CN111803061B - R wave identification method and device based on target detection - Google Patents
R wave identification method and device based on target detection Download PDFInfo
- Publication number
- CN111803061B CN111803061B CN202010596685.6A CN202010596685A CN111803061B CN 111803061 B CN111803061 B CN 111803061B CN 202010596685 A CN202010596685 A CN 202010596685A CN 111803061 B CN111803061 B CN 111803061B
- Authority
- CN
- China
- Prior art keywords
- wave
- target detection
- sequence
- interval
- electrocardiogram
- 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.)
- Active
Links
Images
Landscapes
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention relates to the technical field of electrocardiosignal R wave identification, and discloses an R wave identification method based on target detection, which comprises the following steps: acquiring electrocardiosignals, and drawing an electrocardiogram according to the electrocardiosignals; marking a marking frame at the position of the R wave in the electrocardiogram to obtain a sample data set; training a target detection network by adopting the sample data set to obtain a target detection model; and identifying the R wave position of the electrocardiosignal to be detected according to the target detection model. The invention has the technical effect that the R wave recognition effect of the electrocardiosignals does not depend on the signal processing process of the electrocardiosignals.
Description
Technical Field
The invention relates to the technical field of electrocardiosignal R wave identification, in particular to a target detection-based R wave identification method and device and a computer storage medium.
Background
The current main starting point for the identification of ECG signals (electrocardiographic signals) is the determination of QRS waves, and the most important of these is the accurate location of R waves. The R wave identification method widely used at present comprises the following steps: the method is widely applied to PT (Pan-moments) algorithm, novel R wave identification technology based on amplitude and slope, R wave identification by adopting sampling point second-order differential value extraction characteristics and the like.
In the prior art, methods such as difference threshold, wavelet transformation or waveform feature identification have high requirements on ECG signal quality. If the identified ECG signal has serious error condition, the identification accuracy rate is greatly reduced. Therefore, before the method is carried out, multiple signal processing needs to be carried out on the electrocardiosignal waveform, and the recognition effect depends on the signal processing effect seriously.
Disclosure of Invention
The invention aims to overcome the technical defects and provide an R wave identification method, an R wave identification device and a computer storage medium based on target detection, so as to solve the technical problem that the R wave identification precision depends on the electrocardiosignal processing effect in the prior art.
In order to achieve the technical purpose, the technical scheme of the invention provides an R-wave identification method based on target detection, which comprises the following steps:
acquiring electrocardiosignals, and drawing an electrocardiogram according to the electrocardiosignals;
marking a marking frame at the position of the R wave in the electrocardiogram to obtain a sample data set;
training a target detection network by adopting the sample data set to obtain a target detection model;
and identifying the R wave position of the electrocardiosignal to be detected according to the target detection model.
The invention also provides an R wave identification device based on target detection, which comprises a processor and a memory, wherein the memory is stored with a computer program, and when the computer program is executed by the processor, the R wave identification device based on target detection is realized.
The present invention also provides a computer storage medium having stored thereon a computer program which, when executed by a processor, implements the object detection-based R-wave identification method.
Compared with the prior art, the invention has the beneficial effects that: the invention firstly draws an electrocardiogram according to the electrocardiosignals, converts the electrocardiosignals into image signals and further converts the problems of the electrocardiosignals identification into the problems of the image identification. The application of the target detection technology in image processing has higher accuracy, so that after the R wave identification problem is converted into the image identification problem, the target frame at the R wave position in the electrocardiogram can be identified by adopting the target detection technology, and the identification result of the R wave position is further realized. The method carries out R wave position identification based on target detection, and the target detection has higher identification accuracy, so that the requirement on the quality of the electrocardiosignal is lower during identification, and the R wave identification effect does not depend on a signal processing means seriously. When the method is applied, signal processing is not needed, so that the recognition rate is high, the recognition can be carried out while the acquisition is carried out, and the real-time recognition is realized.
Drawings
Fig. 1 is a flowchart of an embodiment of an R-wave identification method based on target detection according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention.
Example 1
As shown in fig. 1, embodiment 1 of the present invention provides an R-wave identification method based on target detection, including the following steps:
s1, acquiring electrocardiosignals, and drawing an electrocardiogram according to the electrocardiosignals;
s2, marking a marking frame at the R wave position in the electrocardiogram to obtain a sample data set;
s3, training a target detection network by adopting the sample data set to obtain a target detection model;
and S4, identifying the R wave position of the electrocardiosignal to be detected according to the target detection model.
