CN117898740A - Electrocardiogram processing method, device, equipment and medium - Google Patents
Electrocardiogram processing method, device, equipment and medium Download PDFInfo
- Publication number
- CN117898740A CN117898740A CN202410188663.4A CN202410188663A CN117898740A CN 117898740 A CN117898740 A CN 117898740A CN 202410188663 A CN202410188663 A CN 202410188663A CN 117898740 A CN117898740 A CN 117898740A
- Authority
- CN
- China
- Prior art keywords
- heart beat
- electrocardiogram
- waves
- waveform
- processing
- 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.)
- Granted
Links
- 238000003672 processing method Methods 0.000 title claims description 19
- 238000000034 method Methods 0.000 claims abstract description 63
- 238000012545 processing Methods 0.000 claims abstract description 60
- 238000001514 detection method Methods 0.000 claims abstract description 49
- 230000011218 segmentation Effects 0.000 claims abstract description 48
- 238000012549 training Methods 0.000 claims description 61
- 230000002159 abnormal effect Effects 0.000 claims description 20
- 230000000747 cardiac effect Effects 0.000 claims description 13
- 238000012797 qualification Methods 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 description 11
- 230000008569 process Effects 0.000 description 11
- 238000012360 testing method Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 8
- 238000000718 qrs complex Methods 0.000 description 8
- 238000013527 convolutional neural network Methods 0.000 description 7
- 238000013473 artificial intelligence Methods 0.000 description 6
- 238000011156 evaluation Methods 0.000 description 6
- 238000010606 normalization Methods 0.000 description 6
- 230000001746 atrial effect Effects 0.000 description 5
- 230000002861 ventricular Effects 0.000 description 5
- 206010003662 Atrial flutter Diseases 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 206010003658 Atrial Fibrillation Diseases 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000002565 electrocardiography Methods 0.000 description 2
- 238000011478 gradient descent method Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000013139 quantization Methods 0.000 description 2
- 208000024172 Cardiovascular disease Diseases 0.000 description 1
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 1
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000010009 beating Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 230000010247 heart contraction Effects 0.000 description 1
- 230000007721 medicinal effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 230000001020 rhythmical effect Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
- A61B5/353—Detecting P-waves
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
- A61B5/355—Detecting T-waves
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
- A61B5/357—Detecting U-waves
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
- A61B5/358—Detecting ST segments
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
- A61B5/361—Detecting fibrillation
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
- A61B5/363—Detecting tachycardia or bradycardia
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
- A61B5/366—Detecting abnormal QRS complex, e.g. widening
-
- 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
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Cardiology (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Biophysics (AREA)
- Surgery (AREA)
- Veterinary Medicine (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Molecular Biology (AREA)
- Medical Informatics (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Signal Processing (AREA)
- Psychiatry (AREA)
- Physiology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The application discloses a method, a device, equipment and a medium for processing an electrocardiogram, which relate to the technical field of data processing. By utilizing a pre-trained example segmentation model, two types of information, namely the position of a heart beat waveform and the heart beat type, can be extracted from an electrocardiogram, so that richer information is obtained. The electrocardiogram is divided into heart beat waveform images according to heart beat waveform position coordinates. And processing the heart beat waveform image by using the key point detection model to obtain the position information of the key points of the heart beat waveform image. The key point detection model is utilized to determine the position information of the key points of the heart beat waveform image, so that the recognition accuracy of the heart beat waveform can be improved, and especially the accuracy of the recognition of the electrocardiogram with poor image quality can be improved, thereby meeting the actual requirement of analyzing the electrocardiogram.
Description
Technical Field
The present application relates to the field of data processing technologies, and in particular, to a method, an apparatus, a device, and a medium for processing an electrocardiogram.
Background
Electrocardiography is a technique for recording, from a body surface, a pattern of changes in electrical activity produced by the heart for each cardiac cycle. With an electrocardiogram, the condition of the heart beat can be analyzed.
Along with the development of artificial intelligence technology, an electrocardiogram is processed by using an artificial intelligence model, and information contained in the electrocardiogram is automatically extracted. The information contained in the electrocardiogram extracted by the artificial intelligent model can be used as an objective basis to assist medical activities. However, the existing information extracted from the electrocardiogram by using the artificial intelligence model has the problems of inaccurate information and single information type, and is difficult to meet the actual demands.
Disclosure of Invention
In view of the above, the present application provides a method, apparatus, device and medium for processing an electrocardiogram, which aims to extract relatively abundant and relatively accurate information from an electrocardiogram, so as to meet practical requirements.
Based on the above, the technical scheme provided by the application is as follows:
in a first aspect, the present application provides a method for processing an electrocardiogram, the method comprising:
Processing an electrocardiogram by using an example segmentation model to obtain the position coordinates of a heart beat waveform of the electrocardiogram and the heart beat type; the heart beat waveform position coordinates comprise a P wave, a QRS wave group and a TU wave group, wherein the example segmentation model is obtained by pre-training first training data, and the first training data comprises a sample electrocardiogram, a heart beat waveform position label and a heart beat type label;
dividing the electrocardiogram into heart beat waveform images according to the heart beat waveform position coordinates;
processing the heart beat waveform image by using a key point detection model to obtain position information of key points of the heart beat waveform image; the key points comprise a starting point, an ending point and a peak point of an electrocardiographic waveform, the electrocardiographic waveform comprises one or more of P waves, Q waves, R waves, S waves, T waves and U waves, the key point detection model is obtained by training second training data in advance, and the second training data comprise sample heartbeat waveform images and key point labels.
In one possible implementation, the method further includes:
and generating characteristic parameters of the electrocardiogram by utilizing the coordinates of the key points.
In one possible implementation, the characteristic parameters of the electrocardiogram include one or more of a time parameter and an amplitude parameter of a cardiac cycle, the cardiac cycle being composed of adjacent P-waves, Q-waves, R-waves, S-waves, T-waves and U-waves.
In one possible implementation, the method further includes:
If the characteristic parameters do not accord with the preset qualification conditions corresponding to the characteristic parameters, generating an abnormal identifier, wherein the abnormal identifier is used for identifying that the characteristic parameters are abnormal.
In one possible implementation, the location information of the keypoints is coordinates of the keypoints.
In one possible implementation, the location information of the keypoints is a keypoint location heat map.
In one possible implementation, the method further includes:
Processing the heat map of the key point position by adopting a characteristic processing method to obtain a characteristic map; the feature processing method comprises one or more of normalizing the keypoint location heat map and convolving the keypoint location heat map;
and generating the position coordinates of the key points by using the feature map.
In a second aspect, the present application provides an electrocardiogram processing apparatus, the apparatus comprising:
A first processing unit for processing an electrocardiogram by using an example segmentation model to obtain a heart beat waveform position coordinate and a heart beat type of the electrocardiogram; the heart beat waveform position coordinates comprise a P wave, a QRS wave group and a TU wave group, wherein the example segmentation model is obtained by pre-training first training data, and the first training data comprises a sample electrocardiogram, a heart beat waveform position label and a heart beat type label;
A dividing unit for dividing the electrocardiogram into heart beat waveform images according to the heart beat waveform position coordinates;
the second processing unit is used for processing the heart beat waveform image by utilizing a key point detection model to obtain the position information of the key point of the heart beat waveform image; the key points comprise a starting point, an ending point and a peak point of an electrocardiographic waveform, the electrocardiographic waveform comprises one or more of P waves, Q waves, R waves, S waves, T waves and U waves, the key point detection model is obtained by training second training data in advance, and the second training data comprise sample heartbeat waveform images and key point labels.
In one possible implementation, the apparatus further includes:
and the first generation unit is used for generating characteristic parameters of the electrocardiogram by utilizing the coordinates of the key points.
In one possible implementation, the characteristic parameters of the electrocardiogram include one or more of a time parameter and an amplitude parameter of a cardiac cycle, the cardiac cycle being composed of adjacent P-waves, Q-waves, R-waves, S-waves, T-waves and U-waves.
In one possible implementation, the apparatus further includes:
the second generation unit is used for generating an abnormal identifier if the characteristic parameters do not accord with preset qualification conditions corresponding to the characteristic parameters, and the abnormal identifier is used for identifying that the characteristic parameters are abnormal.
In one possible implementation, the location information of the keypoints is coordinates of the keypoints.
In one possible implementation, the location information of the keypoints is a keypoint location heat map.
In one possible implementation, the apparatus further includes:
The third processing unit is used for processing the heat map of the key point position by adopting a characteristic processing method to obtain a characteristic map; the feature processing method comprises one or more of normalizing the keypoint location heat map and convolving the keypoint location heat map; and generating the position coordinates of the key points by using the feature map.
In a third aspect, the present application provides an electronic device comprising:
one or more processors;
a storage device having one or more programs stored thereon,
The one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of any of the first aspects.
In a fourth aspect, the present application provides a computer readable medium having stored thereon a computer program, wherein the program when executed by a processor implements the method of any of the first aspects.
From this, the application has the following beneficial effects:
The application provides an electrocardiogram processing method, device, equipment and medium, wherein the method utilizes an example segmentation model which is trained in advance to process the electrocardiogram so as to obtain a heart beat waveform position coordinate and a heart beat type. The heart beat waveform position coordinates can be used to determine the position of the different waves comprised by the electrocardiogram. The heart beat type can be used to determine the type of heart beat reflected by the electrocardiogram. By utilizing a pre-trained example segmentation model, two types of information, namely the position of a heart beat waveform and the heart beat type, can be extracted from an electrocardiogram, so that richer information is obtained. The electrocardiogram is divided into heart beat waveform images according to heart beat waveform position coordinates. And processing the heart beat waveform image by using the key point detection model to obtain the position information of the key points of the heart beat waveform image. The key point detection model is utilized to determine the position information of the key points of the heart beat waveform image, so that the recognition accuracy of the heart beat waveform can be improved, and especially the accuracy of the recognition of the electrocardiogram with poor image quality can be improved, thereby meeting the actual requirement of analyzing the electrocardiogram.
Drawings
Fig. 1 is a flowchart of a method for processing an electrocardiogram according to an embodiment of the present application;
fig. 2 is a schematic structural diagram of a Yolact network according to an embodiment of the present application;
FIG. 3 is a schematic diagram of a HRnet according to an embodiment of the present application;
FIG. 4 is a flowchart illustrating another method for processing an electrocardiogram according to an embodiment of the present application;
Fig. 5 is a schematic diagram of an electrocardiogram processing device according to an embodiment of the present application;
Fig. 6 is a schematic diagram of an apparatus according to an embodiment of the present application.
Detailed Description
In order to facilitate understanding and explanation of the technical solutions provided by the embodiments of the present application, the following description will first explain the background art of the present application.
Electrocardiography is a common clinical examination method for cardiovascular diseases. The image of the electrocardiogram can reflect the condition of the heart beating. Along with the development of the artificial intelligence model, part of the artificial intelligence model can be used for analyzing an electrocardiogram, so as to realize the identification of information such as waveforms in the electrocardiogram. At present, the information of the electrocardiogram obtained by utilizing the artificial intelligent model has the problems of single information type and inaccurate information type, and is difficult to meet the actual requirements.
Based on this, the embodiment of the application provides a processing method of an electrocardiogram, which processes the electrocardiogram by using an example segmentation model which is trained in advance to obtain the position coordinates of a heart beat waveform and the heart beat type. The heart beat waveform position coordinates can be used to determine the position of the different waves comprised by the electrocardiogram. The heart beat type can be used to determine the type of heart beat reflected by the electrocardiogram. By utilizing a pre-trained example segmentation model, two types of information, namely the position of a heart beat waveform and the heart beat type, can be extracted from an electrocardiogram, so that richer information is obtained. The electrocardiogram is divided into heart beat waveform images according to heart beat waveform position coordinates. And processing the heart beat waveform image by using the key point detection model to obtain the position information of the key points of the heart beat waveform image. The key point detection model is utilized to determine the position information of the key points of the heart beat waveform image, so that the recognition accuracy of the heart beat waveform can be improved, and especially the accuracy of the recognition of the electrocardiogram with poor image quality can be improved, thereby meeting the actual requirement of analyzing the electrocardiogram.
In order to facilitate understanding of the technical solution provided by the embodiments of the present application, the following describes an electrocardiogram processing method provided by the embodiments of the present application with reference to the accompanying drawings.
Referring to fig. 1, the flowchart of an electrocardiogram processing method according to an embodiment of the present application is shown. As shown in FIG. 1, the method for processing an electrocardiogram according to the embodiment of the application includes S101-S103.
S101: and processing the electrocardiogram by using the example segmentation model to obtain the position coordinates of the heart beat waveform of the electrocardiogram and the heart beat type.
An example segmentation model is a model that has been trained in advance. The embodiment of the application is not limited to the model type of the example segmentation model. By way of example, the example segmentation model is composed of a convolutional neural network (Convolutional Neural Networks, CNN), a mask region-based convolutional neural network (Mask Region-based Convolutional Neural Network,Mask R-CNN)、SOLO(Segmenting Objects by Locations,)、Yolact(You Only Look At CoefficienTs), or blendmask (an example segmentation network), which is not limited by the embodiments of the application. As an example, referring to fig. 2, a schematic structural diagram of a Yolact network according to an embodiment of the present application is shown.
The example segmentation model is used to process an input electrocardiogram, and output the heart beat waveform position coordinates and heart beat type of the electrocardiogram. The heart beat waveform position coordinates can be the abscissa in the electrocardiogram of the start and stop points of the heart beat waveform position. The heart beat type is a heart beat type that the example segmentation model can identify.
The example segmentation model is pre-trained from first training data. The first training data includes a sample electrocardiogram, a heart beat waveform location tag of the sample electrocardiogram, and a heart beat type tag.
The sample electrocardiogram can be an electrocardiogram obtained by preprocessing an original electrocardiogram as a training sample.
As an example, the original electrocardiogram is preprocessed using signal processing techniques to obtain a sample electrocardiogram. The signal processing technology comprises the steps of removing signal noise, normalizing and the like, so that the quality of a sample electrocardiogram is improved, and the training efficiency and accuracy of an example segmentation model are improved. A data enhancement method can also be used to obtain a sample electrocardiogram, and a new sample electrocardiogram can be generated. The data enhancement method includes, for example, random up-and-down scaling, random left-and-right inversion, random up-and-down inversion, random left-and-right inversion, etc. The sample can be expanded by using the data enhancement method, and the generalization capability of the instance segmentation model is improved.
A heart beat waveform position label of the sample electrocardiogram is used to mark the position of the heart beat waveform. Heart beat is a rhythmic contractile movement of the heart. The heart beat waveform is a waveform representing the beating of the heart in an electrocardiogram. The heart beat waveform position label is, for example, heart beat waveform position coordinates. The heart beat waveform position coordinates are, for example, the abscissa in the electrocardiogram of the start and stop points of the heart beat waveform position. The abscissa in the electrocardiogram is used to represent the time of waveform generation. The heart beat waveform position coordinates include a start position coordinate of the P wave, a stop position coordinate of the P wave, a start position coordinate of the QRS complex, a stop position coordinate of the QRS complex, a start position coordinate of the TU complex, and a stop position coordinate of the TU complex.
The heart beat type label of the sample electrocardiogram is used for marking the heart beat type of the sample electrocardiogram. The heart beat types can be divided based on the need to classify heart beats. By way of example, the heart beat types include one or more of sinus beats, atrial beats, ventricular beats, border zone beats, atrial flutter, atrial fibrillation, atrial pacing, ventricular pacing, atrioventricular sequential pacing, artifacts, and noise. The type of heart beat can be identified with corresponding letters, such as sinus beat (N), atrial beat (S), ventricular beat (V), junctional regional beat (J), atrial Flutter (AF), atrial Fibrillation (AF), atrial pacing (P), ventricular pacing (a), atrioventricular sequential pacing (D), artifact or noise (X).
In addition, the image of the sample electrocardiogram is resized. The image size of the sample electrocardiogram is adjusted to a fixed size which can be processed by a preset example segmentation model. In the case where the sample electrocardiogram has marked the heart beat waveform position label, the heart beat waveform position label is modified according to the adjustment ratio of the image size.
The penalty functions for the example segmentation model are mask (mask) confidence class penalty L c1, mask class penalty L c2, bounding box segmentation penalty L mask.
