CN108464827A - It is a kind of it is Weakly supervised under electrocardio image-recognizing method - Google Patents
It is a kind of it is Weakly supervised under electrocardio image-recognizing method Download PDFInfo
- Publication number
- CN108464827A CN108464827A CN201810190859.1A CN201810190859A CN108464827A CN 108464827 A CN108464827 A CN 108464827A CN 201810190859 A CN201810190859 A CN 201810190859A CN 108464827 A CN108464827 A CN 108464827A
- Authority
- CN
- China
- Prior art keywords
- heart
- picture
- claps
- electrocardiosignal
- clapped
- 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
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Animal Behavior & Ethology (AREA)
- Veterinary Medicine (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Pathology (AREA)
- Signal Processing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Cardiology (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses it is a kind of it is Weakly supervised under electrocardio image-recognizing method, belong to field of image recognition.Its feature is to include the following steps：1）Using Denoising Algorithm, electrocardiosignal noise is removed；2）By location algorithm, each heart positioned in electrocardiosignal is clapped, then electrocardiosignal is cut into the single heart and is clapped, and ensures that each heart claps all information for including a heartbeat；3）The one-dimensional heart is clapped into conversion and deliberately claps picture, then heart bat picture is divided into trained picture, verification picture and test pictures three parts；4）Heart bat training picture is input to convolutional neural networks to be trained, the structure heart claps picture recognition model；5）The heart is clapped verification picture to be input in the identification model in step 4, the heart for verifying model claps the adjusting of picture recognition accuracy and each key parameter numerical value.6）Finally, heart bat test pictures are input in step 5 heart after adjustment parameter to clap in picture recognition model, are classified.The classification accurate rate that the present invention claps the heart in picture is high, and model construction only needs a small amount of image data, and important in inhibiting is precisely identified to electrocardiosignal.
Description
Technical field
The invention belongs to field of image recognition, be related to it is a kind of it is Weakly supervised under electrocardio image-recognizing method.
Background technology
Deep learning is a kind of based on the method for carrying out representative learning to data in machine learning.The concept of deep learning is
It was proposed in 2006 by Hinton et al..The concept of deep learning is derived from the research of artificial neural network.It is hidden containing multiple
The multilayer perceptron of layer is exactly a kind of deep learning structure.Deep learning forms more abstract be higher by by combining low-level feature
Feature, to find that the distributed nature of data indicates.
Convolutional neural networks have been developed in recent years efficient identification method.In the 1960s, Hubel and Wiese
Find that its unique network structure can be effective when being used for the neuron of local sensitivity and set direction in studying cat cortex
Ground reduces the complexity of Feedback Neural Network, then proposes convolutional neural networks.Convolutional neural networks include two layers, and one is
Feature extraction layer, secondly being characterized mapping layer.The feature extraction that input picture is realized by first layer, makes structure by the second layer
It is non-linear.
Electrocardiosignal is concentrated expression of the cardiac electrical activity in body surface, and arrhythmia cordis be it is a kind of extremely common but very
Important electrocardio-activity abnormality, therefore automatic classification is carried out to electrocardiosignal in heart disease diagnosis and is of great significance.
But since the same type of heart bat of different people has various forms, the heart bat of different types morphologically each other may also be similar,
Lead to the feature extraction difficulty and the classification not high problem of accuracy that the heart is clapped.
With the development of deep learning, a large amount of scientific research personnel are by one-dimensional convolutional neural networks（1-D CNN）It is applied to
Automatic electrocardiosignal is classified in this field, obtains good effect, but there are still the not high problems of accuracy.
With regard to the above problem, due to advantage of the convolutional neural networks in picture recognition, and retrieves pertinent literature and find do not have
The bat of the one-dimensional heart is converted to the heart and claps picture by researcher, therefore we use Alexnet convolution god using the method for deep learning
Through network, the time amplitude information of signal is clapped by obtaining the one-dimensional heart, is converted into the heart and is clapped picture, is re-fed into convolutional Neural net
In network, the beat classification of pinpoint accuracy can be obtained as a result, identifying heart disease type important in inhibiting to auxiliary doctor.
