer (max pool), 4 residual unit blocks, and 1 global average pooling layer (avg pool). The input of the convolution layer is used as the input of the premature beat identification model, the output of the convolution layer is connected with the input of the maximum pooling layer, the output of the maximum pooling layer is connected with the input of the first residual error unit block, the output of the first residual error unit block is connected with the input of the second residual error unit block, the output of the second residual error unit block is connected with the input of the third residual error unit block, the output of the third residual error unit block is connected with the input of the fourth residual error unit block, the output of the fourth residual error unit block is connected with the input of the global average pooling layer, and the output of the global average pooling layer is used as the output of the premature beat identification model.
The 4 residual unit blocks are formed by respectively connecting 2,3 and 2 residual modules in series. The first residual error unit block consists of 2 direct jump residual error modules respectively; wherein the input of the first direct jump residual module is used as the input of the first residual unit block, the output of the first direct jump residual module is connected with the input of the second direct jump residual module, and the output of the second direct jump residual module is used as the output of the first residual unit block. The second residual unit block consists of 1 convolution jump residual module and 1 direct jump residual module respectively; the input of the first convolution jump residual error module is used as the input of the second residual error unit block, the output of the first convolution jump residual error module is connected with the input of the third direct jump residual error module, and the output of the third direct jump residual error module is used as the output of the second residual error unit block. The third residual error unit block consists of 1 convolution jump residual error module and 2 direct jump residual error modules respectively; the input of the second convolution jump residual error module is used as the input of the third residual error unit block, the output of the second convolution jump residual error module is connected with the input of the fourth direct jump residual error module, the output of the fourth direct jump residual error module is connected with the input of the fifth direct jump residual error module, and the output of the fifth direct jump residual error module is used as the output of the third residual error unit block. The fourth residual error unit block is respectively composed of 1 convolution jump residual error module and 1 direct jump residual error module; the input of the third convolution jump residual error module is used as the input of the fourth residual error unit block, the output of the third convolution jump residual error module is connected with the input of the sixth direct jump residual error module, and the output of the sixth direct jump residual error module is used as the output of the fourth residual error unit block.
Each residual module consists of 2 convolutional layers (conv), 1 SimAM attention mechanism layers, and 1 skip branch. SimAM the attention mechanism layer is shown in figure 2. The input vector x of SimAM attentiveness mechanism layer is subjected to SimAM attentiveness mechanism to obtain a weight value e t of each neuron in the output input vector x, and then the weight value e t is multiplied by the input vector x to obtain the output y of SimAM attentiveness mechanism layer. The jump branch comprises 2 types: the jump branch is represented by a solid line, is a direct jump branch, and the input and the output of the direct jump branch are identical mappings, and the jump branch is formed by a direct jump residual error module; a jump branch is indicated by a dashed line, and is a convolution jump branch, the input of the convolution jump branch obtains the output of the convolution jump branch through a 1*1 convolution layer, and the convolution jump residual error module is formed.
The direct jump residual error module consists of 2 convolution layers, 1 SimAM attention mechanism layers and 1 direct jump branch; the input of the first convolution layer and the direct jump branch are used as the input of the direct jump residual error module, the output of the first convolution layer is connected with the input of the second convolution layer, the output of the second convolution layer is connected with the input of the first SimAM attention mechanism layer, and the output of the first SimAM attention mechanism layer and the output of the direct jump branch are added to be used as the output of the direct jump residual error module.
The convolution jump residual error module consists of 2 convolution layers, 1 SimAM attention mechanism layers and 1 convolution jump branch; the input of the third convolution layer and the convolution jump branch are used as the input of the convolution jump residual error module, the output of the third convolution layer is connected with the input of the fourth convolution layer, the output of the fourth convolution layer is connected with the input of the second SimAM attention mechanism layer, and the output of the second SimAM attention mechanism layer and the output of the convolution jump branch are added to be used as the output of the convolution jump residual error module.
In all the convolution layers and the maximum pooling layer, "1*7" or "1*3" represents the size of the convolution kernel of the convolution layer, "36", "60", "80" or "128" represents the number of feature maps of the convolution layer, and "/1" or "/2" represents the movement step size of the convolution layer.
