CN113537214A - Automatic phase singularity identification method based on fast R-CNN and SRGAN - Google Patents
Automatic phase singularity identification method based on fast R-CNN and SRGAN Download PDFInfo
- Publication number
- CN113537214A CN113537214A CN202110804112.2A CN202110804112A CN113537214A CN 113537214 A CN113537214 A CN 113537214A CN 202110804112 A CN202110804112 A CN 202110804112A CN 113537214 A CN113537214 A CN 113537214A
- Authority
- CN
- China
- Prior art keywords
- network
- image
- srgan
- phase singularity
- cnn
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000012549 training Methods 0.000 claims description 19
- 238000013527 convolutional neural network Methods 0.000 claims description 10
- 238000010586 diagram Methods 0.000 claims description 7
- 238000002372 labelling Methods 0.000 claims description 7
- 230000001902 propagating effect Effects 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 230000004936 stimulating effect Effects 0.000 claims description 6
- BSYNRYMUTXBXSQ-UHFFFAOYSA-N Aspirin Chemical compound CC(=O)OC1=CC=CC=C1C(O)=O BSYNRYMUTXBXSQ-UHFFFAOYSA-N 0.000 claims description 5
- 238000011176 pooling Methods 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 5
- 238000012795 verification Methods 0.000 claims description 5
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 230000002107 myocardial effect Effects 0.000 claims description 4
- 230000036982 action potential Effects 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 3
- 238000009792 diffusion process Methods 0.000 claims description 3
- 230000036279 refractory period Effects 0.000 claims description 3
- 230000005764 inhibitory process Effects 0.000 claims description 2
- 230000003387 muscular Effects 0.000 claims description 2
- 238000012545 processing Methods 0.000 claims description 2
- 238000001514 detection method Methods 0.000 abstract description 10
- 230000006870 function Effects 0.000 description 10
- 238000007674 radiofrequency ablation Methods 0.000 description 5
- 210000001519 tissue Anatomy 0.000 description 4
- 206010003658 Atrial Fibrillation Diseases 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000001746 atrial effect Effects 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 208000000059 Dyspnea Diseases 0.000 description 1
- 206010013975 Dyspnoeas Diseases 0.000 description 1
- 206010019280 Heart failures Diseases 0.000 description 1
- 108020004566 Transfer RNA Proteins 0.000 description 1
- 238000002679 ablation Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008602 contraction Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 210000005003 heart tissue Anatomy 0.000 description 1
- 210000003205 muscle Anatomy 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 238000002560 therapeutic procedure Methods 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a method for automatically identifying a phase singularity based on fast R-CNN and SRGAN, and provides an fast R-CNN algorithm in the field of target detection as an automatic detection model of the phase singularity, and an anti-network SRGAN is generated by combining super-resolution to improve the spatial resolution of the phase singularity and better improve the identification precision of the phase singularity. The invention has great significance for automatically identifying the phase singularity of the ECG image acquired by multiple electrodes.
Description
Technical Field
The invention relates to the technical field of artificial intelligence medical image aided diagnosis, in particular to a method for automatically identifying a phase singularity based on fast R-CNN and SRGAN.
Background
Atrial fibrillation leads to increased heart rate and ineffective atrial contraction, which can lead to patient dyspnea and severe heart failure or death. Currently, radio frequency ablation is the first therapy for atrial defibrillation. The radiofrequency ablation technology cures atrial fibrillation by eliminating diseased myocardial cells, and in order to improve the treatment effect of the radiofrequency ablation technology, the determination of an ablation target point of the radiofrequency ablation is crucial. While the phase singularity of the helicon wave produced in atrial fibrillation is often the diseased tissue in the heart tissue. Therefore, determining the phase singularity of the helicon wave allows the determination of the target point for the RF ablation.
The existing algorithms for detecting the phase singularities are all used for identifying the phase singularities manually, and because the spatial resolution of the collected phase singularity images is too low, the identification accuracy is not high. During treatment, a physician acquires a patient's ECG signal that records the spatiotemporal distribution of electrophysiological activity throughout the interior wall of the heart. Thus, the acquired ECG signal can be regarded as an image signal in fact, and the detection of phase singularities based on these signals can be further regarded as an object detection problem in the field of computer vision.
Due to the development of deep neural networks in recent years, target detection based on regional convolutional neural networks (R-CNN) has been considered as an ideal method for assisting medical diagnosis. The physician can use the automatic detection results of the medical images to gain further insight into the specific pathological characteristics of the patient and make more accurate diagnoses.
