CN110693488A - Electrocardiosignal processing system, electrocardiosignal processing method, electronic equipment and readable storage medium - Google Patents
Electrocardiosignal processing system, electrocardiosignal processing method, electronic equipment and readable storage medium Download PDFInfo
- Publication number
- CN110693488A CN110693488A CN201911007785.4A CN201911007785A CN110693488A CN 110693488 A CN110693488 A CN 110693488A CN 201911007785 A CN201911007785 A CN 201911007785A CN 110693488 A CN110693488 A CN 110693488A
- Authority
- CN
- China
- Prior art keywords
- heartbeat
- module
- model
- fine tuning
- classification model
- 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
- A61B5/72—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
-
- 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
- A61B5/24—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
- A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Veterinary Medicine (AREA)
- Public Health (AREA)
- Animal Behavior & Ethology (AREA)
- Surgery (AREA)
- Molecular Biology (AREA)
- Medical Informatics (AREA)
- Heart & Thoracic Surgery (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Cardiology (AREA)
- Signal Processing (AREA)
- Psychiatry (AREA)
- Physiology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Fuzzy Systems (AREA)
- Power Engineering (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The application discloses an electrocardiosignal processing system, an electrocardiosignal processing method, electronic equipment and a storage medium. The system comprises: the method comprises the steps of storing original electrocardiosignals through a storage module, reading the original electrocardiosignals through a denoising module, denoising the original electrocardiosignals, segmenting the denoised electrocardiosignals through a segmentation module to obtain a plurality of heartbeats, extracting global features of the denoised electrocardiosignals through a global feature extraction module to obtain the global features of each heartbeat, classifying each heartbeat through a heartbeat classifier according to the heartbeats and the corresponding global features thereof by using a heartbeat classification model to obtain the prediction probability that each heartbeat belongs to each class. Through the electrocardiosignal processing system, electrocardiosignals can be rapidly processed, the prediction probability that each heart beat belongs to each type is obtained, and therefore a doctor can be assisted in classifying the electrocardiosignals, and the classification efficiency is improved.
Description
Technical Field
The present invention relates to the field of information processing technologies, and in particular, to an electrocardiograph signal processing system, an electrocardiograph signal processing method, an electrocardiograph signal processing electronic device, and a readable storage medium.
Background
Arrhythmia typically refers to a too slow, too fast, or irregular heartbeat [8 ]. In clinical diagnosis of arrhythmia, an electrocardiogram is usually used as a diagnostic tool, and heart beats of a patient are classified in a period of time by means of the form of the electrocardiogram, so as to obtain a diagnosis conclusion. An Electrocardiogram (ECG) is a series of time-series electrical signals that reflect the time-varying changes in bioelectric signals measured at different sites in the body.
Classification criteria of different granularities may be specified according to clinical needs. Among them, AAMI (Association for the Advancement of Medical instrumentation) in the United states proposes a relatively coarse-grained classification criterion for an automatic preliminary diagnosis system of electrocardiographs [9 ]. It classifies non-fatal arrhythmic heartbeats into five categories: non-aberrant (N), supraventricular arrhythmias (SVEB or S), ventricular arrhythmias (VEB or V), fused cardiac beats (F), and other classes (Q). Correctly divide the heartbeat of the patient into the above types, so that the doctor can integrally grasp the condition of the patient, and the basis for further accurate diagnosis is provided.
However, the prior art mainly uses manual diagnosis, which results in high cost of manual diagnosis and high requirement on professional level of diagnosticians due to the excessive amount of data of dynamic electrocardiogram.
Disclosure of Invention
In view of the above problems, embodiments of the present invention are proposed to provide a cardiac electrical signal processing system, a method, an electronic device and a readable storage medium that overcome or at least partially solve the above problems.
In a first aspect, an embodiment of the present application provides an electrocardiograph signal processing system, where the system includes: the heartbeat prediction device comprises a storage module, a preprocessor and a heartbeat classifier, wherein the storage module is connected with the preprocessor, the preprocessor is connected with the heartbeat classifier, and the preprocessor comprises a denoising module, a segmentation module and a global feature extraction module;
the storage module is used for storing original electrocardiosignals;
the denoising module is used for reading the original electrocardiosignals from the storage module and denoising the original electrocardiosignals to obtain denoised electrocardiosignals;
the segmentation module is used for carrying out heartbeat segmentation on the denoised electrocardiosignals to obtain a plurality of heartbeats;
the global feature extraction module is used for carrying out global feature extraction on the denoised electrocardiosignals to obtain the global feature of each heartbeat;
and the heartbeat classifier is used for classifying each heartbeat by using a heartbeat classification model according to the plurality of heartbeats and the corresponding global characteristics thereof to obtain the prediction probability of each heartbeat belonging to each class.
Optionally, the heartbeat classification model comprises a convolutional layer, an activation function layer, a pooling layer, a first fully-connected layer, a second fully-connected layer, a third fully-connected layer, and a softmax layer; the heart beat classifier includes:
the characteristic representation module is used for carrying out characteristic extraction on each heartbeat through the convolutional layer, the activation function layer and the pooling layer to obtain a plurality of characteristic representations of each heartbeat;
the feature compression and splicing module is used for compressing the plurality of feature representations of each heartbeat through the first full-connection layer to obtain a compact feature representation of each heartbeat, and splicing the compact feature representation of each heartbeat with the global feature of the corresponding heartbeat to obtain a splicing feature of each heartbeat;
the transformation module is used for transforming the splicing characteristics of each heartbeat through the second full-connection layer to obtain the transformation characteristics of each heartbeat;
an alignment and identification module for aligning and identifying the transformation features through the third fully-connected layer and the softmax layer, resulting in a probability that each heartbeat belongs to each class.