At present, the target detection deep learning technology is mainly applied to image processing, such as face recognition, walking recognition and the like, good results have been obtained in the fields, and higher accuracy can be realized. Based on the point, the electrocardiosignal is converted into the form of the picture, and then the target detection technology is applied to electrocardiosignal identification to realize the identification of the R wave position. Moreover, the electrocardiosignals are converted into pictures which are more in accordance with the diagnosis scene of doctors, and the target detection and the electrocardiogram are combined, so that the advantages are better exerted, and the R wave detection accuracy is improved.
Specifically, a desired ECG signal, i.e., an ECG signal, is first acquired. The embodiment of the invention does not require the quality of the acquired electrocardiosignals, preferably enriches the types of the acquired electrocardiosignals as much as possible, so that the sample data set contains the electrocardiosignals acquired under different conditions, the training complexity of the model is improved, the target detection model obtained by training can cope with various electrocardiograms, and the application range is wider in practical application.
Although the embodiment does not make a requirement on the quality of the electrocardiosignal and the recognition effect does not depend on the quality of the electrocardiosignal, the electrocardiosignal is processed before modeling, which is certainly beneficial to improving the recognition accuracy of the model. Therefore, a user can select whether to perform signal processing according to requirements, if the requirement on the identification accuracy rate is high, the signal processing is preferably performed, and if the requirement on the identification rate is high, the signal processing is preferably not performed. The signal processing includes baseline filtering, electromyographic interference, and the like.
After the electrocardiosignals are collected, an electrocardiogram is drawn according to the collected electrocardiosignals, so that the electrocardiosignals are visualized, on one hand, the electrocardiosignals are convenient to be identified from the aspect of image processing, and on the other hand, a doctor can conveniently and directly carry out R wave labeling on the visualized electrocardiogram.
After the electrocardiosignals are converted into electrocardiograms, marking and marking frames at the R wave positions on the electrocardiograms so as to obtain sample data for training. And training the target detection network by adopting the sample data to obtain a target detection model capable of identifying the R wave position in the electrocardiogram, thereby realizing the R wave identification of the electrocardiosignal. The target detection network can be realized by adopting any target detection deep learning network, such as Mask-RCNN, RCN, faster-RCNN and the like. The method has the advantages that good results are obtained in the recognition of the target frame in the image recognition, basic parameters and evaluation indexes are set in the recognition of the target frame at the R wave position, the electrocardiogram marked in the previous step is adopted and input into the model, and the model is trained.
After the target detection model is obtained through the target detection algorithm, the target frame at the R wave position of the electrocardiosignal to be detected can be detected according to the target detection model, and the R wave position can be obtained according to the target frame coordinates.
After the R wave is identified, the QRS wave position, the P wave position and the like can be further determined according to the R wave according to requirements.
The method carries out R wave position identification based on the target detection of the image, has lower requirement on the quality of the electrocardiosignal when carrying out R wave identification because the target detection has higher identification accuracy, and the R wave identification effect does not depend on a signal processing means seriously, thereby reducing the difficulty of a signal processing technology and improving the algorithm processing speed. When the method is applied, signal processing is not needed, so that the recognition rate is high, and the method can realize real-time recognition by collecting and recognizing.
Preferably, the method for drawing an electrocardiogram by collecting an electrocardiogram signal further comprises:
the image pixels of the electrocardiogram are increased.
The image pixels of the electrocardiogram are improved, the image quality of the electrocardiogram can be improved, and the detection accuracy of the target detection model can be further improved. Super-resolution image technologies such as RAISR can be adopted to convert low-resolution images into high-resolution images, so that image pixels are improved, and image quality is guaranteed.
Preferably, a marking frame is marked at the position of the R wave in the electrocardiogram, and the marking frame specifically comprises the following steps:
and taking the R wave position as a reference point, and marking the marking frame at a position with a fixed distance from the front to the back of the reference point.