Disclosure of Invention
The invention aims to overcome the defects and provide the automatic phase singularity identification method based on the fast R-CNN and the SRGAN, and the automatic phase singularity identification method based on the fast R-CNN and the SRGAN based on deep learning is adopted to replace the traditional manual identification method, so that the identification precision of the phase singularity can be better improved. The invention has great significance for automatically identifying the phase singularity of the ECG image acquired by multiple electrodes.
The invention achieves the aim through the following technical scheme: a method for automatically identifying a phase singularity based on fast R-CNN and SRGAN comprises the following steps:
(1) constructing a heart model, and simulating heart spiral wave disorder to obtain a phase singularity image;
(2) obtaining a plurality of phase singular point images in a data expansion mode; carrying out category labeling on the image to obtain a plurality of pieces of training data with labels, and making the training data into a phase singularity data set;
(3) establishing a Faster R-CNN network model, and inputting a phase singular point data set into a Faster R-CNN neural network for training;
(4) constructing an SRGAN network model, and training the SRGAN network model by using a phase singular point data set;
(5) and inputting the phase singularity image to be identified into a trained SRGAN network model, improving the spatial resolution of the phase singularity image, inputting the image with improved resolution into an Faster R-CNN network model, and outputting a result of the position of the phase singularity.
Preferably, the step (1) is specifically as follows:
(1.1) constructing a myocardial tissue by the following formula:
wherein V represents action potential in mV; t represents time in ms; cm＝1.0μF/cm2Is an ideal capacitance between films, D is 0.001cm2Perms is diffusion current coefficient, total ion current IionIs determined by an ion gate, and a gate control variable of the ion gate is obtained as a solution of a nonlinear ordinary differential equation coupled system;
(1.2) after a stimulating electrode is added on the left side of the muscular tissue, exciting a row of plane waves propagating from left to right, and applying the stimulating electrode on the bottom part vertical to the plane waves when the plane waves are conducted to the central position so as to initiate a second plane wave propagating upwards;
(1.3) the second plane wave enters the end of the refractory period of the first plane and forms a sharp curvature, thereby forming a helicon wave.
Preferably, the step (2) of creating the phase singularity data set comprises: processing and labeling the phase singularity image according to a VOC2007 data set format; adopting a plurality of conventional data expansion modes, wherein the conventional data expansion modes comprise but are not limited to rotation, brightness, color and noise, and finally obtaining a plurality of spiral wave spot patterns; marking the image by using a marking tool to generate a target coordinate information file of the image; wherein, the data set is divided into three parts, namely a training set, a verification set and a testing machine; and randomly distributing the marked images to a training set, a verification set and a test set by using a python program, wherein the proportion is 8: 1: 1.
preferably, the procedure for establishing the Faster R-CNN network model in the step (3) is as follows:
(3.1) building a convolutional neural network, adopting ResNet50 as a backbone network of Faster R-CNN for extracting the characteristics of the phase singularities, and obtaining a characteristic diagram through a ResNet50 network, wherein the characteristic diagram is used for a subsequent RPN network and a full connection layer;
(3.2) constructing an RPN network for recommending a candidate region, and selecting a candidate frame with a higher score by adopting a non-maximum value inhibition method;
(3.3) building an interested area pooling layer, and converting the inputs with different sizes into the outputs with fixed lengths;
and (3.4) classifying and regressing by using a softmax function, and outputting the class to which the candidate region belongs and the accurate position of the candidate region in the image.
Preferably, the method for constructing the SRGAN network model in step (4) is as follows:
(4.1) building a generating network of the SRGAN network, wherein the generating network consists of three parts, namely a convolutional layer and a RELU function, then the generating network passes through a residual error network and a residual error edge and finally enters an up-sampling part, the length and the width are amplified, and after two times of up-sampling, the original length and the width are changed into 4 times, so that the resolution is improved;
(4.2) constructing a discrimination network of the SRGAN network, wherein the discrimination network consists of a convolution layer, a LeakyRELU function and BN standardization which are repeated continuously, and the discrimination network has the function of obtaining the probability of a predicted natural image and is obtained by two full-connection layers and a final sigmoid activation function.
The invention has the beneficial effects that: the invention provides a Faster R-CNN algorithm in the field of target detection as an automatic detection model of phase singularities, and a countermeasure network SRGAN generated by combining super-resolution is used for improving the spatial resolution of the phase singularities and better improving the identification precision of the phase singularities. The invention has great significance for automatically identifying the phase singularity of the ECG image acquired by multiple electrodes.