Invention content
The present invention propose it is a kind of it is Weakly supervised under electrocardio image-recognizing method, this method be intended to use a small amount of electrocardio picture
Signal constructs convolutional neural networks model, realizes the electrocardiosignal classification of pinpoint accuracy.
The technical solution adopted in the present invention is as follows：
It is a kind of it is Weakly supervised under electrocardio image-recognizing method, be as follows：
Step 1 uses Denoising Algorithm, removal electrocardiosignal noise；
It is clapped containing multiple hearts in step 2, electrocardiosignal, by location algorithm, each heart positioned in electrocardiosignal is clapped, then by the heart
Electric signal cuts into the single heart and claps, and ensures that each heart claps all information for including a heartbeat；
The one-dimensional heart is clapped to convert and deliberately claps picture by step 3, then heart bat picture is divided into trained picture, verification picture and test chart
Piece three parts；
The heart is clapped training picture and is input to convolutional neural networks and is trained by step 4, and the structure heart claps picture recognition model；
Heart bat verification picture is input in the identification model in step 4 by step 5, and the heart for verifying model claps picture recognition
The adjusting of accuracy and each key parameter numerical value；
Heart bat test pictures are input in step 5 in the bat picture recognition model of the heart after adjustment parameter by step 6, are classified；
It is according to claim 1 it is a kind of it is Weakly supervised under electrocardio image-recognizing method, it is characterised in that：It is used in step 1
Denoising Algorithm removes electrocardiosignal noise.Electrocardiosignal includes mainly Hz noise, myoelectricity noise and baseline drift noise, is adopted
Electrocardiosignal is subjected to frequency decomposition with wavelet algorithm, the electrocardiosignal in each frequency range is obtained, band segment is removed, then make
Electrocardiosignal is reconstructed with remaining frequency range, the effect of removal Hz noise, myoelectricity noise and baseline drift noise is had reached, to obtain
Obtain the electrocardiosignal after denoising.
It is according to claim 1 it is a kind of it is Weakly supervised under electrocardio image-recognizing method, it is characterised in that：In step 2
It is clapped containing multiple hearts in electrocardiosignal, each heart need to be positioned using location algorithm and claps position.Each heart bat generally comprise P, Q,
R, S, T wave, this method use R wave location algorithms, if being done if intercepting forward to do and intercept backward by R waves position, obtain
The heart is clapped, and each heart is made to clap all information for including a heartbeat.
It is according to claim 1 it is a kind of it is Weakly supervised under electrocardio image-recognizing method, it is characterised in that：In step 3
The one-dimensional heart is clapped into conversion and deliberately claps picture, the x-axis that the heart claps picture is built by the time shaft that the one-dimensional heart is clapped, then by time shaft
Numerical information structure the heart clap picture y-axis, obtain the heart clap picture, then by the heart bat picture be divided into trained picture, verification picture and
Test pictures three parts.
It is according to claim 1 it is a kind of it is Weakly supervised under electrocardio image-recognizing method, it is characterised in that：Step 4 will
Heart bat training picture is input to convolutional neural networks and is trained, and the structure heart claps picture recognition model, and network structure uses
AlexNet convolutional neural networks, key parameter are as follows：Learning rate value is 0.001, and weight pad value is 0.005, is used
" step " learning strategy.
It is according to claim 1 it is a kind of it is Weakly supervised under electrocardio image-recognizing method, it is characterised in that：Step 5 will
The heart is clapped verification picture and is input in the heart bat picture recognition model generated in step 4, crosses and obtains the knowledge that every heart claps verification picture
Not as a result, being used for the recognition accuracy of computation model.By calculated recognition accuracy come the structure situation of reaction model, then
The numerical value for adjusting each key parameter obtains the best heart and claps picture recognition model.
It is according to claim 1 it is a kind of it is Weakly supervised under electrocardio image-recognizing method, it is characterised in that：Step 6 will
The heart is clapped test pictures and is input in step 5 in the bat picture recognition model of the heart after adjustment parameter, and the output heart claps the identification knot of picture
Fruit realizes that the heart claps the classification of picture.