The loss function expressions are as follows:
Lc1＝-(ylogp(x)+(1-y)log(1-p(x)) (1)
Lc2＝-(ylogp(x)+(1-y)log(1-p(x)) (2)
Ly＝Lc1+Lc2+αLmask (4)
mask confidence classification penalty L c1 employs binary cross entropy (Binary Cross Entropy, BCE) penalty, P (x) is model class probability, and y is heart beat type label.
Mask class classification loss L c2 adopts BCE loss, P (x) is model classification probability, and y is heart beat waveform position label.
Mask penalty L mask, the bounding box partition penalty, employs overlap (Intersection over Union, IOU) penalty. Omega c is the training scale of channel c, and the value range is (0, 1); is a binary mask model; /(I) Is a soft prediction mask model.
Alpha is a weight coefficient. The value of α is, for example, 1.
And performing repeated iterative training on the example segmentation model by using the first training data, and adjusting model parameters. The training of the example segmentation model is stopped until the number of iterations reaches a predetermined value, or when the value of the loss function no longer decreases or falls below a predetermined value.
In some possible implementations, a random number is employed as an initial value for the example segmentation model. An example segmentation model is trained using the first training data. The first training data is data in a training set. The model parameters of the instance segmentation model are adjusted through a gradient descent method, the model parameters are optimized, and the accuracy of the instance segmentation model is improved.
After training is completed with the training data in the training set, the accuracy of the instance segmentation model can also be tested using the test data in the validation set. The test data comprises a sample electrocardiogram, and heart beat waveform position coordinates and heart beat types corresponding to the sample electrocardiogram.
And inputting the test data included in the verification set into the example segmentation model to obtain the heart beat waveform position coordinates and the heart beat type output by the example segmentation model. And comparing the heart beat waveform position coordinates and the heart beat types output by the example segmentation model with the heart beat waveform position coordinates and the heart beat types corresponding to the test data respectively to realize performance evaluation of the example segmentation model. As an example, performance assessment of an example segmentation model using average accuracy (mAP) and Jaccard index. The average accuracy is used for evaluating the prediction effect of classification under different thresholds, and the Jaccard index is used for evaluating the difference between the output result and the real result according to the model. In addition, the distribution of the absolute percentage error of the test data is also in the evaluation range, so that the accuracy of the performance evaluation of the example segmentation model can be improved.
S102: the electrocardiogram is divided into heart beat waveform images according to heart beat waveform position coordinates.
The heart beat waveform position coordinates can identify a start position and an end position of the heart beat waveform. The heart beat waveform image included in the electrocardiogram is divided by using the heart beat waveform position coordinates.
As an example, the heart beat waveform position coordinates include a start position coordinate of a P wave, a stop position coordinate of a P wave, a start position coordinate of a QRS complex, a stop position coordinate of a QRS complex, a start position coordinate of a TU complex, and a stop position coordinate of a TU complex. The heart beat waveform image includes a P-wave waveform image, a QRS complex waveform image, and a TU complex waveform image. And dividing the P wave waveform image according to the starting position coordinate of the P wave and the ending position coordinate of the P wave. The QRS complex waveform image is divided according to the start position coordinates of the QRS complex and the end position coordinates of the QRS complex. And dividing the TU wave group waveform image according to the initial position coordinates of the TU wave groups and the final position coordinates of the TU wave groups.
S103: and processing the heart beat waveform image by using the key point detection model to obtain the position information of the key points of the heart beat waveform image.
The heart beat waveform image is obtained by roughly dividing an electrocardiogram. The divided heart beat waveform image contains less information, so that key points of the heart beat waveform image can be conveniently identified.
And processing the heart beat waveform image by using the key point detection model. The keypoint detection model is used to identify keypoints of the heart beat waveform image. Key points of the heart beat waveform image include a start point, an end point and a peak point of the electrocardiographic waveform. The electrocardiographic waveform includes one or more of P-wave, Q-wave, R-wave, S-wave, T-wave, and U-wave. As an example, for an input P-wave waveform image, the key point detection model outputs the start point, end point, and peak point of the identified P-wave. For an input QRS wave group waveform image, the key point detection model outputs the starting point, the ending point and the peak point of the Q wave, the starting point, the ending point and the peak point of the R wave and the starting point, the ending point and the peak point of the S wave which are obtained by identification. And outputting the starting point, the ending point and the peak point of the T wave obtained by identification by the key point detection model for the input TU wave group waveform image. It should be noted that in some possible implementations, the number of peak points of each waveform may be one or more.
The key point detection model outputs position information of key points of the heart beat waveform image.
The embodiment of the application is not limited to the model type of the key point detection model. As an example, the keypoint detection model is composed of a network structure such as CNN, simple baseline (simple baseline) network, hourglass (hourglass) network, high-Resolution Network, HRnet, u-net, or feature pyramid network (Feature Pyramid Networks, FPN). Referring to fig. 3, a schematic structural diagram of HRnet according to an embodiment of the present application is shown;
The key point detection model is obtained by training the second training data in advance. The second training data includes a sample heartbeat waveform image and a keypoint tag.
The sample heart beat waveform image can be an electrocardiogram obtained by preprocessing an original electrocardiogram as a training sample.
As an example, the original electrocardiogram is preprocessed using signal processing techniques to obtain a sample electrocardiogram. The signal processing technology comprises the steps of removing signal noise, normalizing and the like, so that the quality of a sample electrocardiogram is improved, and the training efficiency and accuracy of an example segmentation model are improved. A data enhancement method can also be used to obtain a sample electrocardiogram, and a new sample electrocardiogram can be generated. The data enhancement method includes, for example, random up-and-down scaling, random left-and-right inversion, random up-and-down inversion, random left-and-right inversion, etc. The sample can be expanded by using the data enhancement method, and the generalization capability of the instance segmentation model is improved.