And determining a marking frame by taking the R wave position as a reference point and fixing the distance between the front and the back of the reference point according to the R wave position marked by the doctor. The width, i.e. duration, of the label box is set according to the normal range of human heart rate. Specifically, the normal range of heart rate is generally 60-100 times in 1 minute, so that the duration of each time is 0.6-1.0 second, and the width of the labeling box is in the range of 0.6-1.0 second. This example takes 0.7 seconds. Therefore, the range of 0.35s is taken before and after the reference point respectively to be used as the range of the marking frame, and the marking frame is marked in the range of the marking frame to form the marking frame taking the R wave as the middle point.
Preferably, the target detection network is trained by using the sample data set to obtain a target detection model, which specifically comprises:
and training the target detection network by taking the electrocardiogram as input and the labeling box as output to obtain the target detection model.
Preferably, the method further comprises;
dividing the sample data set into two parts to obtain a first sample data set and a second sample data set;
training a target detection network by adopting the first sample data set to obtain the target detection model;
inputting the electrocardiogram in the second sample data set into the target detection model to obtain a prediction frame of the R wave, obtaining a prediction coordinate of the position of the R wave according to the prediction frame, and generating an R-R interval prediction sequence;
acquiring a labeling coordinate of the R wave position according to a labeling frame on the electrocardiogram of the second sample data set, and generating an R-R interval labeling sequence;
judging the prediction correctness of the R-R interval prediction sequence according to the R-R interval labeling sequence, and generating a judgment sequence of an R wave position;
taking the R-R interval prediction sequence as input, taking the judgment sequence as output, and training a regression model to obtain an R-wave correction model;
and identifying the R wave position of the electrocardiosignal to be detected by combining the target detection model and the R wave correction model.
The electrocardiogram with large noise has a large amount of convex waveforms, does not belong to R waves, but has similarity with the R waves in form, so that when R wave identification is carried out only based on target detection, the influence of noise is received, the problem of false identification still possibly occurs, and the identification precision is influenced. Therefore, in this embodiment, based on the target detection, the target detection result is further corrected by using a regression model training method.
The R-R interval can be calculated according to the R wave position in a longitudinal view of the complete electrocardiogram, so that the heart rate value of the section of the electrocardiosignal can be obtained. The human heart rate values have a certain range, even tachycardia can regularly follow, and the tachycardia is not randomly changed, so the heart rate values obtained by R wave positions with too close or too far intervals do not conform to the human heart rate range obtained by statistics at present, and the heart rate values should be deleted. This embodiment is just to correct the recognition result obtained by the target detection model by using this point. The specific method is that the sample data in the sample data set is divided into two parts, one part is used for training the target detection model, and the other part is input into the trained target detection model to obtain a prediction result, namely a prediction box of the R wave. And when the marking frame is marked, the boundary of the marking frame is away from the fixed distance of the R wave, and the position coordinate of the R wave is obtained. For example, in the present embodiment, the R-wave position is located at the center of the labeling frame, and therefore, the midpoint coordinate of the prediction frame of the R-wave in the time axis direction is taken as the prediction coordinate of the R-wave position, and the prediction coordinates of all R-waves of the entire electrocardiogram are integrated to obtain the R-R interval prediction sequence. By the same method, the labeled boxes in the sample data are utilized to obtain the R-R interval labeling sequence. And taking the R-R interval prediction sequence as input data, inputting a regression model for continuous variable prediction, obtaining a judgment sequence according to the R-R interval labeling sequence as output, training the regression model to obtain model parameters, and thus obtaining the R-wave correction model capable of correcting the R-R interval sequence. The regression model can be realized by linear regression, logistic regression, principal component regression and the like, and can also be realized by an RNN model and an RNN variant model in the deep learning field. RNN variant models include LSTM, GRU, and the like. This is preferably accomplished using RNN models as well as RNN variant models.
In the step, the label is a real R wave position sequence which is manually marked, and after a selected regression model method is determined, training is carried out to obtain a model for correcting the R wave position. And the R wave identification with higher precision can be realized by combining the R wave correction model and the target detection model. The R wave correction model can correct the detection result of the target detection model, eliminates the interference of R wave noise on R wave identification caused by strong myoelectric interference and the like, so that the R wave identification result is more accurate, and particularly has more effective identification effect on electrocardiosignal identification with strong interference. Meanwhile, the dependence of R wave identification on the signal processing effect is further reduced on the basis of an R wave correction model, and the R wave can still be accurately identified when the electrocardiosignal is not subjected to signal processing and large interference noise exists.