Advantageous effect：
（1）The present invention carries out feature extraction and classification using convolutional neural networks to transformed electrocardio picture, realizes high precision
The image classification of degree；
（2）The present invention has structure heart bat identification model data volume few, and the feature that Model Identification accuracy is high, is suitble to apply to
Accurately electrocardiosignal identifies, to intelligent medical important in inhibiting.
Description of the drawings
Fig. 1 is present system schematic diagram
Fig. 2 is that the heart of the present invention claps picture
Fig. 3 is convolutional neural networks schematic diagram of the present invention
Fig. 4 is electrocardio picture classification results figure of the present invention.
Specific implementation mode
The present invention is described in further detail With reference to embodiment, it should be understood that described below
Preferred embodiment is only for the purpose of illustrating and explaining the present invention and is not intended to limit the present invention：
1. using Denoising Algorithm, electrocardiosignal noise is removed
Using the decomposition and reconstruction of wavelet algorithm, mother wavelet function ' bior2.6 ' is selected, electrocardiosignal is decomposed into 8 layers, by the
One layer and the 8th layer coefficients zero setting remove highest frequency and low-limit frequency, reach removal Hz noise, myoelectricity noise and baseline drift
The effect of shifting, the reconstructed residual number of plies obtain the electrocardiosignal after denoising.
2. electrocardiosignal cuts into the single heart and claps
Each heart bat generally comprises P, Q, R, S, T wave, and this method uses R wave location algorithms, 100 are intercepted forward by R waves position
A sampled point and backward 200 sampled points of interception obtain the single heart and clap, ensure that each heart beat of data includes the institute of a heartbeat
There is information.
3. the one-dimensional heart claps conversion and deliberately claps picture, then heart bat picture is divided into trained picture, verification picture and test pictures
Three parts
The one-dimensional heart is clapped into conversion and deliberately claps picture, building the heart by the time shaft that the one-dimensional heart is clapped claps the x-axis of picture, then passes through the time
The numerical information structure heart on axis claps the y-axis of picture, obtains the heart and claps picture.Then, heart bat picture is divided into trained picture, verification
Picture and test pictures three parts.
It is trained 4. the heart is clapped training picture and is input to convolutional neural networks, the structure heart claps picture recognition model
The heart is clapped into training picture（Per class tens to hundreds of）It is input to convolutional neural networks and carries out model construction, as shown in figure 3,
The figure is convolutional neural networks schematic diagram, which is that heart bat picture passes through convolution after the heart claps picture input network
Layer carries out picture feature extraction, reduces feature quantity using pond layer, constantly repeats convolution pond and simplify feature, finally pass through
All features are input in specific classification device by full articulamentum classifies.This method network structure uses AlexNet convolution god
Through network, key parameter is as follows：Base_lr values are 0.001, and weight pad value is that 0.005, gamma values are 0.1,
Stepsize values are 128, and maximum iteration value is 400, uses " step " learning strategy." step " strategy formula such as following formula
（1）：
Wherein, lr_policy is learning rate, and the initial learning rates of base_lr, iter is current iteration number（Value is according to maximum
Iterations are from 1 to 400）.Floor function performances are downward rounding.
It is clapped in identification model 5. the heart is clapped verification picture and is input to the heart, carries out parameter regulation
The heart is clapped into verification picture（Per class tens to hundreds of）The heart that input generates is clapped in picture recognition model, by obtaining every
The heart claps the recognition result of verification picture, is used for the recognition accuracy of computation model.It is reacted by calculated recognition accuracy
The structure situation of model, then in regulating step 4 key parameter numerical value, obtain more preferably the heart clap picture recognition model.