As an example, the keypoint labels are coordinates of keypoints of the sample heart beat waveform image. The key points of the sample heart beat waveform image include the following key points:
A P wave starting point P0, a P wave crest value point P1, a P wave ending point P2 and other peak points P1 of the P wave;
A Q wave starting point Q0, a Q wave crest value point Q1 and a Q wave ending point Q2; r wave start point R0, R wave peak value point R1, R wave end point R2, R wave other peak points R1; an S wave starting point S0, an S wave crest value point S1 and an S wave ending point S2;
t wave start point T0, T wave peak value point T1, T wave end point T2, T wave other peak points T1; a U wave starting point U0, a U wave crest value point U1 and a U wave ending point U2. For waveforms lacking keypoints, no label is set.
In some possible implementations, the results output by part of the keypoint detection model are the coordinates of the keypoints. That is, the key point position information output by the key point detection model is the key point coordinates. The result of the partial keypoint detection model output is a keypoint location heat map (heatmap) that represents the confidence of the keypoint. The keypoint location heatmap needs to be converted to keypoint coordinates in order to provide a more intuitive recognition result of the keypoints.
The embodiment of the application provides a mode for converting a key point position heat map into key point coordinates, which comprises the following steps:
A1: and processing the heat map of the key point position by adopting a characteristic processing method to obtain a characteristic map.
The feature processing method comprises one or more of normalization processing of the keypoint location heat map and convolution operation of the keypoint location heat map.
The keypoint location heat map normalization process is normalized, for example, according to equation (5).
X min and X max are the minimum and maximum confidence values, respectively, in the keypoint location heat map. X is the confidence of any point in the keypoint location heat map.
And carrying out normalization processing on the confidence coefficient of the heat map of the key point position to obtain a feature map. Normalization enables the numerical range in the key point position heat map to be between 0 and 1, and reduces quantization precision loss.
The convolution operation on the keypoint location heatmap refers to a grouping convolution operation on the keypoint location heatmap. The grouping convolution operation is to divide input data into a plurality of groups in the convolution operation, then carry out the convolution operation on each group, and finally combine the convolution results of the groups together to obtain a feature map. Features of the key point position heat map can be further extracted through grouping convolution operation, and accurate coordinate conversion results can be obtained.
The embodiments of the present application are not limited in the type of feature processing method employed. In one possible implementation, only the keypoint location heat map is normalized, or only the keypoint location heat map is convolved in groups, or the keypoint location heat map is normalized first and then convolved in groups.
A2: and generating the position coordinates of the key points by using the feature map.
And determining the coordinates of the point with the maximum confidence included in the feature map to obtain the coordinates of the key point. As an example, the feature map is processed using an argmax function to obtain the key point location coordinates.
By adopting the feature processing method, the quantization precision loss generated by processing the heat map and the error caused by argmax function can be reduced, the precision loss is reduced, and the accuracy of the coordinates of the key points is improved. The method for converting the key point position heat map into the key point coordinates can be applied to the model training stage, and the generated key point position coordinates can be back-propagated in the key point detection model, so that the performance of the key point detection model is improved.
The embodiment of the application provides a possible implementation mode for training a key point detection model. The keypoint detection model uses a modified linear unit (RECTIFIED LINEAR unit, reLU) as an activation function and batch normalization (batch normalization, BN) to prevent overfitting.
As an example, the loss function of the keypoint detection model uses ELASTIC NET loss functions. ELASTIC NET loss corresponds to the addition of the L1 loss function (L1 loss) and the L2 loss function (L2 loss), expressed as:
Wherein y n is the key point label corresponding to the nth key point, that is, the coordinates of the nth key point, And J is the total number of the key points, wherein the predicted coordinates are the n-th key points.
And stopping training the key point detection model when the training iteration times reach the preset times or when the value of the L y of the loss function is no longer reduced or is lower than a preset value.
In some possible implementations, a random number is employed as an initial value for model parameters of the keypoint detection model. And training a key point detection model by using the second training data. The second training data is data in a training set. And the model parameters of the key point detection model are regulated by a gradient descent method, the model parameters are optimized, and the accuracy of the key point detection model is improved.
After training is completed by using the training data in the training set, the accuracy of the key point detection model can also be tested by using the test data in the verification set. The test data includes a sample heart beat waveform image, and coordinates of key points of the sample heart beat waveform image.
And inputting the sample heart beat waveform image included in the test data included in the verification set into the key point detection model to obtain the position information of the key point output by the key point detection model. And comparing the position information of the key points output by the key point detection model with coordinates of the key points of the sample heart beat waveform image included in the test data to realize performance evaluation of the key point detection model. As an example, the performance of the keypoint detection model is evaluated using a mean square error (mean squared error, MSE) and an average absolute percentage error (mean absolute percentage error, MAPE). The mean square error is used for evaluating the prediction effect of the coordinates of the key points, and the average absolute percentage error is used for evaluating the prediction effect of the coordinates of the key points. In addition, the distribution condition of the absolute percentage error of the test data is also in the evaluation range, so that the accuracy of the performance evaluation of the key point detection model can be improved.
In one possible implementation, the heart beat waveform position coordinates, heart beat type, and location information of key points are displayed for viewing, enabling automated electrocardiographic processing and display of processing results.