In the preferred embodiment, the strong interference of R wave-like noise on R wave identification is eliminated through the R wave distribution of the whole electrocardiogram and a regression model, so that a more accurate R wave position is obtained, the QRS wave position and related data are determined based on the more accurate R wave position, and a more accurate judgment basis is provided for the next step of disease identification. In the process, the method is not limited to the implementation by using a regression model, and the like, and the wrong R-wave position can be deleted by means of example segmentation and the like, and the deletion can also be manually performed according to the theory, but in view of accuracy, the implementation by using regression model training is more recommended.
Preferably, the prediction correctness of the R-R interval prediction sequence is judged according to the R-R interval labeling sequence, and a judgment sequence of R-wave positions is generated, specifically:
and setting the time when the R wave exists as 1 and the time when the R wave does not exist as 0 according to the R-R interval labeling sequence to obtain the judgment sequence.
For example, if the cardiac signal is a sequence including 50 sampling values, and the R-R interval prediction sequence of the cardiac signal predicted by the object detection model is {1,3,6,8,10,15,20,23,30}, it indicates that the cardiac signal has R waves at 1,3,6,8,10,15,20,23, and 30 sampling positions as a result of the prediction. And the R-R interval mark sequence corresponding to the electrocardiosignal is {1,6,15,20,30}, which indicates that the electrocardiosignal actually has R waves only at the 1 st, 6 th, 15 th, 20 th and 30 th sampling positions, namely the prediction result at the 3 rd, 8 th, 10 th and 23 th sampling positions is wrong, so that the judgment sequence is {1,0,1,0, 1}.
Preferably, the R-wave position of the electrocardiographic signal to be detected is identified by combining the target detection model and the R-wave correction model, and the method specifically comprises the following steps:
drawing an electrocardiogram to be detected according to the electrocardiosignals to be detected, inputting the electrocardiogram to be detected into the target detection model to obtain a target frame at the R-wave position of the electrocardiosignals to be detected, and generating an R-R interval sequence of the electrocardiosignals to be detected according to the target frame at the R-wave position;
inputting the R-R interval sequence into the R-wave correction model to obtain a corrected R-R interval sequence;
and obtaining the R wave position of the electrocardiosignal to be detected according to the corrected R-R interval sequence.
And correcting the prediction result of the target detection model through the R wave correction model, and deleting the position of the R wave with the recognition error to obtain a final result.
Preferably, the R-R interval sequence is input into the R-wave correction model to obtain a corrected R-R interval sequence, specifically:
inputting the R-R interval sequence into the R wave correction model to obtain the R wave probability at each position;
and screening out the positions where the R-wave probability is greater than the set probability to obtain the corrected R-R interval sequence.
The output of the regression model is the existence probability of the R wave at each moment, a set probability, for example 0.9, is set, and the positions of the moments which are greater than the set probability are screened out to obtain a corrected R-R interval sequence.
Example 2
Embodiment 2 of the present invention provides an R-wave recognition apparatus based on object detection, which includes a processor and a memory, where the memory stores a computer program, and when the computer program is executed by the processor, the R-wave recognition apparatus based on object detection as provided in embodiment 1 is implemented.
The R-wave identification device based on target detection provided in the embodiments of the present invention is used to implement an R-wave identification method based on target detection, and therefore, the R-wave identification device based on target detection also has the technical effects that the R-wave identification method based on target detection has, and details thereof are not repeated herein.
Example 3
Embodiment 3 of the present invention provides a computer storage medium having stored thereon a computer program that, when executed by a processor, implements the R-wave identifying method based on object detection provided in embodiment 1.
The computer storage medium provided in the embodiments of the present invention is used to implement an R-wave identification method based on target detection, and therefore, the computer storage medium also has the technical effects of the R-wave identification method based on target detection, and details thereof are not repeated herein.
The above-described embodiments of the present invention should not be construed as limiting the scope of the present invention. Any other corresponding changes and modifications made according to the technical idea of the present invention should be included in the protection scope of the claims of the present invention.