6. heart heart bat test pictures being input to after adjustment parameter is clapped in picture recognition model, classify
Heart bat test pictures are input to the heart after adjustment parameter to clap in picture recognition model, the output heart claps the identification knot of picture
Fruit realizes that the heart claps the classification of picture
After above-mentioned six step, the heart can be obtained and clap picture recognition accuracy.The method of the present invention discloses number by MIT-BIH-AR
It is verified according to library, clapping picture to the 4 class hearts is identified, and the heart is clapped training picture and opened per class 100, and the heart claps verification picture per class 50
, the heart claps test pictures number 2000, identification accuracy 99.3%.Wherein, the heart claps picture categories and the heart is clapped picture number and can be added
Add.Experimental result is shown in Table 1, wherein " N ", " L ", " A ", " W " are four kinds of heart disease classifications, sees Fig. 2.
Claims (7)
1. it is a kind of it is Weakly supervised under electrocardio image-recognizing method, characterized in that including following steps：
Step 1 uses Wavelet Algorithm, removal electrocardiosignal noise；
It is clapped containing multiple hearts in step 2, electrocardiosignal, by location algorithm, each heart positioned in electrocardiosignal is clapped, then by the heart
Electric signal cuts into the single heart and claps, and ensures that each heart claps all information for including a heartbeat；
The one-dimensional heart is clapped to convert and deliberately claps picture by step 3, then heart bat picture is divided into trained picture, verification picture and test chart
Piece three parts；
The heart is clapped training picture and is input to convolutional neural networks and is trained by step 4, and the structure heart claps picture recognition model；
Heart bat verification picture is input in the identification model in step 4 by step 5, and the heart for verifying model claps picture recognition
The adjusting of accuracy and each key parameter numerical value；
Heart bat test pictures are input in step 5 in the bat picture recognition model of the heart after adjustment parameter by step 6, are classified.
2. it is according to claim 1 it is a kind of it is Weakly supervised under electrocardio image-recognizing method, it is characterised in that：Make in step 1
With Denoising Algorithm, electrocardiosignal noise is removed, electrocardiosignal includes mainly Hz noise, myoelectricity noise and baseline drift noise,
Electrocardiosignal is carried out by frequency decomposition using wavelet algorithm, the electrocardiosignal in each frequency range is obtained, band segment is removed, then
Electrocardiosignal is reconstructed using remaining frequency range, has reached the effect of removal Hz noise, myoelectricity noise and baseline drift noise, to
Obtain the electrocardiosignal after denoising.
3. it is according to claim 1 it is a kind of it is Weakly supervised under electrocardio image-recognizing method, it is characterised in that：Step 2 center
Clapped containing multiple hearts in electric signal, each heart need to be positioned using location algorithm and claps position, each heart bat generally comprise P, Q, R,
S, T waves, this method use R wave location algorithms, if being done if intercepting forward to do and intercept backward by R waves position, obtain the heart
It claps, each heart is made to clap all information for including a heartbeat.
4. it is according to claim 1 it is a kind of it is Weakly supervised under electrocardio image-recognizing method, it is characterised in that：It will in step 3
The one-dimensional heart claps conversion and deliberately claps picture, builds the x-axis that the heart claps picture by the time shaft that the one-dimensional heart is clapped, then by time shaft
Numerical information builds the y-axis that the heart claps picture, obtains the heart and claps picture, then heart bat picture is divided into trained picture, verification picture and survey
Attempt piece three parts.
5. it is according to claim 1 it is a kind of it is Weakly supervised under electrocardio image-recognizing method, it is characterised in that：Step 4 is by the heart
Bat training picture is input to convolutional neural networks and is trained, and the structure heart claps picture recognition model, and network structure uses AlexNet
Convolutional neural networks, key parameter are as follows：Base_lr values are 0.001, and weight pad value is that 0.005, gamma values are 0.1,
Stepsize values are 128, and maximum iteration value is 400, uses " step " learning strategy.
6. it is according to claim 1 it is a kind of it is Weakly supervised under electrocardio image-recognizing method, it is characterised in that：Step 5 is by the heart
It claps verification picture to be input in the heart bat picture recognition model generated in step 4, crosses and obtain the identification that every heart claps verification picture
As a result, the recognition accuracy for computation model, by calculated recognition accuracy come the structure situation of reaction model, then is adjusted
The numerical value for saving each key parameter obtains the best heart and claps picture recognition model.