And 2, preprocessing the classified electrocardiograms to be used as training samples, and training the premature beat recognition model constructed in the step1 by using the training samples to obtain a trained premature beat recognition model.
The preprocessing process of the ECG mainly uses a filter to remove noise such as baseline drift, power frequency interference and the like in the ECG. Classification as labeled in the electrocardiogram is either premature or non-premature.
And step 3, preprocessing the electrocardiogram to be classified to obtain a prediction sample, and sending the prediction sample into the premature beat recognition model trained in the step 2 to realize classification of the electrocardiogram to be classified.
The way of preprocessing the electrocardiogram to be classified is the same as the way of preprocessing the electrocardiogram marked with classification. The output of the predicted sample obtained by the model is a record of premature beat or non-premature beat.
The performance of the invention is illustrated by the following example:
based on Chinese cardiovascular disease database (CCDD database, http://58.210.56.164/ccdd /), the ECG record in the database is subjected to band-pass filtering of 0.5-40 Hz for denoising.
Since the premature beat of the present invention is classified into a classification problem (premature beat or non-premature beat), sensitivity (Se), specificity (Sp), accuracy (Acc) can be used to measure the quality of the classification effect. The two-class confusion matrix is shown in table 1 below.
TABLE 1 confusion matrix
The definition of each index is as follows:
Sensitivity (Se):
Se＝TP/(TP+FN)
Specificity (Sp):
Sp＝TN/(TN+FP)
accuracy (Acc):
Acc＝(TP+TN)/(TP+TN+FP+FN)
1) Identification of ventricular premature beat (PVC)
35840 Pre-processed ECG recordings were used as training samples, containing 3112 PVC recordings. 141046 pre-processed ECG recordings were used as test samples, which contained 2148 PVC recordings. After training the NEWRESNET model with the training sample, the results obtained by the trained model on the test sample are shown in table 2 below, where NPVC represents non-ventricular premature beat record, PVC is ventricular premature beat record, se is sensitivity, sp is specificity, acc is accuracy.
Table 2 PVC identification results
2) Identification of atrial premature beat (PAB)
Taking 44800 preprocessed ECG records as training samples, wherein the training samples comprise more than 4 thousand PAB records; 132087 pre-processed ECG recordings were used as test samples, containing 4172 PAB recordings. After the improved training of NEWRESNET model using training sample pair NEWRESNET, the results obtained by the trained model on the test sample are shown in table 2 below, where NAPB represents non-atrial premature beat record, APB represents atrial premature beat record, se is sensitivity, sp is specificity, and Acc is accuracy.
Table 3 APB identification results
As can be seen from tables 2 and 3, the sensitivity, specificity and accuracy of classifying the electrocardiogram by the premature beat recognition model ResNet provided by the invention all reach more than 90%.
It should be noted that, although the examples described above are illustrative, this is not a limitation of the present invention, and thus the present invention is not limited to the above-described specific embodiments. Other embodiments, which are apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein, are considered to be within the scope of the invention as claimed.
