CN110897628A - Deep neural network-based electrocardiogram signal feature extraction method, device, system, equipment and classification method - Google Patents
Deep neural network-based electrocardiogram signal feature extraction method, device, system, equipment and classification method Download PDFInfo
- Publication number
- CN110897628A CN110897628A CN201811075514.8A CN201811075514A CN110897628A CN 110897628 A CN110897628 A CN 110897628A CN 201811075514 A CN201811075514 A CN 201811075514A CN 110897628 A CN110897628 A CN 110897628A
- Authority
- CN
- China
- Prior art keywords
- feature extraction
- electrocardiogram signal
- electrocardiogram
- neural network
- deep neural
- 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
Images
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
-
- 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
-
- 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)
- Artificial Intelligence (AREA)
- Animal Behavior & Ethology (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Pathology (AREA)
- Veterinary Medicine (AREA)
- Cardiology (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention discloses an electrocardio feature extraction method, a device, a system, equipment and a classification method based on a deep neural network, wherein the electrocardio feature extraction method based on the deep neural network comprises the following steps: randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, wherein the electrocardiogram signal at least comprises two cardiac cycles; inputting the intercepted electrocardiogram signal into a feature extraction model in a picture form, and extracting to obtain electrocardiogram signal features; the feature extraction model is obtained based on VGG model training. The electrocardio-feature extraction method, the electrocardio-feature extraction device, the electrocardio-feature extraction system, the electrocardio-feature extraction equipment and the electrocardio-feature extraction classification method based on the deep neural network provided by the invention can reduce the incompleteness brought by artificial design features and improve the accuracy and diversity of the electrocardio-feature extraction based on the deep neural network.
Description
Technical Field
The invention relates to the technical field of electrocardiogram signal processing, in particular to an electrocardiogram signal feature extraction method, device, system, equipment and classification method based on a deep neural network.
Background
Electrocardiographic (ECG) is the most important means for detecting and diagnosing heart diseases at present, and human electrocardiographic signals (ECG) are the comprehensive expression of the electrical activity of the heart on the body surface, and the physiological conditions of each part of the heart can be obtained by extracting the characteristic information in the ECG.
Taking arrhythmia as an example, the current arrhythmia analysis mainly adopts a waveform analysis method and a template matching method. The waveform analysis method firstly obtains characteristic waveform parameters, such as amplitude, time length, rising/falling time, waveform interval and the like of the characteristic waveform, and the waveform parameters are compared with a judgment threshold value obtained according to clinical experience to obtain an arrhythmia analysis result. The template matching method is mainly characterized in that an average R-R interval and an average R-wave form in a electrocardiosignal of a detected person are calculated to serve as templates, the R-R interval and the R-wave form of each heartbeat of the detected person are compared with the templates, and if the difference between the R-R interval and the R-wave form exceeds a certain range, arrhythmia is considered to occur.
The waveform analysis method adopts the empirical threshold of the solidified characteristic waveform parameters as a judgment basis, is simpler and more intuitive, has fewer characteristic values and limited classification types, is very sensitive to noise in the form of the electrocardiogram, has disordered definitions of various transform domains and statistical methods on arrhythmia types, and has different classification results and effects. The template matching method can only make effective judgment when the R wave form of the detected person is greatly different from the template, and cannot effectively judge the arrhythmia waveform with unobvious difference.
The human body is a nonlinear complex system, the form of individual electrocardiosignals can change along with time, the health condition of the body also has great influence on the electrocardiosignals, the electrocardiosignal data is very large and complex, and more sufficient characteristic information extraction is required.
Disclosure of Invention
The invention provides an electrocardiogram signal feature extraction method, device, system, equipment and classification method based on a deep neural network, which improve the accuracy and diversity of electrocardiogram signal feature extraction.
An electrocardiogram signal feature extraction method based on a deep neural network comprises the following steps:
randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, wherein the electrocardiogram signal at least comprises two cardiac cycles;
inputting the intercepted electrocardiogram signal into a feature extraction model in a picture form, and extracting to obtain electrocardiogram signal features; the feature extraction model is obtained based on VGG model training.