Drawings
FIG. 1 is a flow chart of the detection of the present invention;
FIG. 2 is a simulated phase singularity image of the present invention;
FIG. 3 is a Faster R-CNN network model structure of the present invention;
fig. 4 is a resulting network structure of the SRRAN network of the present invention;
FIG. 5 is a discriminant network structure of an SRRAN network according to the present invention;
FIG. 6 is a phase singularity image with improved resolution by the SRGAN network model of the present invention;
FIG. 7 is a graph showing the effect of the fast R-CNN of the present invention.
Detailed Description
The invention will be further described with reference to specific examples, but the scope of the invention is not limited thereto:
example (b): a method for automatically identifying a phase singularity based on fast R-CNN and SRGAN is disclosed, wherein a detection process is shown in FIG. 1, and the method specifically comprises the following steps:
step 1: constructing a heart model, and simulating heart spiral wave disorder to obtain a phase singularity image; a typical sample of the image is shown in figure 2. The method comprises the following specific steps:
step 1.1: a piece of myocardial tissue was constructed by equation 1, where V represents action potential (unit: mV), t represents time (unit: ms), and Cm＝1.0μF/cm2Is an ideal capacitance between films, D is 0.001cm2Perms is diffusion current coefficient, total ion current IionIs determined by an ion gate whose gating variables are obtained as solutions to a coupled system of nonlinear ordinary differential equations.
Step 1.2: when a stimulating electrode is added on the left side of the muscle tissue, a row of plane waves propagating from left to right are excited, and when the plane waves are conducted to the central position, the stimulating electrode is applied on the bottom part perpendicular to the plane waves, so that a second plane wave propagating upwards is triggered.
Step 1.3: the second plane wave enters the end of the refractory period of the first plane and forms a sharp curvature. This results in the interruption of a second plane wave which begins to curl to the right, forming a helicon wave.
Step 2: through several conventional data expansion modes, a plurality of phase singularity images are obtained. And carrying out category labeling on the images to obtain a plurality of pieces of training data with labels, and making the training data into a phase singularity data set. The method specifically comprises the following steps: the invention processes and labels the phase singularity images according to the VOC2007 data set format. In order to avoid the situation of unobvious characteristics, the invention carries out data expansion operation, and adopts several conventional modes such as rotation, brightness, color, noise and the like to finally obtain a plurality of spiral wave spot patterns. And (4) labeling the image by using a labeling tool to generate a target coordinate information file of the image. The data set is divided into three parts, namely a training set, a validation set and a tester. And randomly distributing the marked images to a training set, a verification set and a test set by using a python program, wherein the proportion is 8: 1: 1.
and step 3: establishing a Faster R-CNN network model, and inputting a phase singular point data set into a Faster R-CNN neural network for training; the method comprises the following specific steps:
step 3.1: a convolutional neural network is built, ResNet50 is used as a backbone network of Faster R-CNN for extracting the characteristics of phase singularities, and a model structure diagram is shown in figure 3, wherein a specific module of ResNet50 is in Conv layers. A profile is obtained across the ResNet50 network for subsequent RPN networks and full connectivity layers.
Step 3.2: and constructing an RPN (resilient packet network) for recommending the candidate area, and selecting a candidate frame with a higher score by adopting a Non-Maximum Suppression (NMS) method.
Step 3.3: building a Region of Interest pooling layer (Region of Interest pool-ing, RoI pooling), and converting inputs of different sizes into outputs of fixed length;
step 3.4: and classifying and regressing by using a softmax function, and outputting the class to which the candidate region belongs and the accurate position of the candidate region in the image.
And 4, step 4: and constructing an SRGAN network model, and training the SRGAN network model by using the phase singularity data set. The generated network structure diagram of the srna network is shown in fig. 4, and the judged network structure diagram is shown in fig. 5, which is specifically as follows:
step 4.1: the method comprises the steps of building a generating network of the SRGAN network, wherein the generating network consists of three parts, namely a convolutional layer and a RELU function, passing through a residual error network and a residual error edge, entering an up-sampling part, amplifying the length and the width, and changing the length and the width into 4 times of the original length and the width after two times of up-sampling, so that the resolution is improved.
Step 4.2: and (3) constructing a discrimination network of the SRGAN network, wherein the discrimination network consists of a convolution layer, a LeakyRELU function and BN standardization which are repeated continuously, and the discrimination network has the function of obtaining the probability of a predicted natural image and is obtained by two full-connection layers and a final sigmoid activation function.