Claims (1)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210507194.9A CN114886436B (en) | 2022-05-10 | 2022-05-10 | Premature beat recognition method based on improved convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210507194.9A CN114886436B (en) | 2022-05-10 | 2022-05-10 | Premature beat recognition method based on improved convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114886436A CN114886436A (en) | 2022-08-12 |
| CN114886436B true CN114886436B (en) | 2024-11-22 |
Family
ID=82721262
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210507194.9A Active CN114886436B (en) | 2022-05-10 | 2022-05-10 | Premature beat recognition method based on improved convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114886436B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106725420A (en) * | 2015-11-18 | 2017-05-31 | 中国科学院苏州纳米技术与纳米仿生研究所 | VPB recognition methods and VPB identifying system |
| CN109730652A (en) * | 2019-03-08 | 2019-05-10 | 江苏霆升科技有限公司 | Ventricular premature beat positioning identifying method and device |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10706350B1 (en) * | 2017-08-11 | 2020-07-07 | Facebook, Inc. | Video analysis using convolutional networks |
| EP3870039A4 (en) * | 2018-10-26 | 2022-08-10 | Mayo Foundation for Medical Education and Research | NEURAL NETWORKS FOR SCREENING OF ATRIAL FIBERS |
| CN110516670B (en) * | 2019-08-26 | 2022-04-22 | 广西师范大学 | Target detection method based on scene level and area suggestion self-attention module |
| CN110974214A (en) * | 2019-12-20 | 2020-04-10 | 华中科技大学 | A deep learning-based automatic electrocardiogram classification method, system and device |
| CN111354030B (en) * | 2020-02-29 | 2023-08-04 | 同济大学 | Unsupervised Monocular Image Depth Map Generation Method Embedded with SENet Unit |
| US11809519B2 (en) * | 2020-08-19 | 2023-11-07 | Lg Electronics Inc. | Semantic input sampling for explanation (SISE) of convolutional neural networks |
| CN114387539B (en) * | 2021-12-21 | 2025-04-04 | 南京信息工程大学 | A behavior recognition method based on SimAM attention mechanism |
-
2022
- 2022-05-10 CN CN202210507194.9A patent/CN114886436B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106725420A (en) * | 2015-11-18 | 2017-05-31 | 中国科学院苏州纳米技术与纳米仿生研究所 | VPB recognition methods and VPB identifying system |
| CN109730652A (en) * | 2019-03-08 | 2019-05-10 | 江苏霆升科技有限公司 | Ventricular premature beat positioning identifying method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114886436A (en) | 2022-08-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yang et al. | Ensemble of kernel extreme learning machine based random forest classifiers for automatic heartbeat classification | |
| Mitra et al. | A rough-set-based inference engine for ECG classification | |
| CN106805965A (en) | A kind of electrocardiosignal sorting technique and device | |
| CN113229825A (en) | Deep neural network-based multi-label multi-lead electrocardiogram classification method | |
| CN108256453A (en) | A kind of method based on one-dimensional ECG signal extraction two dimension CNN features | |
| CN112842349B (en) | Intelligent analysis method for supraventricular premature beat based on attention transfer mechanism | |
| CN115281688B (en) | A multi-label detection system for cardiac hypertrophy based on multimodal deep learning | |
| US20230225663A1 (en) | Method for predicting multi-type electrocardiogram heart rhythms based on graph convolution | |
| CN116361688A (en) | Multi-mode feature fusion model construction method for automatic classification of electrocardiographic rhythms | |
| Wang et al. | QRS detection in ECG signal based on residual network | |
| CN114091530B (en) | ECG classification method, ECG classification model training method based on domain adaptation | |
| Deevi et al. | HeartNetEC: a deep representation learning approach for ECG beat classification | |
| CN115881308A (en) | Method for constructing heart disease classification model based on spatiotemporal joint detection | |
| CN114668375A (en) | A Cuffless Blood Pressure Prediction System Based on Deep Neural Network Model | |
| Li et al. | Arrhythmia classification using biased dropout and morphology-rhythm feature with incremental broad learning | |
| Peng et al. | A deep learning framework for ECG denoising and classification | |
| CN116226724A (en) | Non-invasive coronary heart disease detection system | |
| CN114176600B (en) | Electrocardiogram ST segment abnormality discrimination system based on causal analysis | |
| CN111557660A (en) | A method for arrhythmia identification under the framework of subpopulation deep learning | |
| CN115718867A (en) | Shappelet-based classification method for converting electrocardiosignals into image structures | |
| CN114886436B (en) | Premature beat recognition method based on improved convolutional neural network | |
| CN114692698A (en) | A one-dimensional ECG data classification method based on residual network | |
| Roland et al. | An automated system for arrhythmia detection using ECG records from MITDB | |
| CN114886435B (en) | Ventricular premature beat recognition method based on improved MobileNetV3 | |
| CN115429284B (en) | Electrocardiogram signal classification method, system, computer device and readable storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |