CN109303560A - A kind of atrial fibrillation recognition methods of electrocardiosignal in short-term based on convolution residual error network and transfer learning - Google Patents
A kind of atrial fibrillation recognition methods of electrocardiosignal in short-term based on convolution residual error network and transfer learning Download PDFInfo
- Publication number
- CN109303560A CN109303560A CN201811293013.7A CN201811293013A CN109303560A CN 109303560 A CN109303560 A CN 109303560A CN 201811293013 A CN201811293013 A CN 201811293013A CN 109303560 A CN109303560 A CN 109303560A
- Authority
- CN
- China
- Prior art keywords
- network
- residual error
- atrial fibrillation
- layer
- short
- 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
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
- A61B5/361—Detecting fibrillation
-
- 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
- A61B5/725—Details of waveform analysis using specific filters therefor, e.g. Kalman or adaptive filters
-
- 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)
- Cardiology (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Pathology (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
A kind of atrial fibrillation recognition methods of electrocardiosignal in short-term based on convolution residual error network and transfer learning, this method can realize the atrial fibrillation arrhythmia cordis identified in electrocardiosignal in short-term, the treating method comprises following steps: (1) obtaining the segment of electrocardiogram (ECG) data in short-term of tape label as training data;(2) convolutional neural networks and residual unit are constructed respectively and are combined into convolution residual error network;(3) it using four disaggregated model of convolution residual error net structure and is trained, extracts the network parameter under optimal training pattern;(4) continue Optimized model using transfer learning, save the network parameter under optimal models;(5) disaggregated model for obtaining electrocardiogram (ECG) data input step (4) to be measured obtains differentiating result.Creativeness of the invention is, convolution residual error network and transfer learning are combined, and it is difficult that the atrial fibrillation caused by solving the problems, such as by signal difference alienation identifies, improves the generalization ability and robustness of model.
Description
Technical field
The present invention relates to atrial fibrillation electrocardiosignals in short-term to know field, in particular to a kind of to be moved based on convolution residual error network and depth
Move the atrial fibrillation recognition methods of electrocardiosignal in short-term of study.
Background technique
Scientific research surface, the whole world have tens of millions of people to die of heart disease every year, and heart disease is in all diseases
Disease incidence highest, the maximum disease of harm.Auricular fibrillation, abbreviation atrial fibrillation are the most common perpetual arrhythmia heart diseases.
According to statistics, the disease incidence of atrial fibrillation is 1%-2%, and the illness rate of atrial fibrillation also gradually increases with advancing age.Heart sheet
Disease of body such as heart failure, valve disease, myocardial infarction etc. is significant related to atrial fibrillation disease.Atrial fibrillation can be such that the stroke of patient sends out
Raw rate increases by 500, and is to lead to death and invalid one of the major reasons.Therefore, the timely diagnosis of the heart diseases such as atrial fibrillation
It is of great significance to improving human health, saving human life.
In recent years, with the research and development of portable intelligent ambulatory medical device, the measurement and diagnosis of electrocardiosignal are come into
People's lives.However, the rapid growth needs of electrocardiosignal take a substantial amount of time and undergo and carry out artificial screening, this increase
People's sees a doctor cost and diagnosis efficiency.Along with the development of artificial intelligence, machine is carried out by machine learning techniques and is examined automatically
The science and technology of disconnected heart disease, which is changed, becomes possibility.
Summary of the invention
In view of the defects existing in the prior art, the present invention, which provides, a kind of is learnt based on convolution residual error network and depth migration
Atrial fibrillation electrocardiosignal recognition methods in short-term, this method solve conventional machines first with the convolution residual error network in deep learning
Learn the problem low to atrial fibrillation electrocardiosignal accuracy of identification in short-term, then solves portable mobile apparatus institute using transfer learning
The problem of the sample of signal deficiency of atrial fibrillation in short-term of acquisition.
In order to solve the above technical problem, the present invention provides the following technical solutions:
A kind of atrial fibrillation recognition methods of electrocardiosignal in short-term based on convolution residual error network and transfer learning, including following step
It is rapid:
(1) segment of electrocardiogram (ECG) data in short-term of tape label is obtained as training data;
(2) convolutional neural networks and residual unit are constructed respectively and are combined into convolution residual error network;
(3) it using four disaggregated model of convolution residual error net structure and is trained, extracts the network under optimal training pattern
Parameter;
(4) continue Optimized model using transfer learning, save the network parameter under optimal models;
(5) disaggregated model for obtaining electrocardiogram (ECG) data input step (4) to be measured obtains differentiating result.