Claims (9)
1. An R wave identification method based on target detection is characterized by comprising the following steps:
acquiring electrocardiosignals, and drawing an electrocardiogram according to the electrocardiosignals;
marking a marking frame at the position of the R wave in the electrocardiogram to obtain a sample data set;
training a target detection network by adopting the sample data set to obtain a target detection model;
identifying the R wave position of the electrocardiosignal to be detected according to the target detection model;
also comprises the following steps;
dividing the sample data set into two parts to obtain a first sample data set and a second sample data set;
training a target detection network by adopting the first sample data set to obtain the target detection model;
inputting the electrocardiogram in the second sample data set into the target detection model to obtain a prediction frame of R waves, acquiring a prediction coordinate of the position of the R waves according to the prediction frame, and generating an R-R interval prediction sequence;
acquiring a labeling coordinate of the R wave position according to a labeling frame on the electrocardiogram of the second sample data set, and generating an R-R interval labeling sequence;
judging the prediction correctness of the R-R interval prediction sequence according to the R-R interval labeling sequence, and generating a judgment sequence of an R wave position;
taking the R-R interval prediction sequence as input and the judgment sequence as output, and training a regression model to obtain an R wave correction model;
and identifying the R wave position of the electrocardiosignal to be detected by combining the target detection model and the R wave correction model.
2. The R-wave identifying method based on object detection according to claim 1, wherein an electrocardiogram is drawn by acquiring an electrocardiogram signal, and further comprising:
the image pixels of the electrocardiogram are enhanced.
3. The R-wave identification method based on target detection according to claim 1, wherein a labeling frame is marked at the R-wave position in the electrocardiogram, specifically:
and taking the R wave position as a reference point, and marking the marking frame at a position with a fixed distance from the front to the back of the reference point.
4. The R-wave recognition method based on target detection according to claim 1, wherein the target detection network is trained by using the sample data set to obtain a target detection model, specifically:
and training the target detection network by taking the electrocardiogram as input and the labeling box as output to obtain the target detection model.
5. The target detection-based R-wave identification method according to claim 1, wherein the prediction correctness of the R-R interval prediction sequence is judged according to the R-R interval labeling sequence, and a judgment sequence of R-wave positions is generated, specifically:
and judging whether the R wave positions in the R-R interval prediction sequence are predicted correctly or not according to the R-R interval labeling sequence, setting the positions predicted correctly in the R-R interval prediction sequence as '1', and setting the positions predicted incorrectly as '0', so as to obtain the judgment sequence.
6. The R-wave identification method based on target detection according to claim 1, wherein the R-wave position of the electrocardiographic signal to be detected is identified by combining the target detection model and the R-wave correction model, specifically:
drawing an electrocardiogram to be detected according to the electrocardiosignals to be detected, inputting the electrocardiogram to be detected into the target detection model to obtain a target frame at the R-wave position of the electrocardiosignals to be detected, and generating an R-R interval sequence of the electrocardiosignals to be detected according to the target frame at the R-wave position;
inputting the R-R interval sequence into the R-wave correction model to obtain a corrected R-R interval sequence;
and obtaining the R wave position of the electrocardiosignal to be detected according to the corrected R-R interval sequence.
7. The target detection-based R-wave identification method of claim 6, wherein the R-R interval sequence is input into the R-wave correction model to obtain a corrected R-R interval sequence, specifically:
inputting the R-R interval sequence into the R wave correction model to obtain the R wave probability at each position;
and screening out the positions where the R-wave probability is greater than the set probability to obtain the corrected R-R interval sequence.
8. An R-wave recognition apparatus based on object detection, comprising a processor and a memory, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements the R-wave recognition method based on object detection according to any one of claims 1 to 7.