Several alternatives are provided below, but not as an additional limitation to the above general solution, but merely as a further addition or preference, each alternative being combinable individually for the above general solution or among several alternatives without technical or logical contradictions.
The features extracted by the feature extraction model are to be understood broadly, and include both the information extracted directly from the electrocardiogram, and the information obtained by further processing the directly extracted information, including processing such as classification.
Optionally, the feature extraction model comprises a convolutional layer, a pooling layer and a fully-connected layer, wherein the convolutional layer and the pooling layer both use a Relu activation function or a variant thereof, and the fully-connected layer uses a softmax function.
Optionally, a first convolutional layer of the convolutional layers comprises 3 convolutional channels, and a convolution kernel of each convolutional channel is { {1, 1, 1}, {1, 1, 1}, {1, 1, 1} }.
Optionally, the pooling layer adopts a maximum value method, and the pooling window size is (2, 2).
Optionally, the fully-connected layer includes three layers, the number of neurons in the first layer is 4096, the number of neurons in the second layer is 1024, and the number of neurons in the third layer is 20.
An electrocardiogram signal feature extraction device based on a deep neural network comprises:
the system comprises a preprocessing module, a processing module and a processing module, wherein the preprocessing module is used for randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, and the electrocardiogram signal at least comprises two cardiac cycles;
the extraction module is used for inputting the intercepted electrocardiogram signal into the feature extraction model in a picture form and extracting the feature of the electrocardiogram signal; the feature extraction model is obtained based on VGG model training.
A computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, wherein the electrocardiogram signal at least comprises two cardiac cycles;
inputting the intercepted electrocardiogram signal into a feature extraction model in a picture form, and extracting to obtain electrocardiogram signal features; the feature extraction model is obtained based on VGG model training.
An electrocardiogram signal feature extraction system based on a deep neural network comprises a terminal and a server, wherein the server comprises a memory and a processor, the memory stores a computer program, and the server acquires an electrocardiogram from the terminal; when the processor executes the computer program, the following steps are realized:
randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, wherein the electrocardiogram signal at least comprises two cardiac cycles;
inputting the intercepted electrocardiogram signal into a feature extraction model in a picture form, and extracting to obtain electrocardiogram signal features; the feature extraction model is obtained based on VGG model training.
A computer-readable storage medium having a computer program stored therein, the computer program when executed by a computer processor implementing the steps of:
randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, wherein the electrocardiogram signal at least comprises two cardiac cycles;
inputting the intercepted electrocardiogram signal into a feature extraction model in a picture form, and extracting to obtain electrocardiogram signal features; the feature extraction model is obtained based on VGG model training.
An electrocardiogram signal classification method based on a deep neural network comprises the following steps:
randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, wherein the electrocardiogram signal at least comprises two cardiac cycles;
inputting the intercepted electrocardiogram signals into a classification model in a picture mode to obtain a classification result; the classification model is obtained based on VGG model training.
The electrocardiogram signal feature extraction method, the device, the system, the equipment and the classification method provided by the invention can reduce the incompleteness brought by artificial design features and improve the accuracy and diversity of electrocardiogram signal feature extraction and classification.
Drawings
FIG. 1 is a flow chart of a deep neural network-based ECG signal feature extraction method in one embodiment;
FIG. 2 is a schematic diagram of an intercepted ECG signal;
FIG. 3 is a schematic diagram of a computer device in one embodiment;
FIG. 4 is a flowchart of a deep neural network-based ECG signal feature extraction method in one embodiment.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1 and 4, an electrocardiogram signal feature extraction method based on a deep neural network includes the following steps:
in step S1, a continuous electrocardiogram signal is randomly captured from the twelve-lead electrocardiogram to be processed, wherein the electrocardiogram signal comprises at least two cardiac cycles.
The ECG signal (i.e., electrocardiogram signal) may be a unipolar ECG signal or a bipolar ECG signal. For example, in a 12-lead ECG, the electrical heart activity of an individual may be recorded by one or more electrodes placed on the surface of the body. A 12-lead ECG may provide spatial information about the electrical heart activity, with each of the 12 leads representing a particular position in space at which the electrical heart activity of the heart is measured.