7. it is according to claim 1 it is a kind of it is Weakly supervised under electrocardio image-recognizing method, it is characterised in that：Step 6 is by the heart
Test pictures to be clapped to be input in step 5 in the bat picture recognition model of the heart after adjustment parameter, the output heart claps the recognition result of picture,
Realize that the heart claps the classification of picture.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810190859.1A CN108464827A (en) | 2018-03-08 | 2018-03-08 | It is a kind of it is Weakly supervised under electrocardio image-recognizing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810190859.1A CN108464827A (en) | 2018-03-08 | 2018-03-08 | It is a kind of it is Weakly supervised under electrocardio image-recognizing method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108464827A true CN108464827A (en) | 2018-08-31 |
Family
ID=63264254
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810190859.1A Pending CN108464827A (en) | 2018-03-08 | 2018-03-08 | It is a kind of it is Weakly supervised under electrocardio image-recognizing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108464827A (en) |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109480825A (en) * | 2018-12-13 | 2019-03-19 | 武汉中旗生物医疗电子有限公司 | The processing method and processing device of electrocardiogram (ECG) data |
| CN109497986A (en) * | 2018-11-22 | 2019-03-22 | 杭州脉流科技有限公司 | Electrocardiogram intelligent analysis method, device, computer equipment and system based on deep neural network |
| CN109620211A (en) * | 2018-11-01 | 2019-04-16 | 吉林大学珠海学院 | A kind of intelligent abnormal electrocardiogram aided diagnosis method based on deep learning |
| CN109636786A (en) * | 2018-12-11 | 2019-04-16 | 杭州嘉楠耘智信息科技有限公司 | Verification method and device of image recognition module |
| CN110141214A (en) * | 2019-04-23 | 2019-08-20 | 首都师范大学 | A kind of mask method of electrocardiogram identification and its application |
| CN110432892A (en) * | 2019-08-05 | 2019-11-12 | 苏州米特希赛尔人工智能有限公司 | Machine learning ECG Automatic Diagnosis System |
| CN110811608A (en) * | 2019-11-19 | 2020-02-21 | 中电健康云科技有限公司 | Atrial fibrillation monitoring method based on ECG (ECG) signals |
| CN110897626A (en) * | 2018-09-14 | 2020-03-24 | 杭州脉流科技有限公司 | Deep neural network-based electrocardiogram analysis method, device, computer equipment and system |
| CN110897628A (en) * | 2018-09-14 | 2020-03-24 | 杭州脉流科技有限公司 | Deep neural network-based electrocardiogram signal feature extraction method, device, system, equipment and classification method |
| CN110897629A (en) * | 2018-09-14 | 2020-03-24 | 杭州脉流科技有限公司 | Deep learning algorithm-based electrocardiogram feature extraction method, device, system, equipment and classification method |
| CN111476282A (en) * | 2020-03-27 | 2020-07-31 | 东软集团股份有限公司 | Data classification method and device, storage medium and electronic equipment |
| CN111723622A (en) * | 2019-03-22 | 2020-09-29 | 安徽华米信息科技有限公司 | Heart beat classification method and device, wearable device and storage medium |
| CN111803061A (en) * | 2020-06-28 | 2020-10-23 | 武汉中旗生物医疗电子有限公司 | R wave identification method and device based on target detection |
| CN114648049A (en) * | 2022-05-20 | 2022-06-21 | 合肥心之声健康科技有限公司 | Method, device and system for constructing and classifying electrocardio image classification model |
| TWI810901B (en) * | 2022-04-20 | 2023-08-01 | 義大醫療財團法人 | Remote electrocardiogram monitoring device and electrocardiogram monitoring method |
| CN117745808A (en) * | 2024-02-19 | 2024-03-22 | 南通市计量检定测试所 | Electrocardiogram image positioning comparison method based on photogrammetry |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105125206A (en) * | 2015-09-15 | 2015-12-09 | 中山大学 | Intelligent electrocardio monitoring method and device |