The present invention can be achieved for one section in short-term electrocardiosignal analyze, and judge its whether have the rhythm of the heart of atrial fibrillation mistake
Chang Tezheng, and then it is these four types of to provide normal signal, atrial fibrillation signal, the other diseases signal in addition to atrial fibrillation and noise signal
One of classification results.
Further, in the step (1), training dataset, can from the ecg signal data for intentionally clapping type label
It can also be the database comprising multi-source electrocardiosignal for self-built ecg signal data library, the ecg signal data library of open source;Together
When, it needs in database to include at least normal signal, atrial fibrillation signal, the other diseases signal in addition to atrial fibrillation and noise signal
These four types of labels, the label need to be marked to each electrocardiogram (ECG) data segment in short-term, which chooses with the short of above-mentioned four classes label
When electrocardiogram (ECG) data segment as training data, and need to guarantee that the quantity of above-mentioned four classes signal segment in training data is equal.
Further, in the step (4), mutually isostructural network is reinitialized, and carry out feature migration, will extracted
To optimized parameter be assigned to new migration network, including following sub-step:
(4-1) selected part testing data constructs electrocardiosignal fragment data collection, and data set includes normal signal, atrial fibrillation letter
Number, the other diseases signal in addition to atrial fibrillation and noise signal these fourth types label;
(4-2) construction migrates networks with the mutually isostructural more classification of reel product residual error network and carries out feature migration, that is, mentions
It takes the parameter under reel product residual error network optimal models and is assigned to new migration network;
(4-3) utilizes step (4-1) data set obtained, and the part convolution residual error layer of migration network is continued to do and optimized
Training, adjusting parameter;
(4-4) repeats step (4-3), after carrying out multiple parameter optimization, saves optimal models structure.
Further, the treatment process of the step (4-3) is as follows:
(4-3-1) freezes to migrate the preceding N1 layer parameter in network, i.e., fixed preceding N1 layers of parameter is not processed;
(4-3-2) is finely adjusted the preceding N2 layer parameter in rest network, is optimized to optimized parameter;
(4-3-3) reinitializes last N3 layer network in migration network, and re -training is to optimized parameter.
Beneficial effects of the present invention are mainly manifested in: solving the hardly possible of electrocardiosignal classification in short-term using convolutional neural networks
Problem, and the more common convolutional neural networks of convolution residual error network have more advantage, arithmetic accuracy is higher;Meanwhile transfer learning and volume
Product residual error network integration can well solve by signal difference alienation caused by the low problem of atrial fibrillation discrimination, improve model
Generalization ability and robustness.
Detailed description of the invention
Fig. 1 is that the present invention is based on the atrial fibrillation recognition methods processes of electrocardiosignal in short-term of convolution residual error network and transfer learning
Figure.
Fig. 2 is the convolution residual error network diagram constructed in the embodiment of the present invention.
Fig. 3 is the three kinds of migration network structures constructed in the embodiment of the present invention.
Specific embodiment
Specific implementation process of the present invention is described further with reference to the accompanying drawings of the specification.
Referring to Fig.1~Fig. 3, a kind of atrial fibrillation of electrocardiosignal in short-term identification side based on convolution residual error network and transfer learning
Method, comprising the following steps:
(1) segment of electrocardiogram (ECG) data in short-term of tape label is obtained as training data;
(2) convolutional neural networks and residual unit are constructed respectively and are combined into convolution residual error network;
(3) it using four disaggregated model of convolution residual error net structure and is trained, extracts the network under optimal training pattern
Parameter;
(4) continue Optimized model using transfer learning, save the network parameter under optimal models;
(5) disaggregated model for obtaining electrocardiogram (ECG) data input step (4) to be measured obtains differentiating result.
Referring to Fig.1, this method can realize for one section in short-term electrocardiosignal analyzed, and judge whether it has atrial fibrillation
Arrhythmia cordis feature, and then provide normal signal, atrial fibrillation signal, the other diseases signal in addition to atrial fibrillation and noise letter
One of number these four types of classification results.
Referring to Fig.1, in the step (1), training dataset, can from the ecg signal data for intentionally clapping type label
It can also be the database comprising multi-source electrocardiosignal for self-built ecg signal data library, the ecg signal data library of open source;Together
When, it needs in database to include at least normal signal, atrial fibrillation signal, the other diseases signal in addition to atrial fibrillation and noise signal
These four types of labels, the label need to be marked to each electrocardiogram (ECG) data segment in short-term, which chooses with the short of above-mentioned four classes label
When electrocardiogram (ECG) data segment as training data, and need to guarantee that the quantity of above-mentioned four classes signal segment in training data is equal.