Of the 12 leads, lead I (right arm to left arm), lead II (right arm to left leg) and lead III (left arm to left leg) are bipolar limb leads, and leads V1, V2, V3, V4, V5 and V6 are unipolar chest leads. To form a bipolar limb lead for measurement, two electrodes are required to contact two skin surfaces of two limbs (e.g., right and left arms), respectively. To form a monopolar chest lead for measurement, an electrode is required to be in contact with the skin surface of the chest. The term "contact" as used herein may refer to direct contact of an electrode with the exposed skin, or indirect contact between an electrode and the exposed skin with a conductive material (e.g., a conductive patch or conductive garment).
The method comprises the steps of taking originally acquired electrocardiogram data as discrete data, fitting by adopting a Bslpine interpolation method to obtain a one-dimensional ECG signal, preprocessing the one-dimensional ECG signal to generate a noise reduction signal, and then extracting characteristics of the electrocardiogram signal.
Noise sources of the ECG signal include: the pretreatment can adopt the prior art, such as removing the baseline drift, power frequency interference and high frequency noise in the electrocardiogram signal by wavelet transformation or a filter.
Step S2, inputting the captured electrocardiogram signal into a feature extraction model in a picture mode, and extracting the features of the electrocardiogram signal; the feature extraction model is obtained based on VGG model training.
The electrocardiogram signal is input to the feature extraction model in the form of a picture as shown in fig. 2, and the picture size of the electrocardiogram signal is 1200 × 600 pixels.
In the prior art, the characteristics are generally required to be artificially designed in the process of extracting the electrocardiogram characteristics, missing detection and misjudgment are easy to occur, the characteristic extraction model in the embodiment is obtained based on VGG model training, the characteristics of comprehensively describing the electrocardiogram can be automatically and effectively extracted from a large number of electrocardiograms, the integrity and diversity of electrocardiogram signal characteristic extraction are improved, the robustness is stronger, and the detection precision is better.
After the feature extraction model is used for a preset time, the updated electrocardiogram data is adopted for retraining, and the feature extraction model is guaranteed to have better extraction accuracy.
In one embodiment, the feature extraction model includes a convolutional layer, a pooling layer, and a fully-connected layer, wherein the convolutional layer and the pooling layer both employ a Relu activation function or a variant thereof, and the fully-connected layer employs a softmax function.
In one embodiment, the first convolutional layer of the convolutional layers comprises 3 convolutional channels, and the convolution kernel of each convolutional channel is { {1, 1, 1}, {1, 1, 1}, {1, 1, 1} }.
The first convolution layer is used for enhancing the boundary information of the picture.
In one embodiment, the pooling layer employs a max-measure, pooling window size is (2, 2).
The pooling layer is used for reducing dimensions, so that operation parameters are reduced, and the calculation speed is increased.
In one embodiment, the fully-connected layer comprises three layers, the number of neurons in the first layer is 4096, the number of neurons in the second layer is 1024, and the number of neurons in the third layer is 20.
The full connection layer is used for integrating multi-channel data into one-dimensional feature vectors, and the abstract one-dimensional feature vectors are classified in the full connection layer according to the characteristics of the multilayer neural network.
The number of neurons in the third layer of the full junction layer determines the number of extracted electrocardiographic features, and in this embodiment, 20 electrocardiographic features can be extracted.
Training of the feature extraction model is carried out in a deep learning framework TensorFlow and Keras, and the training process of the feature extraction model comprises the following steps:
step a, constructing a training database and a testing database.
The data of the training database and the testing database can come from hospitals and physical examination centers, the quantity of electrocardiograms of the training database and the testing database is not less than 10 thousands in total, 70% of the electrocardiograms are normal electrocardiograms, the rest are abnormal electrocardiograms, and abnormal characteristics of the electrocardiograms are manually and independently marked by at least two electrocardiogram experts. The electrocardiogram data in the training database and the test database are not crossed with each other, and the number of the electrocardiogram data in the training database is larger than that in the test database.