Optionally, the system further comprises a model trainer connected to the heart beat classifier, the model trainer comprising:
the system comprises a sample acquisition and preprocessing module, a data processing module and a data processing module, wherein the sample acquisition and preprocessing module is used for acquiring a plurality of sections of original electrocardiosignal samples and preprocessing the plurality of sections of original electrocardiosignal samples to obtain a plurality of heartbeat samples, and the plurality of sections of original electrocardiosignal samples are two-lead dynamic electrocardiogram samples;
a sample classification module for classifying the plurality of heartbeat samples into a pre-training set and a fine-tuning set;
the pre-training module is used for pre-training a pre-set model by using the pre-training set to obtain a reference classification model;
and the active learning module is used for carrying out active learning and model fine tuning on the reference classification model by using the fine tuning set to obtain the heart beat classification model.
Optionally, the pre-training module comprises:
the parameter adjusting sub-module is used for adjusting parameters of the pre-training set to a preset model in a cross validation mode, and determining the hyper-parameters and the model scale of the preset model;
and the pre-training sub-module is used for pre-training the preset model with the set hyper-parameters and model scale by using the pre-training set to obtain a reference classification model.
Optionally, the active learning module comprises:
the first fine tuning set classification submodule is used for sampling the fine tuning set by adopting an uncertainty sampling-based method to obtain a first active training sample, and dividing the first active training sample into a first fine tuning subset and a first evaluation subset;
the first active learning sub-module is used for carrying out multi-round active learning and model fine tuning on the reference classification model by adopting the first fine tuning subset;
and the first evaluation and determination submodule is used for performing performance evaluation on the reference classification model subjected to each round of active learning and model fine tuning by adopting the first evaluation subset until the error value does not decrease within the preset number of rounds, stopping the active learning and model fine tuning and determining the reference classification model corresponding to the round with the minimum error value as the heart beat classification model.
Optionally, the active learning module comprises:
a second fine tuning set classification sub-module for classifying the fine tuning set into a fine tuning subset and an evaluation subset;
the second active learning sub-module is used for performing multiple rounds of active learning and model fine tuning on the reference classification model by adopting a committee-based method on the fine tuning subset, a Dropout layer is introduced into both the first full-connection layer and the second full-connection layer of the preset model, and a committee is generated by adopting the Dropout layer;
and the second evaluation and determination submodule is used for evaluating the performance of the reference classification model subjected to each round of active learning and model fine tuning by adopting the evaluation subset until the error value does not decrease within the preset number of rounds, stopping the active learning and model fine tuning and determining the reference classification model corresponding to the round with the minimum error value as the heartbeat classification model.
Optionally, before the active learning module, the method further comprises:
and the random sampling training module is used for training and evaluating the reference classification model by adopting a random sampling method for the fine adjustment collection until the classification accuracy of the reference classification model is greater than a preset value.
In a second aspect, an embodiment of the present application further provides an electrocardiograph signal processing method, where the method includes:
storing original electrocardiosignals;
reading the original electrocardiosignals and carrying out denoising processing on the original electrocardiosignals to obtain denoised electrocardiosignals;
heart beat segmentation is carried out on the denoised electrocardiosignal to obtain a plurality of heart beats;
carrying out global feature extraction on the denoised electrocardiosignals to obtain the global feature of each heart beat;
and classifying each heartbeat by using a heartbeat classification model according to the plurality of heartbeats and the corresponding global characteristics thereof to obtain the prediction probability of each heartbeat belonging to each class.
Optionally, the heartbeat classification model comprises a convolutional layer, an activation function layer, a pooling layer, a first fully-connected layer, a second fully-connected layer, a third fully-connected layer, and a softmax layer; according to the plurality of heart beats and the corresponding global features thereof, classifying each heart beat by using a heart beat classification model respectively to obtain the prediction probability of each heart beat belonging to each class, comprising the following steps:
performing feature extraction on each heartbeat through a convolutional layer, an activation function layer and a pooling layer to obtain a plurality of feature representations of each heartbeat;
compressing the plurality of feature representations of each heartbeat through the first full-connection layer to obtain a compact feature representation of each heartbeat, and splicing the compact feature representation of each heartbeat with the global feature of the corresponding heartbeat to obtain a splicing feature of each heartbeat;
transforming the splicing characteristic of each heart beat through the second full-connection layer to obtain a transformation characteristic of each heart beat;
aligning and identifying the transformation features through the third fully-connected layer and softmax layer, resulting in a probability that the each heartbeat belongs to each class.
Optionally, the method further comprises:
obtaining a plurality of sections of original electrocardiosignal samples and preprocessing the plurality of sections of original electrocardiosignal samples to obtain a plurality of heartbeat samples, wherein the plurality of sections of original electrocardiosignal samples are two-lead dynamic electrocardiogram samples;
dividing the plurality of heartbeat samples into a pre-training set and a fine-tuning set;
pre-training a preset model by using the pre-training set to obtain a reference classification model;
and performing active learning and model fine tuning on the reference classification model by using the fine tuning set to obtain the heart beat classification model.
Optionally, the pre-training set is used to pre-train a preset model, so as to obtain a reference classification model, including:
performing parameter adjustment on a preset model by adopting a cross validation mode in the pre-training set, and determining the hyper-parameters and the model scale of the preset model;
and pre-training the preset model with the set hyper-parameters and the set model scale by using the pre-training set to obtain a reference classification model.
Optionally, performing active learning and model fine tuning on the reference classification model by using the fine tuning set to obtain the heart beat classification model, including:
sampling the fine tuning set by adopting a method based on uncertainty sampling to obtain a first active training sample, and dividing the first active training sample into a first fine tuning subset and a first evaluation subset;
performing multiple rounds of active learning and model fine tuning on the reference classification model by adopting the first fine tuning subset;
and performing performance evaluation on the reference classification model subjected to each round of active learning and model fine tuning by adopting the first evaluation subset until the error value does not decrease within the preset number of rounds, stopping the active learning and model fine tuning, and determining the reference classification model corresponding to the round with the minimum error value as the heartbeat classification model.