Referring to Fig.1, Fig. 2, in the present embodiment, the convolution residual error network of step (2) building includes 1 non-residual error mould
Block and 16 residual error modules, specific structure are as follows:
Non- 1 structure of residual error module:
1) first layer: one-dimensional convolutional layer, convolution kernel is having a size of N, number of filter M, step-length S;
2) second layer: linear R eLu activation primitive layer;
3) third layer: batch normalization layer.
1 structure of residual error module:
1) first layer: one-dimensional convolutional layer, convolution kernel is having a size of N, number of filter M, step-length S;
2) second layer: batch normalization layer;
3) third layer: linear R eLu activation primitive layer;
4) the 4th layer: random drop layer, drop probability P;
5) layer 5: one-dimensional convolutional layer, convolution kernel is having a size of N, number of filter M, step-length S;
6) layer 6: one-dimensional maximum pond layer, pond window size are 2, step-length 2;
7) residual error layer: one-dimensional maximum pond layer, pond window size are 2, and the input of step-length 2, residual error layer comes from upper one
The last layer of the last layer of module, output and this module carries out linear superposition and constitutes residual error module.
2 structure of residual error module:
1) first layer: batch normalization layer;
2) second layer: linear R eLu activation primitive layer;
3) third layer: random drop layer, drop probability P;
4) the 4th layer: one-dimensional convolutional layer, convolution kernel is having a size of N, number of filter M, step-length S;
5) layer 5: batch normalization layer;
6) layer 6: linear R eLu activation primitive layer;
7) layer 7: random drop layer, drop probability P;
8) the 8th layer: one-dimensional convolutional layer, convolution kernel is having a size of N, number of filter M, step-length S;
9) residual error layer: one-dimensional maximum pond layer, pond window size are 2, and the input of step-length 2, residual error layer comes from upper one
The last layer of the last layer of module, output and this module carries out linear superposition and constitutes residual error module.
3 structure of residual error module:
1) first layer: batch normalization layer;
2) second layer: linear R eLu activation primitive layer;
3) third layer: random drop layer, drop probability P;
4) the 4th layer: one-dimensional convolutional layer, convolution kernel is having a size of N, number of filter M, step-length S;
5) layer 5: batch normalization layer;
6) layer 6: linear R eLu activation primitive layer;
7) layer 7: random drop layer, drop probability P;
8) the 8th layer: one-dimensional convolutional layer, convolution kernel is having a size of N, number of filter M, step-length S;
9) the 9th layer: one-dimensional maximum pond layer, pond window size are 2, step-length 2;
10) residual error layer: one-dimensional maximum pond layer, pond window size are 2, and the input of step-length 2, residual error layer comes from upper one
The last layer of the last layer of module, output and this module carries out linear superposition and constitutes residual error module.
Residual error module 4 and module 2 are identical.
Residual error module 5 and module 3 are identical.
The convolutional layer structure and module 2 of residual error module 6 are identical, and the 8th layer of convolution filter number is 2M.Residual error module 6
Residual error layer is one-dimensional volume base, and convolution kernel is having a size of N, number of filter 2M, step-length S.
7 structure of residual error module and module 3 are identical, and the 4th layer and the 8th layer of convolution filter number is 2M.
8 structure of residual error module and module 2 are identical, and the 4th layer and the 8th layer of convolution filter number is 2M.
Convolution module 9 and module 7 are identical.
10 structure of convolution module is identical with 6 structure of module, and the 4th layer of convolution filter number is 2M, the 8th layer of convolutional filtering
Device number is 3M, and residual error layer convolution filter number is 3M.
11 structure of convolution module and module 3 are identical, and the 4th layer and the 8th layer of convolution filter number is 3M.
12 structure of convolution module and module 2 are identical, and the 4th layer and the 8th layer of convolution filter number is 3M.
13 structure of convolution module and module 11 are identical.
14 structure of convolution module is identical with 6 structure of module, and the 4th layer of convolution filter number is 3M, the 8th layer of convolutional filtering
Device number is 4M, and residual error layer convolution filter number is 4M.
15 structure of convolution module and module 3 are identical, and the 4th layer and the 8th layer of convolution filter number is 4M.