The electrocardiogram data in the training database and the test database come from different individuals, including males aged between 10 and 99 years and females aged between 10 and 99 years, the proportion of the males and the females accounts for half respectively, the training database and the test database are randomly selected from a data set of an electrocardiograph, and the training database and the test database comprise various waveforms and artifact record numbers with representative significance, various unusual but important clinical data, and some complex ventricular, nodal and supraventricular arrhythmias and conduction abnormalities.
Each piece of electrocardiosignal data in the training database and the testing database comprises the format, sampling frequency and length of electrocardiosignals, relevant information of a patient, such as a collection place, the state of illness of the patient, the medication condition and the like, recorded by the electrocardiosignal data, and the result of electrocardiosignal analysis of an electrocardio expert, which mainly comprises information of heartbeat, rhythm, signal quality and the like.
The annotation of each piece of electrocardiographic signal data in the training database and the testing database contains basic information of the electrocardiographic signal and a result of signal analysis by an electrocardiographic diagnostician, such as: signal duration, heart rate, signal quality, location, number of arrhythmia beats, arrhythmia characteristics, and the like.
The abnormal characteristics of the electric signals of the training database and the test data center determine the characteristics which can be extracted by the characteristic extraction model.
The features of the cardiac signal include, but are not limited to, features that embody the following heartbeat types: normal, pacing, sinus tachycardia, sinus bradycardia, sinus arrhythmia, sinus asystole, escape and escape atrial premature beats, supraventricular bigement, ventricular premature beats, fast ventricular rate, slow ventricular rate, supraventricular tachycardia, ventricular tachycardia, atrial flutter, atrial fibrillation, atrioventricular conduction block, intraventricular block, noise, and others.
And b, preprocessing each electrocardiogram in the training database and the testing database.
Randomly intercepting a continuous electrocardiogram signal aiming at one electrocardiogram in the training database and the testing database, wherein the electrocardiogram signal at least comprises two cardiac cycles.
The electrocardiogram signal is converted into a picture form in the same manner as the electrocardiogram signal is converted into a picture form in step S2.
The electrocardiogram signals converted into the image form in the training database are used as the input of the VGG model for training, the model obtained by training is tested by using the electrocardiogram in the test database, and the test results are divided into four types: true Positive (TP), True Negative (TN), False Positive (FP), False Negative (FN).
And evaluating the feature extraction result by using four results, wherein the evaluation adopts the following four parameters:
sensitivity (Sen), the formula for calculation is: sen ═ TP/(TP + FN);
specificity (Spe), calculated by the formula: Spe-TN/(FP + TN);
the positive rate (PPV) is calculated by the formula: PPV ═ TP/(TP + FP);
accuracy (Acc), the calculation formula is: acc ═ TP + TN)/(TP + FP + FN + TN).
The evaluation results are shown in table 1.
TABLE 1
Sen | Spe | Acc | PPV |
98% | 97.6% | 98.0% | 97.1% |
It should be understood that, although the steps in the flowchart of fig. 1 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a portion of the steps in fig. 1 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.
In one embodiment, an apparatus for extracting features of an electrocardiogram signal based on a deep neural network includes:
the system comprises a preprocessing module, a processing module and a processing module, wherein the preprocessing module is used for randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, and the electrocardiogram signal at least comprises two cardiac cycles;
the extraction module is used for inputting the intercepted electrocardiogram signal into the feature extraction model in a picture form and extracting the feature of the electrocardiogram signal; the feature extraction model is obtained based on VGG model training.
For the function definition in each module, reference may be made to the above definition of the electrocardiogram signal feature extraction method, which is not described herein again. All or part of the modules in the electrocardiogram signal feature extraction device can be realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
The electrocardiogram signal feature extraction device provided by the embodiment can be configured at a remote end, and acquires an electrocardiogram signal through a remote terminal connected with the device, or the device of the embodiment itself can be configured at a terminal (such as a computer or medical detection equipment used by a user), and directly acquires an electrocardiogram signal through an electrocardiogram acquisition device.
In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as shown in fig. 3. The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement an electrocardiogram signal feature extraction method. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on the shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like.