| CN105769177A (en) * | 2016-04-01 | 2016-07-20 | 佛山市质量计量监督检测中心 | Electrocardiograph signal examination record analyzing device |
| CN106108889A (en) * | 2016-07-20 | 2016-11-16 | 杨平 | Electrocardiogram classification method based on degree of depth learning algorithm |
| CN106778685A (en) * | 2017-01-12 | 2017-05-31 | 司马大大（北京）智能系统有限公司 | Electrocardiogram image-recognizing method, device and service terminal |
-
2018
- 2018-03-08 CN CN201810190859.1A patent/CN108464827A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105125206A (en) * | 2015-09-15 | 2015-12-09 | 中山大学 | Intelligent electrocardio monitoring method and device |
| CN105769177A (en) * | 2016-04-01 | 2016-07-20 | 佛山市质量计量监督检测中心 | Electrocardiograph signal examination record analyzing device |
| CN106108889A (en) * | 2016-07-20 | 2016-11-16 | 杨平 | Electrocardiogram classification method based on degree of depth learning algorithm |
| CN106778685A (en) * | 2017-01-12 | 2017-05-31 | 司马大大（北京）智能系统有限公司 | Electrocardiogram image-recognizing method, device and service terminal |
Non-Patent Citations (2)
| Title |
|---|
| Y. YAN, X. QIN, Y. WU, N. ZHANG, J. FAN AND L. WANG: "A restricted Boltzmann machine based two-lead electrocardiography classification", 《2015 IEEE 12TH INTERNATIONAL CONFERENCE ON WEARABLE AND IMPLANTABLE BODY SENSOR NETWORKS (BSN)》 * |
| 贺佳，陆健: "《医学统计学中的SAS统计分析》", 31 December 2002 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110897626A (en) * | 2018-09-14 | 2020-03-24 | 杭州脉流科技有限公司 | Deep neural network-based electrocardiogram analysis method, device, computer equipment and system |
| CN110897628A (en) * | 2018-09-14 | 2020-03-24 | 杭州脉流科技有限公司 | Deep neural network-based electrocardiogram signal feature extraction method, device, system, equipment and classification method |
| CN110897629A (en) * | 2018-09-14 | 2020-03-24 | 杭州脉流科技有限公司 | Deep learning algorithm-based electrocardiogram feature extraction method, device, system, equipment and classification method |
| CN109620211A (en) * | 2018-11-01 | 2019-04-16 | 吉林大学珠海学院 | A kind of intelligent abnormal electrocardiogram aided diagnosis method based on deep learning |
| CN109497986A (en) * | 2018-11-22 | 2019-03-22 | 杭州脉流科技有限公司 | Electrocardiogram intelligent analysis method, device, computer equipment and system based on deep neural network |
| CN109636786A (en) * | 2018-12-11 | 2019-04-16 | 杭州嘉楠耘智信息科技有限公司 | Verification method and device of image recognition module |
| CN109636786B (en) * | 2018-12-11 | 2022-11-22 | 嘉楠明芯(北京)科技有限公司 | Verification method and device of image recognition module |
| CN109480825B (en) * | 2018-12-13 | 2021-08-06 | 武汉中旗生物医疗电子有限公司 | Electrocardio data processing method and device |
| CN109480825A (en) * | 2018-12-13 | 2019-03-19 | 武汉中旗生物医疗电子有限公司 | The processing method and processing device of electrocardiogram (ECG) data |
| CN111723622B (en) * | 2019-03-22 | 2024-04-26 | 安徽华米信息科技有限公司 | Heart beat classification method, heart beat classification device, wearable equipment and storage medium |
| CN111723622A (en) * | 2019-03-22 | 2020-09-29 | 安徽华米信息科技有限公司 | Heart beat classification method and device, wearable device and storage medium |
| CN110141214A (en) * | 2019-04-23 | 2019-08-20 | 首都师范大学 | A kind of mask method of electrocardiogram identification and its application |
| CN110432892A (en) * | 2019-08-05 | 2019-11-12 | 苏州米特希赛尔人工智能有限公司 | Machine learning ECG Automatic Diagnosis System |
| CN110811608A (en) * | 2019-11-19 | 2020-02-21 | 中电健康云科技有限公司 | Atrial fibrillation monitoring method based on ECG (ECG) signals |
| CN111476282A (en) * | 2020-03-27 | 2020-07-31 | 东软集团股份有限公司 | Data classification method and device, storage medium and electronic equipment |
| CN111803061A (en) * | 2020-06-28 | 2020-10-23 | 武汉中旗生物医疗电子有限公司 | R wave identification method and device based on target detection |