16 structure of convolution module and module 2 are identical, and the 4th layer and the 8th layer of convolution filter number is 4M.
Referring to Fig.1, in the step (4), mutually isostructural network is reinitialized, and carry out feature migration, will extracted
To optimized parameter be assigned to new migration network, including following sub-step:
(4-1) selected part testing data constructs electrocardiosignal fragment data collection, and data set includes normal signal, atrial fibrillation letter
Number, the other diseases signal in addition to atrial fibrillation and noise signal these fourth types label;
(4-2) construction migrates networks with the mutually isostructural more classification of reel product residual error network and carries out feature migration, that is, mentions
It takes the parameter under reel product residual error network optimal models and is assigned to new migration network;
(4-3) utilizes step (4-1) data set obtained, and the part convolution residual error layer of migration network is continued to do and optimized
Training, adjusting parameter;
(4-4) repeats step (4-3), after carrying out multiple parameter optimization, saves optimal models structure.
Further, the treatment process of the step (4-3) is as follows:
(4-3-1) freezes to migrate the preceding N1 layer parameter in network, i.e., fixed preceding N1 layers of parameter is not processed;
(4-3-2) is finely adjusted the preceding N2 layer parameter in rest network, is optimized to optimized parameter;
(4-3-3) reinitializes last N3 layer network in migration network, and re -training is to optimized parameter.
Referring to Fig.1, Fig. 2, in the present embodiment, the corresponding migration network of step (4) specifically includes following three kinds of migrations knot
Structure:
1) construction is identical with reel product residual error network structure, and the last layer is that the migration network progress feature of four classification is moved
It moves, that is, extract the parameter under reel product residual error network optimal models and is assigned to new migration network, four kinds of classifications are as follows: normal letter
Number, atrial fibrillation signal, other abnormal signals, noise signal.
2.1) construction is identical with reel product residual error network structure, and the last layer is that the migration network progress feature of three classification is moved
It moves, that is, extract the parameter under reel product residual error network optimal models and is assigned to new migration network, three kinds of classifications are respectively as follows: just
Regular signal, abnormal signal and noise signal.
2.2) it is identical with reel product residual error network structure to construct another, the last layer is the migration network of two classification, right
Abnormal electrocardiogram signal is classified, that is, is extracted the parameter under reel product residual error network optimal models and be assigned to new migration net
Network, two categories are respectively atrial fibrillation signal and other abnormal signals.
3.1) construction is identical with reel product residual error network structure, and the last layer is that the migration network progress feature of two classification is moved
It moves, that is, extract the parameter under reel product residual error network optimal models and is assigned to new migration network, two categories, which are respectively as follows:, makes an uproar
Acoustical signal and non-noise signal.
3.2) it is identical with reel product residual error network structure to construct another, the last layer is the migration network of two classification, i.e.,
It extracts the parameter under reel product residual error network optimal models and is assigned to new migration network, two categories are respectively as follows: normal letter
Number and abnormal signal.
3.3) it is identical with reel product residual error network structure to reconstruct another, the last layer is the migration network of two classification,
It extracts the parameter under reel product residual error network optimal models and is assigned to new migration network, two categories are respectively as follows: atrial fibrillation
Signal and other abnormal signals.
Technical solution of the present invention is described in detail in embodiment described above, and the present invention can be directed to short
When electrocardiosignal carry out atrial fibrillation arrhythmia cordis identification.Embodiments above is merely to illustrate the present invention, rather than to of the invention
Limitation.Although the invention is described in detail with reference to an embodiment, those skilled in the art should understand that, to this hair
Bright technical solution carries out various combinations, modification or equivalent replacement, without departure from the spirit and scope of technical solution of the present invention,
It should all cover in scope of the presently claimed invention.
Claims (4)
1. a kind of atrial fibrillation recognition methods of electrocardiosignal in short-term based on convolution residual error network and transfer learning, which is characterized in that institute
State method the following steps are included:
(1) segment of electrocardiogram (ECG) data in short-term of tape label is obtained as training data;
(2) convolutional neural networks and residual unit are constructed respectively and are combined into convolution residual error network;
(3) it using four disaggregated model of convolution residual error net structure and is trained, extracts the network parameter under optimal training pattern;
(4) continue Optimized model using transfer learning, save the network parameter under optimal models;
(5) disaggregated model for obtaining electrocardiogram (ECG) data input step (4) to be measured obtains differentiating result.