Those skilled in the art will appreciate that the architecture shown in fig. 3 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, there is provided a computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, wherein the electrocardiogram signal at least comprises two cardiac cycles;
inputting the intercepted electrocardiogram signal into a feature extraction model in a picture form, and extracting to obtain electrocardiogram signal features; the feature extraction model is obtained based on VGG model training.
In one embodiment, an electrocardiogram signal feature extraction system based on a deep neural network is provided, and comprises a terminal and a server, wherein the server comprises a memory and a processor, the memory stores a computer program, and the server acquires an electrocardiogram from the terminal; when the processor executes the computer program, the following steps are realized:
randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, wherein the electrocardiogram signal at least comprises two cardiac cycles;
inputting the intercepted electrocardiogram signal into a feature extraction model in a picture form, and extracting to obtain electrocardiogram signal features; the feature extraction model is obtained based on VGG model training.
The terminal may be, but is not limited to: the server can be a remote background server or a server in a cloud platform, and can be realized by adopting an independent server or a server cluster formed by a plurality of servers, the electrocardiogram is transmitted between the terminal and the server through a communication network, and the communication network can be, but is not limited to 3G, 4G and 5G, wifi.
In addition to transmitting the original electrocardiogram from the terminal to the server, information related to the electrocardiogram including, but not limited to, user information, time of detection may be transmitted from the terminal to the server. The server receives the electrocardiogram data and transmits the extracted characteristic result to the terminal through the communication network. The server stores the received electrocardiogram data and the feature extraction result.
In one embodiment, a computer readable storage medium is provided, having a computer program stored therein, which when executed by a computer processor, performs the steps of:
randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, wherein the electrocardiogram signal at least comprises two cardiac cycles;
inputting the intercepted electrocardiogram signal into a feature extraction model in a picture form, and extracting to obtain electrocardiogram signal features; the feature extraction model is obtained based on VGG model training.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a non-volatile computer-readable storage medium, and can include the processes of the embodiments of the methods described above when the computer program is executed. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
In one embodiment, a deep neural network-based electrocardiogram signal classification method is provided, which includes the following steps:
randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, wherein the electrocardiogram signal at least comprises two cardiac cycles;
inputting the intercepted electrocardiogram signals into a classification model in a picture mode to obtain a classification result; the classification model is obtained based on VGG model training.
Categorical categories of electrocardiogram signals include, but are not limited to: normal, pacing, sinus tachycardia, sinus bradycardia, sinus arrhythmia, sinus asystole, escape and escape atrial premature beats, supraventricular bigement, ventricular premature beats, fast ventricular rate, slow ventricular rate, supraventricular tachycardia, ventricular tachycardia, atrial flutter, atrial fibrillation, atrioventricular conduction block, intraventricular block, noise, and others.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (9)
1. An electrocardiogram signal feature extraction method based on a deep neural network is characterized by comprising the following steps:
randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, wherein the electrocardiogram signal at least comprises two cardiac cycles;
inputting the intercepted electrocardiogram signal into a feature extraction model in a picture form, and extracting to obtain electrocardiogram signal features; the feature extraction model is obtained based on VGG model training.
2. The deep neural network-based electrocardiographic signal feature extraction method according to claim 1, wherein the feature extraction model comprises a convolutional layer, a pooling layer and a fully-connected layer, wherein the convolutional layer and the pooling layer both use Relu activation functions or variants thereof, and the fully-connected layer uses softmax functions.
3. The method according to claim 2, wherein the first convolutional layer of the convolutional layers comprises 3 convolutional channels, and the convolutional core of each convolutional channel is { {1, 1, 1}, {1, 1, 1}, {1, 1, 1} }.
4. The deep neural network-based electrocardiogram signal feature extraction method according to claim 2, wherein the pooling layer adopts a maximum value method, and the pooling window size is (2, 2).
5. The method according to claim 2, wherein the fully-connected layer comprises three layers, the number of neurons in the first layer is 4096, the number of neurons in the second layer is 1024, and the number of neurons in the third layer is 20.