Optionally, performing active learning and model fine tuning on the reference classification model by using the fine tuning set to obtain the heart beat classification model, including:
sampling the fine tuning set by adopting a committee-based method to obtain a second active training sample, and dividing the second active training sample into a fine tuning subset and an evaluation subset; a Dropout layer is introduced into the first full-connection layer and the second full-connection layer of the preset model, and a committee is generated by adopting the Dropout layer;
performing multiple rounds of active learning and model fine tuning on the reference classification model by using the second fine tuning subset;
and performing performance evaluation on the reference classification model after each round of active learning and model fine tuning by adopting the evaluation subset until the error value is not reduced within the preset round number, stopping the active learning and model fine tuning, and determining the reference classification model corresponding to the round with the minimum error value as the heartbeat classification model.
Optionally, before performing active learning and model fine-tuning on the reference classification model by using the fine-tuning set to obtain the heart beat classification model, the method further includes:
and training and evaluating the reference classification model by adopting a random sampling method in the fine adjustment set until the classification accuracy of the reference classification model is greater than a preset value.
In a third aspect, an embodiment of the present application further provides an electronic device, including: a memory, a processor and a computer program stored on the memory and executable on the processor, the computer program, when executed by the processor, implementing the steps of the method for processing cardiac electrical signals according to the second aspect.
In a fourth aspect, the present application further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the electrocardiograph signal processing method according to the second aspect are implemented.
The embodiment of the invention has the following advantages:
in the embodiment of the invention, the original electrocardiosignals are stored through the storage module, the original electrocardiosignals are read through the denoising module and are denoised, the denoised electrocardiosignals are subjected to heartbeat segmentation through the segmentation module to obtain a plurality of heartbeats, the denoised electrocardiosignals are subjected to global feature extraction through the global feature extraction module to obtain the global feature of each heartbeat, and finally, the heartbeat classification model is used for classifying each heartbeat according to the plurality of heartbeats and the corresponding global feature of the heartbeat to obtain the prediction probability that each heartbeat belongs to each class. Through the electrocardiosignal processing system, electrocardiosignals can be rapidly processed, the prediction probability that each heart beat belongs to each type is obtained, and therefore a doctor can be assisted in classifying the electrocardiosignals, and the classification efficiency is improved.
Drawings
FIG. 1 is a block diagram of an ECG signal processing system according to the present invention;
FIG. 2 is a network architecture diagram of a heart beat classification model of the present invention;
FIG. 3 is a block diagram of a heartbeat classifier of the present invention;
FIG. 4 is a block diagram of another alternative ECG signal processing system of the present invention;
FIG. 5 is a flow chart illustrating the steps of a method for processing an ECG signal according to the present invention.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Referring to fig. 1, fig. 1 is a block diagram illustrating a structural diagram of an electrocardiograph signal processing system according to an embodiment of the present application, and as shown in fig. 1, the system includes the following structures: a storage module 101, a preprocessor 102 and a heart beat classifier 103, wherein the storage module 101 is connected with the preprocessor 102, the preprocessor 102 is connected with the heart beat classifier 103, and the preprocessor 102 comprises a denoising module 1021, a segmentation module 1022 and a global feature extraction module 1023;
the storage module 101 is configured to store an original electrocardiographic signal.
In this embodiment, the storage module is configured to store an original electrocardiographic signal, wherein the original electrocardiographic signal is detected by a medical worker and then input to the storage module for storage.
The first storage module 101 is configured to store electronic medical record information, where the electronic medical record information includes objective indicators, current medical history, and physical examination results.
The denoising module 1021 is configured to read the original electrocardiographic signal from the memory and perform denoising processing on the original electrocardiographic signal to obtain a denoised electrocardiographic signal.
In this embodiment, the acquired original ecg signal mainly includes two types of relatively serious noises: baseline drift, power frequency interference. The baseline drift refers to the change of the baseline of the electrocardiosignal caused by factors such as human respiration and the like, and can be considered that a section of low-frequency noise is superimposed on the original signal. The power frequency interference means that high-frequency noise is inevitably superposed on an original signal due to a cable and the like. In specific implementation, a median filter can be adopted to remove baseline drift, and the median filter can be used to filter out each characteristic wave in the original signal to obtain the low-frequency baseline component of the original signal. Finally, the low-frequency baseline component obtained by subtracting the original signal is used to normalize the baseline of the original signal and remove the baseline drift. In actual operation, a median filter with the width of 200ms is used for removing P waves and QRS complexes, and then a median filter with the width of 600ms is used for removing T waves, so that the denoised electrocardiosignals can be obtained. Denoising the electrocardiosignal, so that the subsequent prediction result is more accurate.
The segmenting module 1022 is configured to perform heartbeat segmentation on the denoised electrocardiographic signal to obtain a plurality of heartbeats.
In the present embodiment, since the unit for prediction classification is a single heartbeat and a single-segment electrocardiographic signal includes a plurality of heartbeats, it is necessary to divide the denoised electrocardiographic signal into a plurality of heartbeats by dividing the heartbeat by the dividing module. In specific implementation, the values of each R peak of the denoised electrocardiosignals are extracted, then 0.278s of data is taken to the left and 0.556s of data is taken to the right (in the MIT-BIH arrhythmia database, the sampling frequency of the electrocardiosignals is 360Hz, that is, 100 points are taken to the left of the R peak and 299 points are taken to the right, and the obtained signal length is 300), and the denoised electrocardiosignals are divided into a plurality of heartbeats.
And the global feature extraction module 1023 is used for performing global feature extraction on the denoised electrocardiosignal to obtain the global feature of each heartbeat.
In the embodiment, one heart beat is classified, and the global features of some heart beats can be extracted to improve the accuracy of classification. In specific implementation, the following four global features are extracted for each heart beat: (1) forward RR interval: distance of the heart beat R peak from the previous heart beat R peak; (2) a backward RR interval: the distance of this heart beat R peak from the next heart beat R peak; (3) local RR interval mean: mean value of RR intervals of the first 10 heartbeats; (4) global RR interval mean: mean value of RR intervals of 300 beats prior to the heart beat.
The heartbeat classifier 103 is configured to classify each heartbeat by using a heartbeat classification model according to the multiple heartbeats and the corresponding global features thereof, so as to obtain a prediction probability that each heartbeat belongs to each class.