2. a kind of atrial fibrillation of electrocardiosignal in short-term identification side based on convolution residual error network and transfer learning as described in claim 1
Method, which is characterized in that in the step (1), training dataset can be from the ecg signal data for intentionally clapping type label
Self-built ecg signal data library, open source ecg signal data library, can also be the database comprising multi-source electrocardiosignal;Together
When, it needs in database to include at least normal signal, atrial fibrillation signal, the other diseases signal in addition to atrial fibrillation and noise signal
These four types of labels, the label need to be marked to each electrocardiogram (ECG) data segment in short-term, which chooses with the short of above-mentioned four classes label
When electrocardiogram (ECG) data segment as training data, and need to guarantee that the quantity of above-mentioned four classes signal segment in training data is equal.
3. as claimed in claim 1 or 2 a kind of based on the knowledge of the atrial fibrillation of electrocardiosignal in short-term of convolution residual error network and transfer learning
Other method, which is characterized in that in the step (4), reinitialize mutually isostructural network, and carry out feature migration, will mention
The optimized parameter got is assigned to new migration network, including following sub-step:
(4-1) selected part testing data construct electrocardiosignal fragment data collection, data set include normal signal, atrial fibrillation signal,
Other diseases signal and noise signal these fourth types label in addition to atrial fibrillation;
(4-2) construction migrates networks with the mutually isostructural more classification of reel product residual error network and carries out feature migration, that is, extracts former
Parameter under convolution residual error network optimal models is simultaneously assigned to new migration network;
(4-3) utilizes step (4-1) data set obtained, continues to do optimization instruction to the part convolution residual error layer of migration network
Practice, adjusting parameter;
(4-4) repeats step (4-3), after carrying out multiple parameter optimization, saves optimal models structure.
4. a kind of atrial fibrillation of electrocardiosignal in short-term identification side based on convolution residual error network and transfer learning as claimed in claim 3
Method, which is characterized in that the treatment process of the step (4-3) is as follows:
(4-3-1) freezes to migrate the preceding N1 layer parameter in network, i.e., fixed preceding N1 layers of parameter is not processed;
(4-3-2) is finely adjusted the preceding N2 layer parameter in rest network, is optimized to optimized parameter;
(4-3-3) reinitializes last N3 layer network in migration network, and re -training is to optimized parameter.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811293013.7A CN109303560A (en) | 2018-11-01 | 2018-11-01 | A kind of atrial fibrillation recognition methods of electrocardiosignal in short-term based on convolution residual error network and transfer learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811293013.7A CN109303560A (en) | 2018-11-01 | 2018-11-01 | A kind of atrial fibrillation recognition methods of electrocardiosignal in short-term based on convolution residual error network and transfer learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109303560A true CN109303560A (en) | 2019-02-05 |
Family
ID=65222692
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811293013.7A Pending CN109303560A (en) | 2018-11-01 | 2018-11-01 | A kind of atrial fibrillation recognition methods of electrocardiosignal in short-term based on convolution residual error network and transfer learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109303560A (en) |
Cited By (24)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110046604A (en) * | 2019-04-25 | 2019-07-23 | 成都信息工程大学 | A kind of single lead ECG arrhythmia detection classification method based on residual error network |
CN110148466A (en) * | 2019-05-15 | 2019-08-20 | 东北大学 | A kind of heart impact signal atrial fibrillation computer aided diagnosing method based on transfer learning |
CN110236525A (en) * | 2019-06-27 | 2019-09-17 | 上海数创医疗科技有限公司 | A kind of Ventricular Tachycardia rhythm of the heart identification neural network based on transfer learning |