Based on the above-mentioned content related to S101-S103, it can be known that an example segmentation model and a keypoint detection model with more stable performance can be obtained by adopting a training manner of an end-to-end artificial intelligence model. Not only can the heart beat waveform position be obtained by using the example segmentation model so as to divide heart beat waveform images, but also the heart beat type can be obtained. Thus, a richer and comprehensive electrocardiogram processing result can be obtained. The key point detection model is utilized to process the divided heart beat waveform image, so that the recognition accuracy of the heart beat waveform can be improved, and the accuracy of the recognition of the electrocardiogram with poor image quality can be improved. The electrocardiogram processing method provided by the embodiment of the application can realize high-automation electrocardiogram processing, avoid interference and artificial error caused by subjective factors, improve consistency and repeatability of processing results and can meet actual demands.
Furthermore, after the electrocardiogram is processed by using the example segmentation model and the key point detection model to obtain the position information of the key points, the characteristic parameters of the electrocardiogram can be generated by using the position information of the key points.
Referring to fig. 4, a flowchart of another method for processing an electrocardiogram according to an embodiment of the present application is shown. In addition to the above S101-S103, the method for processing an electrocardiogram provided by the embodiment of the present application further includes the following steps:
s104: and generating characteristic parameters of the electrocardiogram by using the coordinates of the key points.
The embodiment of the application is not limited to the specific content of the characteristic parameters of the electrocardiogram. The characteristic parameters of the electrocardiogram can be set based on the needs of the electrocardiogram process. As an example, the characteristic parameters of the electrocardiogram comprise one or more of a time parameter and an amplitude parameter of the cardiac cycle. The cardiac cycle is composed of adjacent P-waves, Q-waves, R-waves, S-waves, T-waves and U-waves.
The time parameters include, for example, P-wave time limit, Q-wave time limit, R-wave time limit, S-wave time limit, QRS time limit, T-wave time limit, U-wave time limit, PR interval time limit, QT segment time limit, QT interval correction, JT interval correction, ventricular rate, atrial rate, and the like. The time limit is the length of time from the start point to the end point. I.e. the difference between the abscissa of the end point and the abscissa of the start point in the electrocardiogram. The unit of time limit is, for example, milliseconds.
The amplitude parameters include, for example, P-wave amplitude, Q-wave amplitude, R-wave amplitude, S-wave amplitude, R/S amplitude ratio, T-wave amplitude, U-wave amplitude, and ST-segment amplitude. The amplitude is the difference between the highest peak height and the height of the starting point. The coordinates of both the peak point and the starting point can be determined using the coordinates of the key point. The unit of amplitude is, for example, millivolts (mv).
When calculating the characteristic parameters of the electrocardiogram, the characteristic parameters of the cardiac cycle of the waveform corresponding to each lead can be calculated, and the median of the characteristic parameters of the cardiac cycle included in the waveform can be used as the characteristic parameters of the electrocardiogram.
Therefore, the characteristic parameters of the electrocardiogram can be automatically analyzed and obtained, a comprehensive analysis result of the electrocardiogram can be provided, and the requirement of analyzing the electrocardiogram is met.
After the characteristic parameters of the electrocardiogram are obtained, whether the characteristic parameters of the electrocardiogram meet preset qualification conditions corresponding to the characteristic parameters can be judged. The preset qualification conditions corresponding to the characteristic parameters are preset conditions for determining that the characteristic parameters are normal. If the characteristic parameters do not accord with the corresponding preset qualification conditions, the characteristic parameters are indicated to be abnormal, and an abnormal mark is generated. The anomaly identification is used for identifying that the characteristic parameters are abnormal. The generated abnormality identification facilitates prompting a viewer of the electrocardiogram for abnormalities. Further, for different exception types, the method has corresponding preset qualification conditions. And under the condition that the characteristic parameters meet the preset qualified conditions corresponding to the abnormal types, the generated abnormal identifiers are also used for identifying the abnormal types of the characteristic parameters. Thus, the abnormal situation can be further divided, and more detailed analysis results can be obtained.
In some possible implementations, taking the above-described heart beat type as an example, it is determined whether the heart beat type identified by the example segmentation model includes a target heart beat type. The target heart beat type is a preset heart beat type with possible abnormal type. As an example, the target heart beat type is a heart beat type other than sinus heart beat, artifact, and noise among the above-described heart beat types. If the heart beat type output by the example segmentation model includes the target heart beat type, an anomaly identification is generated indicating that an anomaly exists in the heart beat type.
Based on the method embodiment, the embodiment of the application also provides an electrocardiogram processing device.
Referring to fig. 5, fig. 5 is a schematic diagram of an electrocardiographic processing device according to an embodiment of the present application, where the device includes:
a first processing unit 501, configured to process an electrocardiogram by using an example segmentation model, and obtain a position coordinate of a heartbeat waveform and a heartbeat type of the electrocardiogram; the heart beat waveform position coordinates comprise a P wave, a QRS wave group and a TU wave group, wherein the example segmentation model is obtained by pre-training first training data, and the first training data comprises a sample electrocardiogram, a heart beat waveform position label and a heart beat type label;
A dividing unit 502, configured to divide the electrocardiogram into heart beat waveform images according to the heart beat waveform position coordinates;
a second processing unit 503, configured to process the heartbeat waveform image by using a key point detection model, so as to obtain position information of a key point of the heartbeat waveform image; the key points comprise a starting point, an ending point and a peak point of an electrocardiographic waveform, the electrocardiographic waveform comprises one or more of P waves, Q waves, R waves, S waves, T waves and U waves, the key point detection model is obtained by training second training data in advance, and the second training data comprise sample heartbeat waveform images and key point labels.