And 5: after the training is completed, inputting the phase singularity image to be identified into the SRGAN network model, and improving the spatial resolution of the phase singularity image, as shown in fig. 6; inputting the image with the improved resolution into a Faster R-CNN network model, and outputting a result of the position of the obtained phase singularity; the resulting recognition map is shown in fig. 7.
While the invention has been described in connection with specific embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (5)
1. A method for automatically identifying a phase singularity based on fast R-CNN and SRGAN is characterized by comprising the following steps:
(1) constructing a heart model, and simulating heart spiral wave disorder to obtain a phase singularity image;
(2) obtaining a plurality of phase singular point images in a data expansion mode; carrying out category labeling on the image to obtain a plurality of pieces of training data with labels, and making the training data into a phase singularity data set;
(3) establishing a Faster R-CNN network model, and inputting a phase singular point data set into a Faster R-CNN neural network for training;
(4) constructing an SRGAN network model, and training the SRGAN network model by using a phase singular point data set;
(5) and inputting the phase singularity image to be identified into a trained SRGAN network model, improving the spatial resolution of the phase singularity image, inputting the image with improved resolution into an Faster R-CNN network model, and outputting a result of the position of the phase singularity.
2. The method of claim 1, wherein the method comprises the following steps: the step (1) is specifically as follows:
(1.1) constructing a myocardial tissue by the following formula:
wherein V represents action potential in mV; t represents time in ms; cm＝1.0μF/cm2Is an ideal capacitance between films, D is 0.001cm2Perms is diffusion current coefficient, total ion current IionIs determined by an ion gate, and a gate control variable of the ion gate is obtained as a solution of a nonlinear ordinary differential equation coupled system;
(1.2) after a stimulating electrode is added on the left side of the muscular tissue, exciting a row of plane waves propagating from left to right, and applying the stimulating electrode on the bottom part vertical to the plane waves when the plane waves are conducted to the central position so as to initiate a second plane wave propagating upwards;
(1.3) the second plane wave enters the end of the refractory period of the first plane and forms a sharp curvature, thereby forming a helicon wave.
3. The method of claim 1, wherein the method comprises the following steps: the process for making the phase singularity data set in the step (2) is as follows: processing and labeling the phase singularity image according to a VOC2007 data set format; adopting a plurality of conventional data expansion modes, wherein the conventional data expansion modes comprise but are not limited to rotation, brightness, color and noise, and finally obtaining a plurality of spiral wave spot patterns; marking the image by using a marking tool to generate a target coordinate information file of the image; wherein, the data set is divided into three parts, namely a training set, a verification set and a testing machine; and randomly distributing the marked images to a training set, a verification set and a test set by using a python program, wherein the proportion is 8: 1: 1.
4. the method of claim 1, wherein the method comprises the following steps: the process for establishing the Faster R-CNN network model in the step (3) is as follows:
(3.1) building a convolutional neural network, adopting ResNet50 as a backbone network of Faster R-CNN for extracting the characteristics of the phase singularities, and obtaining a characteristic diagram through a ResNet50 network, wherein the characteristic diagram is used for a subsequent RPN network and a full connection layer;
(3.2) constructing an RPN network for recommending a candidate region, and selecting a candidate frame with a higher score by adopting a non-maximum value inhibition method;
(3.3) building an interested area pooling layer, and converting the inputs with different sizes into the outputs with fixed lengths;
and (3.4) classifying and regressing by using a softmax function, and outputting the class to which the candidate region belongs and the accurate position of the candidate region in the image.