CN110488607A (en) * | 2019-08-13 | 2019-11-22 | 湘潭大学 | A kind of recognition methods that lathe tool is worn based on convolution residual error network and transfer learning |
CN110547792A (en) * | 2019-07-31 | 2019-12-10 | 平安科技(深圳)有限公司 | atrial fibrillation detection method and device, computer equipment and storage medium |
CN110688942A (en) * | 2019-09-25 | 2020-01-14 | 西安邮电大学 | Electrocardiosignal joint identification module and method based on InResNet network |
CN110728654A (en) * | 2019-09-06 | 2020-01-24 | 台州学院 | Automatic pipeline detection and classification method based on deep residual error neural network |
CN110889448A (en) * | 2019-11-26 | 2020-03-17 | 北京华医共享医疗科技有限公司 | Electrocardiogram classification method based on convolutional neural network |
CN110897630A (en) * | 2019-06-27 | 2020-03-24 | 上海数创医疗科技有限公司 | Ventricular tachycardia and heart rhythm identification method and device based on transfer learning |
CN110995475A (en) * | 2019-11-20 | 2020-04-10 | 国网湖北省电力有限公司信息通信公司 | Power communication network fault detection method based on transfer learning |
CN110974214A (en) * | 2019-12-20 | 2020-04-10 | 华中科技大学 | Automatic electrocardiogram classification method, system and equipment based on deep learning |
CN111134662A (en) * | 2020-02-17 | 2020-05-12 | 武汉大学 | Electrocardio abnormal signal identification method and device based on transfer learning and confidence degree selection |
CN111178439A (en) * | 2019-12-31 | 2020-05-19 | 杭州电子科技大学 | SAR image classification method based on convolutional neural network and fine adjustment |
CN111329445A (en) * | 2020-02-20 | 2020-06-26 | 广东工业大学 | Atrial fibrillation identification method based on group convolution residual error network and long-term and short-term memory network |
CN111803028A (en) * | 2020-06-19 | 2020-10-23 | 天津大学 | Individualized epileptic seizure detection device based on network parameter migration |
CN111956212A (en) * | 2020-07-29 | 2020-11-20 | 鲁东大学 | Inter-group atrial fibrillation identification method based on frequency domain filtering-multi-mode deep neural network |
CN112329609A (en) * | 2020-11-03 | 2021-02-05 | 山东大学 | Feature fusion transfer learning arrhythmia classification system based on 2D heart beat |
CN112966591A (en) * | 2021-03-03 | 2021-06-15 | 河北工业职业技术学院 | Knowledge map deep reinforcement learning migration system for mechanical arm grabbing task |
CN113274031A (en) * | 2021-04-30 | 2021-08-20 | 西安理工大学 | Arrhythmia classification method based on deep convolution residual error network |
CN113288161A (en) * | 2021-06-02 | 2021-08-24 | 南京信息工程大学 | Single-lead ECG signal classification method and system based on improved residual error network |
CN113288162A (en) * | 2021-06-03 | 2021-08-24 | 北京航空航天大学 | Short-term electrocardiosignal atrial fibrillation automatic detection system based on self-adaptive attention mechanism |
CN113627553A (en) * | 2021-08-18 | 2021-11-09 | 国网福建省电力有限公司营销服务中心 | Image identification method and system for electric energy meter abnormal label identification |
CN114287884A (en) * | 2021-12-25 | 2022-04-08 | 肇庆星网医疗科技有限公司 | Noise detection method and system based on artificial intelligence |
CN115019893A (en) * | 2022-06-14 | 2022-09-06 | 邵阳学院 | Enhancer identification method based on bidirectional long-and-short-term memory and attention mechanism |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130260357A1 (en) * | 2012-03-27 | 2013-10-03 | Lauren Reinerman-Jones | Skill Screening |
CN107203692A (en) * | 2017-05-09 | 2017-09-26 | 哈尔滨工业大学(威海) | The implementation method of atrial fibrillation detection based on depth convolutional neural networks |
CN107545245A (en) * | 2017-08-14 | 2018-01-05 | 中国科学院半导体研究所 | A kind of age estimation method and equipment |
CN107595276A (en) * | 2017-08-22 | 2018-01-19 | 南京易哈科技有限公司 | A kind of atrial fibrillation detection method based on single lead electrocardiosignal time-frequency characteristics |
CN108564122A (en) * | 2018-04-10 | 2018-09-21 | 北京工业大学 | The integrated training method of image description model based on convolution-cycle hybrid network |
CN108606798A (en) * | 2018-05-10 | 2018-10-02 | 东北大学 | Contactless atrial fibrillation intelligent checking system based on depth convolution residual error network |
-
2018
- 2018-11-01 CN CN201811293013.7A patent/CN109303560A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130260357A1 (en) * | 2012-03-27 | 2013-10-03 | Lauren Reinerman-Jones | Skill Screening |
CN107203692A (en) * | 2017-05-09 | 2017-09-26 | 哈尔滨工业大学(威海) | The implementation method of atrial fibrillation detection based on depth convolutional neural networks |