6. An electrocardiogram signal feature extraction device based on a deep neural network is characterized by comprising:
the system comprises a preprocessing module, a processing module and a processing module, wherein the preprocessing module is used for randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, and the electrocardiogram signal at least comprises two cardiac cycles;
the extraction module is used for inputting the intercepted electrocardiogram signal into the feature extraction model in a picture form and extracting the feature of the electrocardiogram signal; the feature extraction model is obtained based on VGG model training.
7. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor when executing the computer program implements the method for deep neural network-based electrocardiographic signal feature extraction according to any one of claims 1 to 5.
8. An electrocardiogram signal feature extraction system based on a deep neural network comprises a terminal and a server, wherein the server comprises a memory and a processor, the memory stores a computer program, and the server is characterized in that the server acquires an electrocardiogram from the terminal; the processor, when executing the computer program, implements the method for extracting features of an electrocardiogram signal based on a deep neural network according to any one of claims 1 to 5.
9. A deep neural network-based electrocardiogram signal classification method is characterized by comprising the following steps:
randomly intercepting a continuous electrocardiogram signal in a twelve-lead electrocardiogram to be processed, wherein the electrocardiogram signal at least comprises two cardiac cycles;
inputting the intercepted electrocardiogram signals into a classification model in a picture mode to obtain a classification result; the classification model is obtained based on VGG model training.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811075514.8A CN110897628A (en) | 2018-09-14 | 2018-09-14 | Deep neural network-based electrocardiogram signal feature extraction method, device, system, equipment and classification method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811075514.8A CN110897628A (en) | 2018-09-14 | 2018-09-14 | Deep neural network-based electrocardiogram signal feature extraction method, device, system, equipment and classification method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110897628A true CN110897628A (en) | 2020-03-24 |
Family
ID=69812734
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811075514.8A Pending CN110897628A (en) | 2018-09-14 | 2018-09-14 | Deep neural network-based electrocardiogram signal feature extraction method, device, system, equipment and classification method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110897628A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112686091A (en) * | 2020-11-23 | 2021-04-20 | 南京信息工程大学 | Two-step arrhythmia classification method based on deep neural network |
CN113080988A (en) * | 2021-03-26 | 2021-07-09 | 京东方科技集团股份有限公司 | Attention mechanism-based 12-lead electrocardiogram overall classification method and device |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104573458A (en) * | 2014-12-30 | 2015-04-29 | 深圳先进技术研究院 | Identity recognition method, device and system based on electrocardiogram signals |
US20170188971A1 (en) * | 2016-01-06 | 2017-07-06 | Samsung Electronics Co., Ltd. | Electrocardiogram (ecg) authentication method and apparatus |
CN107822622A (en) * | 2017-09-22 | 2018-03-23 | 成都比特律动科技有限责任公司 | Electrocardiographic diagnosis method and system based on depth convolutional neural networks |
CN107951485A (en) * | 2017-11-27 | 2018-04-24 | 乐普(北京)医疗器械股份有限公司 | Ambulatory ECG analysis method and apparatus based on artificial intelligence self study |
US20180144241A1 (en) * | 2016-11-22 | 2018-05-24 | Mitsubishi Electric Research Laboratories, Inc. | Active Learning Method for Training Artificial Neural Networks |
CN108175402A (en) * | 2017-12-26 | 2018-06-19 | 智慧康源(厦门)科技有限公司 | The intelligent identification Method of electrocardiogram (ECG) data based on residual error network |
CN108464827A (en) * | 2018-03-08 | 2018-08-31 | 四川大学 | It is a kind of it is Weakly supervised under electrocardio image-recognizing method |
-
2018
- 2018-09-14 CN CN201811075514.8A patent/CN110897628A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104573458A (en) * | 2014-12-30 | 2015-04-29 | 深圳先进技术研究院 | Identity recognition method, device and system based on electrocardiogram signals |