9. A computer storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the R-wave identification method based on object detection according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010596685.6A CN111803061B (en) | 2020-06-28 | 2020-06-28 | R wave identification method and device based on target detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010596685.6A CN111803061B (en) | 2020-06-28 | 2020-06-28 | R wave identification method and device based on target detection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111803061A CN111803061A (en) | 2020-10-23 |
| CN111803061B true CN111803061B (en) | 2022-12-30 |
Family
ID=72855425
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010596685.6A Active CN111803061B (en) | 2020-06-28 | 2020-06-28 | R wave identification method and device based on target detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111803061B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115770048A (en) * | 2021-09-08 | 2023-03-10 | 上海联影医疗科技股份有限公司 | Electrocardiosignal processing method and device, computer equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003000561A (en) * | 2001-06-18 | 2003-01-07 | Canon Inc | R wave recognizing method, r-r interval measuring method, heart rate measuring method, r-r interval measuring device and heart rate measuring device |
| CN108309284A (en) * | 2018-03-15 | 2018-07-24 | 广州视源电子科技股份有限公司 | ECG T wave end-point detection method and device |
| CN108464827A (en) * | 2018-03-08 | 2018-08-31 | 四川大学 | It is a kind of it is Weakly supervised under electrocardio image-recognizing method |
| CN110916646A (en) * | 2019-12-10 | 2020-03-27 | 电子科技大学 | QRS wave detection method based on FASTER-RCNN |
| CN111110228A (en) * | 2020-01-17 | 2020-05-08 | 武汉中旗生物医疗电子有限公司 | Electrocardiosignal R wave detection method and device |
-
2020
- 2020-06-28 CN CN202010596685.6A patent/CN111803061B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003000561A (en) * | 2001-06-18 | 2003-01-07 | Canon Inc | R wave recognizing method, r-r interval measuring method, heart rate measuring method, r-r interval measuring device and heart rate measuring device |
| CN108464827A (en) * | 2018-03-08 | 2018-08-31 | 四川大学 | It is a kind of it is Weakly supervised under electrocardio image-recognizing method |
| CN108309284A (en) * | 2018-03-15 | 2018-07-24 | 广州视源电子科技股份有限公司 | ECG T wave end-point detection method and device |
| CN110916646A (en) * | 2019-12-10 | 2020-03-27 | 电子科技大学 | QRS wave detection method based on FASTER-RCNN |
| CN111110228A (en) * | 2020-01-17 | 2020-05-08 | 武汉中旗生物医疗电子有限公司 | Electrocardiosignal R wave detection method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111803061A (en) | 2020-10-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107203692B (en) | Electrocardio data digital signal processing method based on deep convolutional neural network | |
| Sayadi et al. | A model-based Bayesian framework for ECG beat segmentation | |
| CN109758145B (en) | Automatic sleep staging method based on electroencephalogram causal relationship | |
| CN108294745A (en) | P waves, T wave start-stop point detecting methods and system in multi-lead ECG signal | |
| CN111110228A (en) | Electrocardiosignal R wave detection method and device | |
| CN113057648A (en) | ECG signal classification method based on composite LSTM structure | |
| CN113080996B (en) | Electrocardiogram analysis method and device based on target detection | |
| CN111053552B (en) | QRS wave detection method based on deep learning | |
| CN111803061B (en) | R wave identification method and device based on target detection | |
| CN112396565A (en) | Method and system for enhancing and segmenting blood vessels of images and videos of venipuncture robot | |
| CN116012568A (en) | System for acquiring cardiac rhythm information through photographing electrocardiogram | |
| CN109770891B (en) | Electrocardiosignal preprocessing method and preprocessing device | |
| Boucheham et al. | Piecewise linear correction of ECG baseline wander: a curve simplification approach | |
| CN111466905B (en) | Electrocardiographic waveform extraction method based on bidirectional communication | |
| Ganesh et al. | Combining optical character recognition with paper ECG digitization | |
| US10043269B2 (en) | Medical image processing apparatus, medical image diagnostic apparatus, medical image processing method, and ultrasonic image processing method | |
| CN110811608A (en) | Atrial fibrillation monitoring method based on ECG (ECG) signals | |
| CN110916645A (en) | QRS wave identification method combining wavelet transformation and image segmentation network | |
| CN114176602B (en) | Method for simultaneously positioning electrocardiograph P wave, QRS wave and T wave based on deep learning multi-target detection | |
| CN108836312B (en) | Clutter rejection method and system based on artificial intelligence | |
| CN114119798A (en) | Radiographic image processing method and device based on respiratory motion | |
| CN116135146A (en) | Dynamic electrocardiograph analysis method and device and medical equipment | |
| EP3806731A1 (en) | Predicting atrial fibrillation or stroke using p-wave analysis | |
| CN111832586A (en) | Deep learning data preprocessing method and device and training system | |
| CN111685754B (en) | Heart rate calculation method and system for wearable ECG acquisition equipment |
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 |