In the embodiment, the heartbeat classifier adopts a trained heartbeat classification model, classifies each heartbeat according to a plurality of heartbeats and corresponding global characteristics thereof, can quickly obtain the prediction probability that each heartbeat belongs to each class, and can be used as a reference to assist a doctor in classifying heartbeats, so that the heartbeat classification rate is improved, and the labor cost is reduced.
In the embodiment of the invention, the original electrocardiosignals are stored through the storage module, the original electrocardiosignals are read through the denoising module and are denoised, the denoised electrocardiosignals are subjected to heartbeat segmentation through the segmentation module to obtain a plurality of heartbeats, the denoised electrocardiosignals are subjected to global feature extraction through the global feature extraction module to obtain the global feature of each heartbeat, and finally, the heartbeat classification model is used for classifying each heartbeat according to the plurality of heartbeats and the corresponding global feature of the heartbeat to obtain the prediction probability that each heartbeat belongs to each class. Through the electrocardiosignal processing system, electrocardiosignals can be rapidly processed, the prediction probability that each heart beat belongs to each type is obtained, and therefore a doctor can be assisted in classifying the electrocardiosignals, and the classification efficiency is improved.
Referring to fig. 2 and fig. 3, fig. 2 shows a network structure diagram of a heart beat classification model according to an embodiment of the present application, and as shown in fig. 2, the heart beat classification model includes a convolutional layer, an activation function layer, a pooling layer, a first fully-connected layer, a second fully-connected layer, a third fully-connected layer, and a softmax layer. Fig. 3 shows a block diagram of a structure of a heart beat classifier according to an embodiment of the present application, and as shown in fig. 3, the heart beat classifier includes:
a feature representation module 301, configured to perform feature extraction on each heartbeat through the convolutional layer, the activation function layer, and the pooling layer, to obtain a plurality of feature representations of each heartbeat.
In the present embodiment, the original electrocardiographic signal used includes two leads, each heartbeat obtained through preprocessing also includes two leads, which may be an MLII lead and a V5 lead, and when feature extraction is performed by the feature representation module, the convolutional layer, the activation function layer, and the pooling layer are divided into two sub-networks, feature extraction is performed on each heartbeat, and there is no fusion of information, so that a plurality of feature representations of each heartbeat are extracted.
A feature compression and concatenation module 302, configured to compress the plurality of feature representations of each heartbeat through the first full connection layer to obtain a compact feature representation of each heartbeat, and concatenate the compact feature representation of each heartbeat with the global feature of the corresponding heartbeat to obtain a concatenated feature of each heartbeat.
In this embodiment, when processing the plurality of feature representations of each heartbeat, the feature compression and concatenation module compresses the plurality of feature representations of each heartbeat through the first full link layer to obtain a compact feature representation of each heartbeat, and then concatenates the compact feature representation of each heartbeat with the global feature of the corresponding heartbeat to obtain a concatenation feature of each heartbeat. The global feature of each heart beat is spliced with the compact feature representation at the position, so that the subsequent classification result can be more accurate.
A transforming module 303, configured to transform the splicing feature of each heartbeat through the second full-connection layer, so as to obtain a transformed feature of each heartbeat.
An alignment and recognition module 304 for aligning and recognizing the transformed features through the third fully-connected layer and softmax layer, resulting in a probability that the heart beat belongs to each class.
In the embodiment, each heartbeat and the corresponding global features are processed through the heartbeat classification model in the heartbeat classifier, so that the probability that each heartbeat belongs to each class can be quickly obtained, a doctor can be assisted in classifying the electrocardiosignals, the classification efficiency is improved, and the labor cost is saved.
Referring to fig. 4, fig. 4 is a block diagram of another electrocardiograph signal processing system according to an embodiment of the present application, and as shown in fig. 4, the system further includes a model trainer connected to the heart beat classifier, where the model trainer includes:
the sample acquiring and preprocessing module 401 is configured to acquire a plurality of segments of original electrocardiographic signal samples and preprocess the plurality of segments of original electrocardiographic signal samples to obtain a plurality of heartbeat samples, where the plurality of segments of original electrocardiographic signal samples are two-lead electrocardiogram samples.
In this embodiment, a plurality of segments of original electrocardiographic signal samples need to be obtained and preprocessed, so as to facilitate subsequent training of a preset model, where the preprocessing mode is the same as the processing mode of the denoising module 1021, the segmentation module 1022, and the global feature extraction module 1023 in the preprocessor, and reference may be made to the above specific explanation, which is not described herein again, and each heartbeat sample in the obtained plurality of heartbeat samples includes a heartbeat and a corresponding global feature thereof.
A sample classification module 402 for classifying the plurality of heartbeat samples into a pre-training set and a micro-tone set;
and a pre-training module 403, configured to perform pre-training on a preset model by using the pre-training set, so as to obtain a reference classification model.
In order to obtain a better model training effect, a part of data needs to be used for pre-training a preset model, in the embodiment, a plurality of heartbeat samples are divided into a pre-training set and a micro-adjustment set, and the pre-training set is used for pre-training the preset model.
In one possible embodiment, the pre-training module comprises:
the parameter adjusting sub-module is used for adjusting parameters of the pre-training set to a preset model in a cross validation mode, and determining the hyper-parameters and the model scale of the preset model;
and the pre-training sub-module is used for pre-training the preset model with the set hyper-parameters and model scale by using the pre-training set to obtain a reference classification model.
In this embodiment, in order to fully utilize the data, the hyper-parameters and model scale of the preset model may be determined by using an n-fold cross validation method, the data is averagely divided into n parts, 1 part of the data is taken as a validation set each time, training of the model is performed by using another (n-1) parts of the data, one evaluation index may be obtained, and the operation is repeated n times, and the average value of the n times is taken as a final evaluation index. By doing so, the data can be relatively fully utilized, and when Dropout is used, the value can be taken as 0.5, the parameter scale of training is increased, and finally, through cross validation, the size of the feature vector obtained after the first full connection layer is selected as 50, and the batch size is selected as 10. On the basis, all data are used again, and pre-training is carried out according to the hyper-parameters and the model scale setting to obtain a standard classification model.