US20170188971A1 (en) * | 2016-01-06 | 2017-07-06 | Samsung Electronics Co., Ltd. | Electrocardiogram (ecg) authentication method and apparatus |
US20180144241A1 (en) * | 2016-11-22 | 2018-05-24 | Mitsubishi Electric Research Laboratories, Inc. | Active Learning Method for Training Artificial Neural Networks |
CN107822622A (en) * | 2017-09-22 | 2018-03-23 | 成都比特律动科技有限责任公司 | Electrocardiographic diagnosis method and system based on depth convolutional neural networks |
CN107951485A (en) * | 2017-11-27 | 2018-04-24 | 乐普(北京)医疗器械股份有限公司 | Ambulatory ECG analysis method and apparatus based on artificial intelligence self study |
CN108175402A (en) * | 2017-12-26 | 2018-06-19 | 智慧康源(厦门)科技有限公司 | The intelligent identification Method of electrocardiogram (ECG) data based on residual error network |
CN108464827A (en) * | 2018-03-08 | 2018-08-31 | 四川大学 | It is a kind of it is Weakly supervised under electrocardio image-recognizing method |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112686091A (en) * | 2020-11-23 | 2021-04-20 | 南京信息工程大学 | Two-step arrhythmia classification method based on deep neural network |
CN112686091B (en) * | 2020-11-23 | 2024-02-23 | 南京信息工程大学 | Two-step arrhythmia classification method based on deep neural network |
CN113080988A (en) * | 2021-03-26 | 2021-07-09 | 京东方科技集团股份有限公司 | Attention mechanism-based 12-lead electrocardiogram overall classification method and device |
CN113080988B (en) * | 2021-03-26 | 2024-01-16 | 京东方科技集团股份有限公司 | Attention mechanism-based 12-lead electrocardiogram overall classification method and device |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11517212B2 (en) | Electrocardiogram information dynamic monitoring method and dynamic monitoring system | |
Taji et al. | Impact of skin–electrode interface on electrocardiogram measurements using conductive textile electrodes | |
CN106214145B (en) | Electrocardiogram classification method based on deep learning algorithm | |
Yang | Multiscale recurrence quantification analysis of spatial cardiac vectorcardiogram signals | |
AL-Ziarjawey et al. | Heart rate monitoring and PQRST detection based on graphical user interface with Matlab | |
CN114901143A (en) | System and method for reduced-lead electrocardiography diagnosis using deep neural networks and rule-based systems | |
CN110720894B (en) | Atrial flutter detection method, device, equipment and storage medium | |
CN111631705A (en) | Electrocardio abnormality detection method, model training method, device, equipment and medium | |
CN109497986A (en) | Electrocardiogram intelligent analysis method, device, computer equipment and system based on deep neural network | |
CN114732419B (en) | Exercise electrocardiogram data analysis method and device, computer equipment and storage medium | |
CN110897629A (en) | Deep learning algorithm-based electrocardiogram feature extraction method, device, system, equipment and classification method | |
US9549681B2 (en) | Matrix-based patient signal analysis | |
Nandagopal et al. | Newly constructed real time ECG monitoring system using labview | |
Singhal et al. | A systematic review on artificial intelligence-based techniques for diagnosis of cardiovascular arrhythmia diseases: challenges and opportunities | |
Khatun et al. | Detection of myocardial infarction and arrhythmia from single-lead ECG data using bagging trees classifier | |
CN114901145A (en) | System and method for electrocardiographic diagnosis using deep neural networks and rule-based systems | |
CN110897628A (en) | Deep neural network-based electrocardiogram signal feature extraction method, device, system, equipment and classification method | |
Mathur et al. | Analysis of CNN and feed-forward ANN model for the evaluation of ECG signal | |
Vizcaya et al. | Standard ECG lead I prospective estimation study from far-field bipolar leads on the left upper arm: A neural network approach | |
CN110179451B (en) | Electrocardiosignal quality detection method and device, computer equipment and storage medium | |
CN111528833B (en) | Rapid identification and processing method and system for electrocardiosignals | |
US20150265174A1 (en) | Non-Invasive Evaluation of Cardiac Repolarisation Instability for Risk Stratification of Sudden Cardiac Death | |
Hegde et al. | A review on ECG signal processing and HRV analysis | |
Rajoub | Machine learning in biomedical signal processing with ECG applications | |
CN110897626A (en) | Deep neural network-based electrocardiogram analysis method, device, computer equipment and system |
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 |