| CN111803061B (en) * | 2020-06-28 | 2022-12-30 | 武汉中旗生物医疗电子有限公司 | R wave identification method and device based on target detection |
| TWI810901B (en) * | 2022-04-20 | 2023-08-01 | 義大醫療財團法人 | Remote electrocardiogram monitoring device and electrocardiogram monitoring method |
| CN114648049A (en) * | 2022-05-20 | 2022-06-21 | 合肥心之声健康科技有限公司 | Method, device and system for constructing and classifying electrocardio image classification model |
| CN117745808A (en) * | 2024-02-19 | 2024-03-22 | 南通市计量检定测试所 | Electrocardiogram image positioning comparison method based on photogrammetry |
| CN117745808B (en) * | 2024-02-19 | 2024-05-03 | 南通市计量检定测试所 | Electrocardiogram image positioning comparison method based on photogrammetry |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108464827A (en) | It is a kind of it is Weakly supervised under electrocardio image-recognizing method | |
| Singh et al. | A new ECG denoising framework using generative adversarial network | |
| Li et al. | Classification of ECG signals based on 1D convolution neural network | |
| Das et al. | ECG beats classification using mixture of features | |
| CN113627518A (en) | Method for realizing multichannel convolution-recurrent neural network electroencephalogram emotion recognition model by utilizing transfer learning | |
| US6546378B1 (en) | Signal interpretation engine | |
| Sarkaleh et al. | Classification of ECG arrhythmias using discrete wavelet transform and neural networks | |
| Javadi et al. | Improving ECG classification accuracy using an ensemble of neural network modules | |
| Alwasiti et al. | Motor imagery classification for brain computer interface using deep metric learning | |
| CN110555468A (en) | Electroencephalogram signal identification method and system combining recursion graph and CNN | |
| Wang et al. | An approach of one-vs-rest filter bank common spatial pattern and spiking neural networks for multiple motor imagery decoding | |
| Zhang et al. | Channel selection in motor imaginary-based brain-computer interfaces: a particle swarm optimization algorithm | |
| Asghar et al. | Semi-skipping layered gated unit and efficient network: hybrid deep feature selection method for edge computing in EEG-based emotion classification | |
| Miah et al. | CluSem: Accurate clustering-based ensemble method to predict motor imagery tasks from multi-channel EEG data | |
| Kumar M et al. | Classification of ECG signal using FFT based improved Alexnet classifier | |
| CN110321856B (en) | Time-frequency multi-scale divergence CSP brain-computer interface method and device | |
| Jain et al. | Sleep stages classification using wavelettransform & neural network | |
| Sinha et al. | HSCAD: Heart sound classification for accurate diagnosis using machine learning and MATLAB | |
| Ahmad et al. | Features optimization for ECG signals classification | |
| Allam et al. | Patient-specific ECG beat classification using EMD and deep learning-based technique | |
| Tahir et al. | Classification of EEG Signal by Training Neural Network with Swarm Optimization for Identification of Epilepsy | |
| Alabandi | Combining Deep Learning with Traditional Machine Learning to Improve Classification Accuracy on Small Datasets | |
| Naghsh-Nilchi et al. | Cardiac arrhythmias classification method based on MUSIC, morphological descriptors, and neural network | |
| Sikder et al. | Heterogeneous hand guise classification based on surface electromyographic signals using multichannel convolutional neural network | |
| Khorasani et al. | Agricultural Harvester Sound Classification using Convolutional Neural Networks and Spectrograms |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180831 |
|
| WD01 | Invention patent application deemed withdrawn after publication |