And an active learning module 404, configured to perform active learning and model fine tuning on the reference classification model by using the fine tuning set, so as to obtain the heart beat classification model.
In one possible embodiment, the active learning module includes:
the first fine tuning set classification submodule is used for sampling the fine tuning set by adopting an uncertainty sampling-based method to obtain a first active training sample, and dividing the first active training sample into a first fine tuning subset and a first evaluation subset;
the first active learning sub-module is used for carrying out multi-round active learning and model fine tuning on the reference classification model by adopting the first fine tuning subset;
and the first evaluation and determination submodule is used for performing performance evaluation on the reference classification model subjected to each round of active learning and model fine tuning by adopting the first evaluation subset until the error value does not decrease within the preset number of rounds, stopping the active learning and model fine tuning and determining the reference classification model corresponding to the round with the minimum error value as the heart beat classification model.
In this embodiment, the fine tuning set used for active learning may be sampled by an uncertainty sampling-based method to obtain a first active training sample, specifically, the fine tuning set may be subjected to uncertainty sampling, the fine tuning set may be classified by using a reference classification model, and an edge confidence determination method may be used, in which the probability p1, p2(p1 is not less than p2) corresponding to the 2 categories with the highest confidence level among the categories is considered, the lower the value of p1-p2, the larger the sample information amount is, 1- (p1-p2) may be recorded as the information score of the sample, so that the first active training sample with the larger information score may be selected from the fine tuning set, in specific implementation, 50% of the information score may be selected from the fine tuning set as the first active training sample, then the first active training sample may be labeled by manual classification and divided into a first fine tuning subset and a first evaluation subset, and performing multi-round training and evaluation on the reference classification model, wherein in specific implementation, the active learning and the model fine tuning are stopped until the error value calculated by evaluation does not decrease within 10 preset rounds, and the reference classification model corresponding to the round with the minimum error value is determined as the heart beat classification model.
In the embodiment, the fine tuning set is sampled by adopting a method based on uncertainty sampling, a sample with large information content is selected to obtain a first active training sample, and active learning and model fine tuning are carried out on the reference classification model, so that a good active learning effect can be achieved.
In another possible implementation, the active learning module includes:
the second fine tuning set classification submodule is used for sampling the fine tuning set by adopting a committee-based method to obtain a second active training sample, and dividing the second active training sample into a fine tuning subset and an evaluation subset; a Dropout layer is introduced into the first full-connection layer and the second full-connection layer of the preset model, and a committee is generated by adopting the Dropout layer;
the second active learning submodule is used for carrying out multi-round active learning and model fine tuning on the reference classification model by adopting the second fine tuning subset;
and the second evaluation and determination submodule is used for evaluating the performance of the reference classification model subjected to each round of active learning and model fine tuning by adopting the evaluation subset until the error value does not decrease within the preset number of rounds, stopping the active learning and model fine tuning and determining the reference classification model corresponding to the round with the minimum error value as the heartbeat classification model.
In the present embodiment, the fine tuning set for active learning may be sampled by a committee-based method to obtain a second active training sample, specifically, as described above, the heartbeat classification model includes a convolutional layer, an activation function layer, a pooling layer, a first fully-connected layer, a second fully-connected layer, a third fully-connected layer, and a softmax layer, and the pre-set model and the pre-trained reference classification model also include a convolutional layer, an activation function layer, a pooling layer, a first fully-connected layer, a second fully-connected layer, a third fully-connected layer, and a softmax layer, and a Dropout layer is introduced into both the first fully-connected layer and the second fully-connected layer of the pre-set model, so that when the fine tuning set is classified by using the reference classification model, the committee is generated by the Dropout layer, and a plurality of different probabilities are obtained for each class of the same sample, the difference of classification results is larger, and the information amount of the sample is larger, and calculating the information score of the sample, so that a second active training sample with a larger information score is selected from the fine adjustment set, in specific implementation, the first 50% of the information score can be selected from the fine adjustment set as the second active training sample, then the second active training sample is labeled by manual classification and is divided into a second fine adjustment subset and a second evaluation subset, and the reference classification model is subjected to multi-round training and evaluation, in specific implementation, the active learning and the model fine adjustment are stopped until the error value calculated by evaluation does not decrease within 10 preset rounds, and the reference classification model corresponding to the round with the smallest error value is determined as the heart beat classification model.
In the embodiment, a committee-based method is adopted to sample the fine tuning set, a sample with a large information amount is selected, a second active training sample is obtained, active learning and model fine tuning are performed on the reference classification model, and a good active learning effect can be achieved.
In a possible implementation, before the active learning module, the method further includes:
and the random sampling training module is used for training and evaluating the reference classification model by adopting a random sampling method for the fine adjustment collection until the classification accuracy of the reference classification model is greater than a preset value.
In this embodiment, the reference classification model is trained and evaluated by a random sampling method to have a certain classification accuracy, for example, the classification accuracy of the reference classification model reaches 50%, and then the reference classification model is actively learned and model fine-tuned, so that a better active learning effect can be obtained, and the classification of the obtained heart beat classification model is more accurate.
Based on the same inventive concept, an embodiment of the present application provides an electrocardiographic signal processing method, referring to fig. 5, fig. 5 is a flowchart illustrating steps of the electrocardiographic signal processing method according to the embodiment of the present application, and as shown in fig. 5, the method includes:
step S501: storing original electrocardiosignals;
step S502: reading the original electrocardiosignals and carrying out denoising processing on the original electrocardiosignals to obtain denoised electrocardiosignals;
step S503: heart beat segmentation is carried out on the denoised electrocardiosignal to obtain a plurality of heart beats;
step S504: carrying out global feature extraction on the denoised electrocardiosignals to obtain the global feature of each heart beat;
step S505: and classifying each heartbeat by using a heartbeat classification model according to the plurality of heartbeats and the corresponding global characteristics thereof to obtain the prediction probability of each heartbeat belonging to each class.