5. The method of claim 1, wherein the method comprises the following steps: the method for constructing the SRGAN network model in the step (4) is as follows:
(4.1) building a generating network of the SRGAN network, wherein the generating network consists of three parts, namely a convolutional layer and a RELU function, then the generating network passes through a residual error network and a residual error edge and finally enters an up-sampling part, the length and the width are amplified, and after two times of up-sampling, the original length and the width are changed into 4 times, so that the resolution is improved;
(4.2) constructing a discrimination network of the SRGAN network, wherein the discrimination network consists of a convolution layer, a LeakyRELU function and BN standardization which are repeated continuously, and the discrimination network has the function of obtaining the probability of a predicted natural image and is obtained by two full-connection layers and a final sigmoid activation function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110804112.2A CN113537214A (en) | 2021-07-16 | 2021-07-16 | Automatic phase singularity identification method based on fast R-CNN and SRGAN |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110804112.2A CN113537214A (en) | 2021-07-16 | 2021-07-16 | Automatic phase singularity identification method based on fast R-CNN and SRGAN |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113537214A true CN113537214A (en) | 2021-10-22 |
Family
ID=78128283
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110804112.2A Pending CN113537214A (en) | 2021-07-16 | 2021-07-16 | Automatic phase singularity identification method based on fast R-CNN and SRGAN |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113537214A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109903839A (en) * | 2019-02-22 | 2019-06-18 | 武汉凯德维斯生物技术有限公司 | A kind of medical imaging diagosis system based on cloud platform |
| CN110728658A (en) * | 2019-09-16 | 2020-01-24 | 武汉大学 | High-resolution remote sensing image weak target detection method based on deep learning |
| CN112347895A (en) * | 2020-11-02 | 2021-02-09 | 北京观微科技有限公司 | Ship remote sensing target detection method based on boundary optimization neural network |
-
2021
- 2021-07-16 CN CN202110804112.2A patent/CN113537214A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109903839A (en) * | 2019-02-22 | 2019-06-18 | 武汉凯德维斯生物技术有限公司 | A kind of medical imaging diagosis system based on cloud platform |
| CN110728658A (en) * | 2019-09-16 | 2020-01-24 | 武汉大学 | High-resolution remote sensing image weak target detection method based on deep learning |
| CN112347895A (en) * | 2020-11-02 | 2021-02-09 | 北京观微科技有限公司 | Ship remote sensing target detection method based on boundary optimization neural network |
Non-Patent Citations (4)
| Title |
|---|
| MERVE TAS¸等: "《Super resolution convolutional neural network based pre-processing for automatic polyp detection in colonoscopy images》", 《COMPUTERS AND ELECTRICAL ENGINEERING》, vol. 90, pages 1 - 9 * |
| 吴洋洋: "《生成对抗网络的超声图像超分辨率重建研究》", 《中国优秀硕士学位论文全文数据库》, no. 12, pages 17 - 21 * |
| 徐志敏: "《基于卷积神经网络的车型识别技术研究》", 《中国优秀硕士学位论文全文数据库》, pages 40 - 54 * |
| 金印彬 等: "《心脏电生理的计算机仿真方法研究》", 《生物医学工程学杂志》, no. 02, pages 419 - 423 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cantwell et al. | Rethinking multiscale cardiac electrophysiology with machine learning and predictive modelling | |
| US20200352652A1 (en) | Systems and methods for improving cardiac ablation procedures | |
| CN109544598A (en) | Method for tracking target, device and readable storage medium storing program for executing | |
| US20210386355A1 (en) | System and method to detect stable arrhythmia heartbeat and to calculate and detect cardiac mapping annotations | |
| CN112784879A (en) | Medical image segmentation or classification method based on small sample domain self-adaption | |
| Fan et al. | Accurate retinal vessel segmentation via octave convolution neural network | |
| US20210378579A1 (en) | Local noise identification using coherent algorithm | |
| CN112426160A (en) | Electrocardiosignal type identification method and device | |
| CN110731773A (en) | abnormal electrocardiogram screening method based on fusion of global and local depth features of electrocardiogram | |
| US20220000410A1 (en) | Mapping efficiency by suggesting map point's location | |
| US20230225663A1 (en) | Method for predicting multi-type electrocardiogram heart rhythms based on graph convolution | |
| Cheng et al. | Adgan: A scalable gan-based architecture for image anomaly detection | |
| US20220008126A1 (en) | Optimized ablation for persistent atrial fibrillation | |
| CN114492519A (en) | Lung ultrasonic special sign B-line identification and classification method based on ultrasonic echo radio frequency signals | |
| US20210369174A1 (en) | Automatic detection of cardiac structures in cardiac mapping | |
| EP4113528A1 (en) | System and method to determine the location of a catheter | |
| CN113537214A (en) | Automatic phase singularity identification method based on fast R-CNN and SRGAN | |
| US20220007989A1 (en) | System and method to determine the location of a catheter | |
| RU2461877C1 (en) | Method for neural network analysis of cardial condition | |
| CN115719648A (en) | Anatomically correct reconstruction of atria | |
| CN113298773A (en) | Heart view identification and left ventricle detection device and system based on deep learning | |
| CN113729728A (en) | Automatic detection of cardiac structures in cardiac mapping | |
| CN113298754A (en) | Detection method for contour line control points of prostate tissue | |
| US20220338939A1 (en) | System and method to determine the location of a catheter | |
| CN110916646A (en) | QRS wave detection method based on FASTER-RCNN |
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 |