CN107545245A (en) * | 2017-08-14 | 2018-01-05 | 中国科学院半导体研究所 | A kind of age estimation method and equipment |
CN107595276A (en) * | 2017-08-22 | 2018-01-19 | 南京易哈科技有限公司 | A kind of atrial fibrillation detection method based on single lead electrocardiosignal time-frequency characteristics |
CN108564122A (en) * | 2018-04-10 | 2018-09-21 | 北京工业大学 | The integrated training method of image description model based on convolution-cycle hybrid network |
CN108606798A (en) * | 2018-05-10 | 2018-10-02 | 东北大学 | Contactless atrial fibrillation intelligent checking system based on depth convolution residual error network |
Cited By (33)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110046604B (en) * | 2019-04-25 | 2023-05-30 | 成都信息工程大学 | Single-lead ECG arrhythmia detection and classification method based on residual error network |
CN110046604A (en) * | 2019-04-25 | 2019-07-23 | 成都信息工程大学 | A kind of single lead ECG arrhythmia detection classification method based on residual error network |
CN110148466A (en) * | 2019-05-15 | 2019-08-20 | 东北大学 | A kind of heart impact signal atrial fibrillation computer aided diagnosing method based on transfer learning |
CN110897630A (en) * | 2019-06-27 | 2020-03-24 | 上海数创医疗科技有限公司 | Ventricular tachycardia and heart rhythm identification method and device based on transfer learning |
CN110236525A (en) * | 2019-06-27 | 2019-09-17 | 上海数创医疗科技有限公司 | A kind of Ventricular Tachycardia rhythm of the heart identification neural network based on transfer learning |
CN110547792A (en) * | 2019-07-31 | 2019-12-10 | 平安科技(深圳)有限公司 | atrial fibrillation detection method and device, computer equipment and storage medium |
CN110547792B (en) * | 2019-07-31 | 2022-02-15 | 平安科技(深圳)有限公司 | Atrial fibrillation detection method and device, computer equipment and storage medium |
CN110488607A (en) * | 2019-08-13 | 2019-11-22 | 湘潭大学 | A kind of recognition methods that lathe tool is worn based on convolution residual error network and transfer learning |
CN110728654A (en) * | 2019-09-06 | 2020-01-24 | 台州学院 | Automatic pipeline detection and classification method based on deep residual error neural network |
CN110728654B (en) * | 2019-09-06 | 2023-01-10 | 台州学院 | Automatic pipeline detection and classification method based on deep residual error neural network |
CN110688942A (en) * | 2019-09-25 | 2020-01-14 | 西安邮电大学 | Electrocardiosignal joint identification module and method based on InResNet network |
CN110995475A (en) * | 2019-11-20 | 2020-04-10 | 国网湖北省电力有限公司信息通信公司 | Power communication network fault detection method based on transfer learning |
CN110889448A (en) * | 2019-11-26 | 2020-03-17 | 北京华医共享医疗科技有限公司 | Electrocardiogram classification method based on convolutional neural network |
CN110974214A (en) * | 2019-12-20 | 2020-04-10 | 华中科技大学 | Automatic electrocardiogram classification method, system and equipment based on deep learning |
CN111178439A (en) * | 2019-12-31 | 2020-05-19 | 杭州电子科技大学 | SAR image classification method based on convolutional neural network and fine adjustment |
CN111134662B (en) * | 2020-02-17 | 2021-04-16 | 武汉大学 | Electrocardio abnormal signal identification method and device based on transfer learning and confidence degree selection |
CN111134662A (en) * | 2020-02-17 | 2020-05-12 | 武汉大学 | Electrocardio abnormal signal identification method and device based on transfer learning and confidence degree selection |
CN111329445A (en) * | 2020-02-20 | 2020-06-26 | 广东工业大学 | Atrial fibrillation identification method based on group convolution residual error network and long-term and short-term memory network |
CN111329445B (en) * | 2020-02-20 | 2023-09-15 | 广东工业大学 | Atrial fibrillation identification method based on group convolution residual error network and long-term and short-term memory network |
CN111803028A (en) * | 2020-06-19 | 2020-10-23 | 天津大学 | Individualized epileptic seizure detection device based on network parameter migration |
CN111956212A (en) * | 2020-07-29 | 2020-11-20 | 鲁东大学 | Inter-group atrial fibrillation identification method based on frequency domain filtering-multi-mode deep neural network |
CN111956212B (en) * | 2020-07-29 | 2023-08-01 | 鲁东大学 | Inter-group atrial fibrillation recognition method based on frequency domain filtering-multi-mode deep neural network |
CN112329609A (en) * | 2020-11-03 | 2021-02-05 | 山东大学 | Feature fusion transfer learning arrhythmia classification system based on 2D heart beat |