Optionally, the heartbeat classification model comprises a convolutional layer, an activation function layer, a pooling layer, a first fully-connected layer, a second fully-connected layer, a third fully-connected layer, and a softmax layer; according to the plurality of heart beats and the corresponding global features thereof, classifying each heart beat by using a heart beat classification model respectively to obtain the prediction probability of each heart beat belonging to each class, comprising the following steps:
performing feature extraction on each heartbeat through a convolutional layer, an activation function layer and a pooling layer to obtain a plurality of feature representations of each heartbeat;
compressing the plurality of feature representations of each heartbeat through the first full-connection layer to obtain a compact feature representation of each heartbeat, and splicing the compact feature representation of each heartbeat with the global feature of the corresponding heartbeat to obtain a splicing feature of each heartbeat;
transforming the splicing characteristic of each heart beat through the second full-connection layer to obtain a transformation characteristic of each heart beat;
aligning and identifying the transformation features through the third fully-connected layer and softmax layer, resulting in a probability that the each heartbeat belongs to each class.
Optionally, the method further comprises:
obtaining a plurality of sections of original electrocardiosignal samples and preprocessing the plurality of sections of original electrocardiosignal samples to obtain a plurality of heartbeat samples, wherein the plurality of sections of original electrocardiosignal samples are two-lead dynamic electrocardiogram samples;
dividing the plurality of heartbeat samples into a pre-training set and a fine-tuning set;
pre-training a preset model by using the pre-training set to obtain a reference classification model;
and performing active learning and model fine tuning on the reference classification model by using the fine tuning set to obtain the heart beat classification model.
Optionally, the pre-training set is used to pre-train a preset model, so as to obtain a reference classification model, including:
performing parameter adjustment on a preset model by adopting a cross validation mode in the pre-training set, and determining the hyper-parameters and the model scale of the preset model;
and pre-training the preset model with the set hyper-parameters and the set model scale by using the pre-training set to obtain a reference classification model.
Optionally, performing active learning and model fine tuning on the reference classification model by using the fine tuning set to obtain the heart beat classification model, including:
sampling the fine tuning set by adopting a method based on uncertainty sampling to obtain a first active training sample, and dividing the first active training sample into a first fine tuning subset and a first evaluation subset;
performing multiple rounds of active learning and model fine tuning on the reference classification model by adopting the first fine tuning subset;
and performing performance evaluation on the reference classification model subjected to each round of active learning and model fine tuning by adopting the first evaluation subset until the error value does not decrease within the preset number of rounds, stopping the active learning and model fine tuning, and determining the reference classification model corresponding to the round with the minimum error value as the heartbeat classification model.
Optionally, performing active learning and model fine tuning on the reference classification model by using the fine tuning set to obtain the heart beat classification model, including:
sampling the fine tuning set by adopting a committee-based method to obtain a second active training sample, and dividing the second active training sample into a fine tuning subset and an evaluation subset; a Dropout layer is introduced into the first full-connection layer and the second full-connection layer of the preset model, and a committee is generated by adopting the Dropout layer;
performing multiple rounds of active learning and model fine tuning on the reference classification model by using the second fine tuning subset;
and performing performance evaluation on the reference classification model after each round of active learning and model fine tuning by adopting the evaluation subset until the error value is not reduced within the preset round number, stopping the active learning and model fine tuning, and determining the reference classification model corresponding to the round with the minimum error value as the heartbeat classification model.
Optionally, before performing active learning and model fine-tuning on the reference classification model by using the fine-tuning set to obtain the heart beat classification model, the method further includes:
and training and evaluating the reference classification model by adopting a random sampling method in the fine adjustment set until the classification accuracy of the reference classification model is greater than a preset value.
Based on the same inventive concept, another embodiment of the present application provides an electronic device, including: a memory, a processor and a computer program stored on the memory and executable on the processor, the computer program, when executed by the processor, implementing the steps of the method according to any of the above embodiments.
Based on the same inventive concept, another embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements the steps in the method according to any of the above-mentioned embodiments of the present application.
As for the method embodiment, since it is basically similar to the system embodiment, the description is simple, and the relevant points can be referred to the partial description of the system embodiment.
The embodiments in the present specification are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, apparatus, or computer program product. Accordingly, embodiments of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
Embodiments of the present invention are described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing terminal to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing terminal to cause a series of operational steps to be performed on the computer or other programmable terminal to produce a computer implemented process such that the instructions which execute on the computer or other programmable terminal provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications of these embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the embodiments of the invention.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or terminal that comprises the element.
The above detailed descriptions of an ecg signal processing system, an ecg signal processing method, an electronic device, and a computer-readable storage medium provided by the present invention have been provided, and specific examples are used herein to explain the principles and embodiments of the present invention, and the descriptions of the above embodiments are only used to help understand the method and the core ideas of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (10)
1. A cardiac signal processing system, the system comprising: the heartbeat prediction device comprises a storage module, a preprocessor and a heartbeat classifier, wherein the storage module is connected with the preprocessor, the preprocessor is connected with the heartbeat classifier, and the preprocessor comprises a denoising module, a segmentation module and a global feature extraction module;
the storage module is used for storing original electrocardiosignals;
the denoising module is used for reading the original electrocardiosignals from the storage module and denoising the original electrocardiosignals to obtain denoised electrocardiosignals;
the segmentation module is used for carrying out heartbeat segmentation on the denoised electrocardiosignals to obtain a plurality of heartbeats;
the global feature extraction module is used for carrying out global feature extraction on the denoised electrocardiosignals to obtain the global feature of each heartbeat;
and the heartbeat classifier is used for classifying each heartbeat by using a heartbeat classification model according to the plurality of heartbeats and the corresponding global characteristics thereof to obtain the prediction probability of each heartbeat belonging to each class.