In one possible implementation, the apparatus further includes:
and the first generation unit is used for generating characteristic parameters of the electrocardiogram by utilizing the coordinates of the key points.
In one possible implementation, the characteristic parameters of the electrocardiogram include one or more of a time parameter and an amplitude parameter of a cardiac cycle, the cardiac cycle being composed of adjacent P-waves, Q-waves, R-waves, S-waves, T-waves and U-waves.
In one possible implementation, the apparatus further includes:
the second generation unit is used for generating an abnormal identifier if the characteristic parameters do not accord with preset qualification conditions corresponding to the characteristic parameters, and the abnormal identifier is used for identifying that the characteristic parameters are abnormal.
In one possible implementation, the location information of the keypoints is coordinates of the keypoints.
In one possible implementation, the location information of the keypoints is a keypoint location heat map.
In one possible implementation, the apparatus further includes:
The third processing unit is used for processing the heat map of the key point position by adopting a characteristic processing method to obtain a characteristic map; the feature processing method comprises one or more of normalizing the keypoint location heat map and convolving the keypoint location heat map; and generating the position coordinates of the key points by using the feature map.
Referring to fig. 6, fig. 6 is a schematic diagram of an apparatus according to an embodiment of the present application.
The apparatus includes: a memory 601 and a processor 602;
the memory 601 is used for storing relevant program codes;
the processor 602 is configured to invoke the program code to perform the method for processing an electrocardiogram according to the method embodiment described above.
Furthermore, the embodiment of the application also provides a computer readable storage medium for storing a computer program for executing the method for processing the electrocardiogram according to the embodiment of the method.
It should be noted that, in the present description, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different manner from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the system or device disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and the relevant points refer to the description of the method section.
It should be understood that in the present application, "at least one (item)" means one or more, and "a plurality" means two or more. "and/or" for describing the association relationship of the association object, the representation may have three relationships, for example, "a and/or B" may represent: only a, only B and both a and B are present, wherein a, B may be singular or plural. The character "/" generally indicates that the context-dependent object is an "or" relationship. "at least one of" or the like means any combination of these items, including any combination of single item(s) or plural items(s). For example, at least one (one) of a, b or c may represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", wherein a, b, c may be single or plural.
It is further noted that relational terms such as first and second, and the like are 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. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus 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 apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. The software modules may be disposed in Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. A method of processing an electrocardiogram, the method comprising:
Processing an electrocardiogram by using an example segmentation model to obtain the position coordinates of a heart beat waveform of the electrocardiogram and the heart beat type; the heart beat waveform position coordinates comprise a P wave, a QRS wave group and a TU wave group, wherein the example segmentation model is obtained by pre-training first training data, and the first training data comprises a sample electrocardiogram, a heart beat waveform position label and a heart beat type label;
dividing the electrocardiogram into heart beat waveform images according to the heart beat waveform position coordinates;
processing the heart beat waveform image by using a key point detection model to obtain position information of key points of the heart beat waveform image; the key points comprise a starting point, an ending point and a peak point of an electrocardiographic waveform, the electrocardiographic waveform comprises one or more of P waves, Q waves, R waves, S waves, T waves and U waves, the key point detection model is obtained by training second training data in advance, and the second training data comprise sample heartbeat waveform images and key point labels.
2. The method according to claim 1, wherein the method further comprises:
and generating characteristic parameters of the electrocardiogram by utilizing the coordinates of the key points.
3. The method of claim 2, wherein the characteristic parameters of the electrocardiogram include one or more of a time parameter and an amplitude parameter of a cardiac cycle, the cardiac cycle being composed of adjacent P-waves, Q-waves, R-waves, S-waves, T-waves, and U-waves.
4. The method according to claim 2, wherein the method further comprises:
If the characteristic parameters do not accord with the preset qualification conditions corresponding to the characteristic parameters, generating an abnormal identifier, wherein the abnormal identifier is used for identifying that the characteristic parameters are abnormal.
5. The method of any one of claims 1-4, wherein the location information of the keypoints is coordinates of the keypoints.
6. The method of any one of claims 1-4, wherein the location information of the keypoints is a keypoint location heat map.
7. The method of claim 6, wherein the method further comprises:
Processing the heat map of the key point position by adopting a characteristic processing method to obtain a characteristic map; the feature processing method comprises one or more of normalizing the keypoint location heat map and convolving the keypoint location heat map;
and generating the position coordinates of the key points by using the feature map.
8. An electrocardiogram processing apparatus, the apparatus comprising:
A first processing unit for processing an electrocardiogram by using an example segmentation model to obtain a heart beat waveform position coordinate and a heart beat type of the electrocardiogram; the heart beat waveform position coordinates comprise a P wave, a QRS wave group and a TU wave group, wherein the example segmentation model is obtained by pre-training first training data, and the first training data comprises a sample electrocardiogram, a heart beat waveform position label and a heart beat type label;
A dividing unit for dividing the electrocardiogram into heart beat waveform images according to the heart beat waveform position coordinates;
the second processing unit is used for processing the heart beat waveform image by utilizing a key point detection model to obtain the position information of the key point of the heart beat waveform image; the key points comprise a starting point, an ending point and a peak point of an electrocardiographic waveform, the electrocardiographic waveform comprises one or more of P waves, Q waves, R waves, S waves, T waves and U waves, the key point detection model is obtained by training second training data in advance, and the second training data comprise sample heartbeat waveform images and key point labels.