CN112966591A (en) * | 2021-03-03 | 2021-06-15 | 河北工业职业技术学院 | Knowledge map deep reinforcement learning migration system for mechanical arm grabbing task |
CN113274031A (en) * | 2021-04-30 | 2021-08-20 | 西安理工大学 | Arrhythmia classification method based on deep convolution residual error network |
CN113274031B (en) * | 2021-04-30 | 2023-12-29 | 西安理工大学 | Arrhythmia classification method based on depth convolution residual error network |
CN113288161A (en) * | 2021-06-02 | 2021-08-24 | 南京信息工程大学 | Single-lead ECG signal classification method and system based on improved residual error network |
CN113288162B (en) * | 2021-06-03 | 2022-06-28 | 北京航空航天大学 | Short-term electrocardiosignal atrial fibrillation automatic detection system based on self-adaptive attention mechanism |
CN113288162A (en) * | 2021-06-03 | 2021-08-24 | 北京航空航天大学 | Short-term electrocardiosignal atrial fibrillation automatic detection system based on self-adaptive attention mechanism |
CN113627553A (en) * | 2021-08-18 | 2021-11-09 | 国网福建省电力有限公司营销服务中心 | Image identification method and system for electric energy meter abnormal label identification |
CN113627553B (en) * | 2021-08-18 | 2024-03-15 | 国网福建省电力有限公司营销服务中心 | Image recognition method and system for recognizing abnormal label of electric energy meter |
CN114287884A (en) * | 2021-12-25 | 2022-04-08 | 肇庆星网医疗科技有限公司 | Noise detection method and system based on artificial intelligence |
CN115019893A (en) * | 2022-06-14 | 2022-09-06 | 邵阳学院 | Enhancer identification method based on bidirectional long-and-short-term memory and attention mechanism |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109303560A (en) | A kind of atrial fibrillation recognition methods of electrocardiosignal in short-term based on convolution residual error network and transfer learning | |
Zhao et al. | Noise rejection for wearable ECGs using modified frequency slice wavelet transform and convolutional neural networks | |
Oh et al. | Classification of heart sound signals using a novel deep WaveNet model | |
CN111657926B (en) | Arrhythmia classification method based on multi-lead information fusion | |
CN107822622B (en) | Electrocardiogram diagnosis method and system based on deep convolutional neural network | |
CN109948647B (en) | Electrocardiogram classification method and system based on depth residual error network | |
US11324455B2 (en) | Artificial intelligence-based interference recognition method for electrocardiogram | |
CN109820525A (en) | A kind of driving fatigue recognition methods based on CNN-LSTM deep learning model | |
CN109998500A (en) | A kind of pulse signal generation method and system based on production confrontation network | |
CN108596069A (en) | Neonatal pain expression recognition method and system based on depth 3D residual error networks | |
CN111297349A (en) | Machine learning-based heart rhythm classification system | |
CN107103182A (en) | A kind of heart disease Warning System and method based on deep learning algorithm | |
CN109864714A (en) | A kind of ECG Signal Analysis method based on deep learning | |
CN110619322A (en) | Multi-lead electrocardio abnormal signal identification method and system based on multi-flow convolution cyclic neural network | |
CN110363090A (en) | Intelligent heart disease detection method, device and computer readable storage medium | |
CN112906748A (en) | 12-lead ECG arrhythmia detection classification model construction method based on residual error network | |
CN111759345A (en) | Heart valve abnormality analysis method, system and device based on convolutional neural network | |
Li et al. | EEG signal classification method based on feature priority analysis and CNN | |
Ullah et al. | [Retracted] An Effective and Lightweight Deep Electrocardiography Arrhythmia Recognition Model Using Novel Special and Native Structural Regularization Techniques on Cardiac Signal | |
CN106355574A (en) | Intra-abdominal adipose tissue segmentation method based on deep learning | |
Jayasekara et al. | Timecaps: Capturing time series data with capsule networks | |
Wang et al. | Multiscale residual network based on channel spatial attention mechanism for multilabel ECG classification | |
EP4149351A1 (en) | System and method for testing for sars-cov-2/covid-19 based on wearable medical sensors and neural networks | |
CN110960207A (en) | Tree model-based atrial fibrillation detection method, device, equipment and storage medium | |
CN114521900B (en) | Arrhythmia classification and identification method based on transfer learning |
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: 20190205 |