2. The system of claim 1, wherein the heartbeat classification model includes a convolutional layer, an activation function layer, a pooling layer, a first fully-connected layer, a second fully-connected layer, a third fully-connected layer, and a softmax layer; the heart beat classifier includes:
the characteristic representation module is used for carrying out characteristic extraction on each heartbeat through the convolutional layer, the activation function layer and the pooling layer to obtain a plurality of characteristic representations of each heartbeat;
the feature compression and splicing module is used for compressing the plurality of feature representations of each heartbeat through the first full-connection layer to obtain a compact feature representation of each heartbeat, and splicing the compact feature representation of each heartbeat with the global feature of the corresponding heartbeat to obtain a splicing feature of each heartbeat;
the transformation module is used for transforming the splicing characteristics of each heartbeat through the second full-connection layer to obtain the transformation characteristics of each heartbeat;
an alignment and identification module for aligning and identifying the transformation features through the third fully-connected layer and the softmax layer, resulting in a probability that each heartbeat belongs to each class.
3. The system of claim 2, further comprising a model trainer connected to the heart beat classifier, the model trainer comprising:
the system comprises a sample acquisition and preprocessing module, a data processing module and a data processing module, wherein the sample acquisition and preprocessing module is used for acquiring a plurality of sections of original electrocardiosignal samples and preprocessing the plurality of sections of original electrocardiosignal samples to obtain a plurality of heartbeat samples, and the plurality of sections of original electrocardiosignal samples are two-lead dynamic electrocardiogram samples;
a sample classification module for classifying the plurality of heartbeat samples into a pre-training set and a fine-tuning set;
the pre-training module is used for pre-training a pre-set model by using the pre-training set to obtain a reference classification model;
and the active learning module is used for carrying out active learning and model fine tuning on the reference classification model by using the fine tuning set to obtain the heart beat classification model.
4. The system of claim 3, wherein the pre-training module comprises:
the parameter adjusting sub-module is used for adjusting parameters of the pre-training set to a preset model in a cross validation mode, and determining the hyper-parameters and the model scale of the preset model;
and the pre-training sub-module is used for pre-training the preset model with the set hyper-parameters and model scale by using the pre-training set to obtain a reference classification model.
5. The system of claim 3, wherein the active learning module comprises:
the first fine tuning set classification submodule is used for sampling the fine tuning set by adopting an uncertainty sampling-based method to obtain a first active training sample, and dividing the first active training sample into a first fine tuning subset and a first evaluation subset;
the first active learning sub-module is used for carrying out multi-round active learning and model fine tuning on the reference classification model by adopting the first fine tuning subset;
and the first evaluation and determination submodule is used for performing performance evaluation on the reference classification model subjected to each round of active learning and model fine tuning by adopting the first evaluation subset until the error value does not decrease within the preset number of rounds, stopping the active learning and model fine tuning and determining the reference classification model corresponding to the round with the minimum error value as the heart beat classification model.
6. The system of claim 3, wherein the active learning module comprises:
the second fine tuning set classification submodule is used for sampling the fine tuning set by adopting a committee-based method to obtain a second active training sample, and dividing the second active training sample into a fine tuning subset and an evaluation subset; a Dropout layer is introduced into the first full-connection layer and the second full-connection layer of the preset model, and a committee is generated by adopting the Dropout layer;
the second active learning submodule is used for carrying out multi-round active learning and model fine tuning on the reference classification model by adopting the second fine tuning subset;
and the second evaluation and determination submodule is used for evaluating the performance of the reference classification model subjected to each round of active learning and model fine tuning by adopting the evaluation subset until the error value does not decrease within the preset number of rounds, stopping the active learning and model fine tuning and determining the reference classification model corresponding to the round with the minimum error value as the heartbeat classification model.
7. The system of claim 3, further comprising, prior to the active learning module:
and the random sampling training module is used for training and evaluating the reference classification model by adopting a random sampling method for the fine adjustment collection until the classification accuracy of the reference classification model is greater than a preset value.
8. A method of processing an electrical cardiac signal, the method comprising:
storing original electrocardiosignals;
reading the original electrocardiosignals and carrying out denoising processing on the original electrocardiosignals to obtain denoised electrocardiosignals;
heart beat segmentation is carried out on the denoised electrocardiosignal to obtain a plurality of heart beats;
carrying out global feature extraction on the denoised electrocardiosignals to obtain the global feature of each heart beat;
and classifying each heartbeat by using a heartbeat classification model according to the plurality of heartbeats and the corresponding global characteristics thereof to obtain the prediction probability of each heartbeat belonging to each class.
9. An electronic device, comprising: memory, a processor and a computer program stored on the memory and executable on the processor, the computer program, when executed by the processor, implementing the steps of the method for cardiac electrical signal processing as claimed in claim 8.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a computer program which, when being executed by a processor, carries out the steps of carrying out the method for cardiac electrical signal processing as claimed in claim 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911007785.4A CN110693488A (en) | 2019-10-22 | 2019-10-22 | Electrocardiosignal processing system, electrocardiosignal processing method, electronic equipment and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911007785.4A CN110693488A (en) | 2019-10-22 | 2019-10-22 | Electrocardiosignal processing system, electrocardiosignal processing method, electronic equipment and readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110693488A true CN110693488A (en) | 2020-01-17 |
Family
ID=69200971
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911007785.4A Pending CN110693488A (en) | 2019-10-22 | 2019-10-22 | Electrocardiosignal processing system, electrocardiosignal processing method, electronic equipment and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110693488A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112022142A (en) * | 2020-08-07 | 2020-12-04 | 上海联影智能医疗科技有限公司 | Electrocardiosignal type identification method, device and medium |
| CN113712525A (en) * | 2020-05-21 | 2021-11-30 | 深圳市理邦精密仪器股份有限公司 | Physiological parameter processing method and device and medical equipment |
| CN113712564A (en) * | 2020-05-12 | 2021-11-30 | 深圳市科瑞康实业有限公司 | Electrocardiosignal classification device and method |
| CN113974649A (en) * | 2021-12-03 | 2022-01-28 | 上海交通大学医学院附属瑞金医院 | Method, apparatus and medium for classification of heart beat signals and training of deep learning models for classification of heart beat signals |
| CN116509414A (en) * | 2023-04-14 | 2023-08-01 | 中国科学院大学 | Electrocardiosignal denoising classification system and method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109223002A (en) * | 2018-08-27 | 2019-01-18 | 上海铱硙医疗科技有限公司 | Self-closing disease illness prediction technique, device, equipment and storage medium |
| CN109805924A (en) * | 2019-02-15 | 2019-05-28 | 济南大学 | ECG's data compression method and cardiac arrhythmia detection system based on CNN |
-
2019
- 2019-10-22 CN CN201911007785.4A patent/CN110693488A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109223002A (en) * | 2018-08-27 | 2019-01-18 | 上海铱硙医疗科技有限公司 | Self-closing disease illness prediction technique, device, equipment and storage medium |
| CN109805924A (en) * | 2019-02-15 | 2019-05-28 | 济南大学 | ECG's data compression method and cardiac arrhythmia detection system based on CNN |
Non-Patent Citations (4)
| Title |
|---|
| BINHANG YUAN ET AL: "Diagnosing Cardiac Abnormalities from 12-Lead Electrocardiograms Using Enhanced Deep Convolutional Neural Networks", 《MACHINE LEARNING AND MEDICAL ENGINEERING FOR CARDIOVASCULAR HEALTH AND INTRAVASCULAR IMAGING AND COMPUTER ASSISTED STENTING》 * |
| MELANIE DUCOFFE ET AL: "Active learning strategy for CNN combining", 《ESANN 2017 PROCEEDINGS, EUROPEAN SYMPOSIUM ON ARTIFICIAL NEURAL NETWORKS, COMPUTATIONAL INTELLIGENCE AND MACHINE LEARNING》 * |
| 王珍钰: "基于不确定性的主动学习算法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 邓力等: "基于心拍的端到端心律失常分类", 《南方医科大学学报》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113712564A (en) * | 2020-05-12 | 2021-11-30 | 深圳市科瑞康实业有限公司 | Electrocardiosignal classification device and method |
| CN113712564B (en) * | 2020-05-12 | 2023-09-01 | 深圳市科瑞康实业有限公司 | Electrocardiogram signal classification equipment and method |
| CN113712525A (en) * | 2020-05-21 | 2021-11-30 | 深圳市理邦精密仪器股份有限公司 | Physiological parameter processing method and device and medical equipment |
| CN113712525B (en) * | 2020-05-21 | 2024-07-12 | 深圳市理邦精密仪器股份有限公司 | Physiological parameter processing method and device and medical equipment |
| CN112022142A (en) * | 2020-08-07 | 2020-12-04 | 上海联影智能医疗科技有限公司 | Electrocardiosignal type identification method, device and medium |
| CN112022142B (en) * | 2020-08-07 | 2023-10-17 | 上海联影智能医疗科技有限公司 | Electrocardiosignal type identification method, device and medium |
| CN113974649A (en) * | 2021-12-03 | 2022-01-28 | 上海交通大学医学院附属瑞金医院 | Method, apparatus and medium for classification of heart beat signals and training of deep learning models for classification of heart beat signals |
| CN116509414A (en) * | 2023-04-14 | 2023-08-01 | 中国科学院大学 | Electrocardiosignal denoising classification system and method |
| CN116509414B (en) * | 2023-04-14 | 2023-12-15 | 中国科学院大学 | Electrocardiosignal denoising classification system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110693488A (en) | Electrocardiosignal processing system, electrocardiosignal processing method, electronic equipment and readable storage medium | |
| CN109171712B (en) | Atrial fibrillation identification method, atrial fibrillation identification device, atrial fibrillation identification equipment and computer readable storage medium | |
| JP6367442B2 (en) | Method and system for disease analysis based on conversion of diagnostic signals | |
| CN111449645B (en) | Intelligent classification and identification method for electrocardiogram and heartbeat | |
| CN109303559B (en) | Dynamic electrocardiogram and heartbeat classification method based on gradient boosting decision tree | |
| CN111107785A (en) | Detecting atrial fibrillation using short single lead ECG recordings | |
| CN113057648A (en) | ECG signal classification method based on composite LSTM structure | |
| CN110522442B (en) | Multi-lead electrocardiographic abnormality detection device, electronic apparatus, and storage medium | |
| CN109948396B (en) | Heart beat classification method, heart beat classification device and electronic equipment | |
| CN112426160A (en) | Electrocardiosignal type identification method and device | |
| CN110840443B (en) | Electrocardiosignal processing method, electrocardiosignal processing device and electronic equipment | |
| CN112869716B (en) | Pulse feature identification system and method based on two-channel convolutional neural network | |
| CN113128585B (en) | Deep neural network based multi-size convolution kernel method for realizing electrocardiographic abnormality detection and classification | |
| CN115120248B (en) | Histogram-based adaptive threshold R peak detection and heart rhythm classification method and device | |
| CN115337018B (en) | Electrocardiogram signal classification method and system based on overall dynamic characteristics | |
| Mahesh et al. | ECG arrhythmia classification based on logistic model tree | |
| CN115281688A (en) | Cardiac hypertrophy multi-label detection system based on multi-mode deep learning | |
| CN111568410A (en) | Electrocardiogram data classification method based on 12-lead and convolutional neural network | |
| CN113647959B (en) | Waveform identification method, device and equipment for electrocardiographic waveform signals | |
| KR20140097039A (en) | Method and apparatus for classifying cardiac arrhythmia using an auto associative neural network | |
| Sultana et al. | MSVM-based classifier for cardiac arrhythmia detection | |
| Rai et al. | Classification of ECG waveforms for abnormalities detection using DWT and back propagation algorithm | |
| Ghosal et al. | Ecg beat quality assessment using self organizing map | |
| CN113712525B (en) | Physiological parameter processing method and device and medical equipment | |
| CN111345815B (en) | Method, device, equipment and storage medium for detecting QRS wave in electrocardiosignal |
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 |
Application publication date: 20200117 |
|
| RJ01 | Rejection of invention patent application after publication |