9. An electronic device, comprising:
one or more processors;
Storage means having stored thereon one or more programs which, when executed by the one or more processors, cause the one or more processors to implement the method of processing an electrocardiogram as claimed in any one of claims 1 to 7.
10. A computer-readable medium, on which a computer program is stored, wherein the program, when being executed by a processor, implements the method of processing an electrocardiogram according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410188663.4A CN117898740B (en) | 2024-02-20 | 2024-02-20 | Electrocardiogram processing method, device, equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410188663.4A CN117898740B (en) | 2024-02-20 | 2024-02-20 | Electrocardiogram processing method, device, equipment and medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117898740A true CN117898740A (en) | 2024-04-19 |
| CN117898740B CN117898740B (en) | 2024-07-09 |
Family
ID=90683933
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410188663.4A Active CN117898740B (en) | 2024-02-20 | 2024-02-20 | Electrocardiogram processing method, device, equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117898740B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107714023A (en) * | 2017-11-27 | 2018-02-23 | 乐普（北京）医疗器械股份有限公司 | Static ecg analysis method and apparatus based on artificial intelligence self study |
| CN110495872A (en) * | 2019-08-27 | 2019-11-26 | 中科麦迪人工智能研究院(苏州)有限公司 | Ecg analysis method, apparatus, equipment and medium based on picture and heart-beat messages |
| CN111772619A (en) * | 2020-06-19 | 2020-10-16 | 厦门纳龙科技有限公司 | Electrocardiogram heart beat identification method, terminal device and storage medium |
| CN113876335A (en) * | 2020-06-17 | 2022-01-04 | 深圳邦健生物医疗设备股份有限公司 | Electrocardiogram heart beat classification method, electrocardiogram heart beat classification device, electronic equipment and medium |
| CN115497116A (en) * | 2021-06-02 | 2022-12-20 | 武汉Tcl集团工业研究院有限公司 | Key point detection method, device, equipment and computer readable storage medium |
-
2024
- 2024-02-20 CN CN202410188663.4A patent/CN117898740B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107714023A (en) * | 2017-11-27 | 2018-02-23 | 乐普（北京）医疗器械股份有限公司 | Static ecg analysis method and apparatus based on artificial intelligence self study |
| CN110495872A (en) * | 2019-08-27 | 2019-11-26 | 中科麦迪人工智能研究院(苏州)有限公司 | Ecg analysis method, apparatus, equipment and medium based on picture and heart-beat messages |
| CN113876335A (en) * | 2020-06-17 | 2022-01-04 | 深圳邦健生物医疗设备股份有限公司 | Electrocardiogram heart beat classification method, electrocardiogram heart beat classification device, electronic equipment and medium |
| CN111772619A (en) * | 2020-06-19 | 2020-10-16 | 厦门纳龙科技有限公司 | Electrocardiogram heart beat identification method, terminal device and storage medium |
| CN115497116A (en) * | 2021-06-02 | 2022-12-20 | 武汉Tcl集团工业研究院有限公司 | Key point detection method, device, equipment and computer readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117898740B (en) | 2024-07-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Shi et al. | A hierarchical method based on weighted extreme gradient boosting in ECG heartbeat classification | |
| Zhou et al. | Beatgan: Anomalous rhythm detection using adversarially generated time series. | |
| US20230131876A1 (en) | Systems and methods of identity analysis of electrocardiograms | |
| CN109948647B (en) | Electrocardiogram classification method and system based on depth residual error network | |
| US11617528B2 (en) | Systems and methods for reduced lead electrocardiogram diagnosis using deep neural networks and rule-based systems | |
| Javadi et al. | Improving ECG classification accuracy using an ensemble of neural network modules | |
| CN111523502B (en) | Electrocardiogram image processing method, equipment, medium and electrocardiograph | |
| CN110974214A (en) | Automatic electrocardiogram classification method, system and equipment based on deep learning | |
| WO2019161611A1 (en) | Ecg information processing method and ecg workstation | |
| CN1951320A (en) | Abnormal electrocardiogram recognition method based on ultra-complete characteristics | |
| Dhyani et al. | Arrhythmia disease classification utilizing ResRNN | |
| Banerjee et al. | A hybrid CNN-LSTM architecture for detection of coronary artery disease from ECG | |
| CN116602642A (en) | Heart rate monitoring method, device and equipment | |
| CN116503673B (en) | Arrhythmia identification and detection method and system based on electrocardiogram | |
| WO2021071646A1 (en) | Systems and methods for electrocardiogram diagnosis using deep neural networks and rule-based systems | |
| Hassan et al. | Performance comparison of CNN and LSTM algorithms for arrhythmia classification | |
| CN117898740B (en) | Electrocardiogram processing method, device, equipment and medium | |
| Hua et al. | An interpretable model for ECG data based on Bayesian neural networks | |
| US20210298625A1 (en) | System and method for detecting and predicting an occurrence of cardiac events from electrocardiograms | |
| Wang et al. | A multi-module algorithm for heartbeat classification based on unsupervised learning and adaptive feature transfer | |
| CN116636858A (en) | ECG sensing signal coding method and system | |
| CN114366116A (en) | Parameter acquisition method based on Mask R-CNN network and electrocardiogram | |
| CN113171102A (en) | ECG data classification method based on continuous deep learning | |
| CN113349790B (en) | Heart beat classification method and system of 2D convolutional neural network based on S-shaped reconstruction | |
| Jayasinghe | A Real-Time Framework for Arrhythmia Classification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |