Atrial fibrillation detection method, device and equipment under high-field-intensity magnetic resonance and storage medium
By extracting the RR interval and P wave matrix signal characteristics of the electrocardiogram signal under a high field and strong magnetic resonance environment, and fusion classification, the problem of poor detection accuracy is solved, and the accuracy and robustness of atrial fibrillation detection are improved.
Patent Information
- Application Number
- CN202410952212.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-16
- Publication Date
- 2024-11-01
AI Technical Summary
In a high-field and strong magnetic resonance environment, the detection accuracy of atrial fibrillation is poor and the robustness is poor, making it difficult to accurately identify the amplitude of the P and F waves and the frequency is too small, which leads to inaccurate electrocardiogram detection results.
By acquiring the electrocardiogram signal, positioning the R wave and P wave, extracting the matrix signals of the RR interval and P wave, extracting traditional and deep features, and performing fusion classification, the atrial fibrillation alarm results are determined using DCCA algorithm and SVM classifier, and enhancing Accuracy and robustness of the detection.
It improves the accuracy of atrial fibrillation detection and the accuracy of detection results, is suitable for monitoring in high-field and strong magnetic resonance environments, and enhances the robustness of the algorithm.
Smart Images

Figure CN118873152A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of medical monitoring technology, and in particular to a method, device, equipment and storage medium for detecting atrial fibrillation under high-field magnetic resonance. Background Art
[0002] Cardiovascular disease is the focus of human attention worldwide because of its extremely high mortality rate. Electrocardiogram is an important means of detecting cardiovascular disease. Electrocardiogram is an important tool for detecting heart activity and diagnosing cardiovascular disease. In certain special cases, it is necessary to continuously monitor the patient's electrocardiogram to detect arrhythmias. There are mature equipment for this in daily environments, but continuous monitoring of the patient's electrocardiogram in a high-field magnetic resonance environment requires further research, because during the magnetic resonance imaging process, it is also necessary to monitor the body's electrocardiogram abnormalities, which is a very critical step.
[0003] Atrial fibrillation is a very common heart disease with cardiac dysfunction. When atrial fibrillation is detected based on an electrocardiogram under high-field magnetic resonance conditions, the amplitude of the P wave and the F wave is too small and the frequency is too fast, which can easily lead to inaccurate test results, poor detection accuracy, and poor robustness. Summary of the invention
[0004] Based on this, it is necessary to propose a method, device, equipment and storage medium for detecting atrial fibrillation under high-field strength magnetic resonance imaging to address the above problems.
[0005] To achieve the above-mentioned object, the present application provides, in a first aspect, a method for detecting atrial fibrillation under high-field magnetic resonance imaging, the method comprising:
[0006] Acquire electrocardiogram signals;
[0007] Positioning feature points of the electrocardiogram signal to obtain R waves and P waves;
[0008] According to the R wave and the P wave, a matrix signal of the RR interval and a matrix signal of the P wave are obtained respectively;
[0009] Extracting features from the matrix signal of the RR interval to obtain traditional features and deep features of the RR interval signal;
[0010] fusing and classifying the traditional features and deep features of the RR interval signal to obtain feature classification of the RR interval signal;
[0011] Extracting features from the matrix signal of the P wave to obtain traditional features and deep features of the P wave signal;
[0012] fusing and classifying the traditional features and deep features of the P-wave signal to obtain feature classification of the P-wave signal;
[0013] The atrial fibrillation alarm result is determined based on the feature classification of the RR interval signal and the feature classification of the P wave signal.
[0014] In some embodiments, the performing feature point positioning on the electrocardiogram signal to obtain the R wave and the P wave specifically includes:
[0015] Preprocessing the electrocardiogram signal;
[0016] Positioning the preprocessed electrocardiogram signal by using a PT algorithm to obtain a position set of the R wave in the electrocardiogram signal;
[0017] Obtaining an HRV signal of the RR interval according to the R wave position set;
[0018] The P wave is determined based on the HRV signal of the RR interval.
[0019] In some embodiments, the preprocessing of the electrocardiogram signal specifically includes:
[0020] Performing baseline processing on the electrocardiogram signal to obtain an electrocardiogram signal baseline,
[0021] The electrocardiogram signal baseline is removed from the original electrocardiogram signal, and then the high-frequency power frequency noise is removed to obtain a clean electrocardiogram signal, and the clean electrocardiogram signal is the preprocessed electrocardiogram signal.
[0022] In some implementations, extracting features from the matrix signal of the RR interval to obtain traditional features and deep features of the RR interval signal specifically includes:
[0023] Extracting features from the matrix signal of the RR interval to obtain ratio features, Poincare map features, and entropy features among traditional features;
[0024] The matrix signal of the RR interval is extracted with a neural network to obtain deep features.
[0025] In some embodiments, the fusing and classifying the traditional features and the deep features of the RR interval signal to obtain the feature classification of the RR interval signal specifically includes:
[0026] The traditional features and deep features of the RR interval signal are connected in series to form a first feature matrix;
[0027] Obtaining a first optimal projection matrix of the first feature matrix according to a class representativeness correlation analysis (DCCA) algorithm;
[0028] Obtaining a first feature projection on the X-axis and the Y-axis according to the first optimal projection matrix;
[0029] Splicing the first feature projections on the X-axis and the Y-axis to obtain the fusion feature of the RR interval signal;
[0030] The fusion features of the RR interval signal are put into the SVM classifier for classification to obtain atrial fibrillation warning information and non-atrial fibrillation warning information of the RR interval signal.
[0031] In some embodiments, extracting features from the matrix signal of the P wave to obtain traditional features and depth features of the P wave signal specifically includes:
[0032] Extracting features from the matrix signal of the P wave to obtain the width of the P wave detection range and the maximum value within the P wave detection range in traditional features;
[0033] The deep features are obtained by extracting features from the matrix signal of the P wave through a neural network.
[0034] In some embodiments, the traditional features and deep features of the P wave signal are fused and classified to obtain feature classification of the P wave signal, specifically including:
[0035] The traditional features and the deep features of the P-wave signal are connected in series to form a second feature matrix;
[0036] Obtaining a second optimal projection matrix of the second feature matrix according to a class representativeness correlation analysis (DCCA) algorithm;
[0037] Obtaining second feature projections on the X and Y axes according to the second optimal projection matrix;
[0038] splicing the second feature projections on the X and Y axes to obtain a fusion feature of the P wave signal;
[0039] The fusion features of the P wave signal are put into the SVM classifier for classification to obtain the atrial fibrillation warning information and non-atrial fibrillation warning information of the P wave signal.
[0040] To achieve the above-mentioned purpose, the second aspect of the present application provides an atrial fibrillation detection device under high-field strength magnetic resonance, the device comprising:
[0041] An electrocardiogram acquisition module, used for acquiring electrocardiogram signals;
[0042] A waveform acquisition module, used to locate characteristic points of the electrocardiogram signal to obtain R waves and P waves;
[0043] A matrix signal acquisition module, used to obtain a matrix signal of the RR interval and a matrix signal of the P wave according to the R wave and the P wave respectively;
[0044] A feature extraction module, used to extract features from the matrix signal of the RR interval to obtain traditional features and deep features of the RR interval signal;
[0045] A classification module, used for fusing and classifying the traditional features and deep features of the RR interval signal to obtain feature classification of the RR interval signal;
[0046] The feature extraction module is also used to extract features from the matrix signal of the P wave to obtain traditional features and deep features of the P wave signal;
[0047] The classification module is further used to fuse and classify the traditional features and deep features of the P-wave signal to obtain feature classification of the P-wave signal;
[0048] A judgment module is used to determine the atrial fibrillation alarm result according to the characteristic classification of the RR interval signal and the characteristic classification of the P wave signal.
[0049] To achieve the above-mentioned object, the third aspect of the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the following steps:
[0050] Acquire electrocardiogram signals;
[0051] Positioning feature points of the electrocardiogram signal to obtain R waves and P waves;
[0052] According to the R wave and the P wave, a matrix signal of the RR interval and a matrix signal of the P wave are obtained respectively;
[0053] Extracting features from the matrix signal of the RR interval to obtain traditional features and deep features of the RR interval signal;
[0054] fusing and classifying the traditional features and deep features of the RR interval signal to obtain feature classification of the RR interval signal;
[0055] Extracting features from the matrix signal of the P wave to obtain traditional features and deep features of the P wave signal;
[0056] fusing and classifying the traditional features and deep features of the P-wave signal to obtain feature classification of the P-wave signal;
[0057] The atrial fibrillation alarm result is determined based on the feature classification of the RR interval signal and the feature classification of the P wave signal.
[0058] To achieve the above-mentioned purpose, the fourth aspect of the present application provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the following steps:
[0059] Acquire electrocardiogram signals;
[0060] Positioning feature points of the electrocardiogram signal to obtain R waves and P waves;
[0061] According to the R wave and the P wave, a matrix signal of the RR interval and a matrix signal of the P wave are obtained respectively;
[0062] Extracting features from the matrix signal of the RR interval to obtain traditional features and deep features of the RR interval signal;
[0063] fusing and classifying the traditional features and deep features of the RR interval signal to obtain feature classification of the RR interval signal;
[0064] Extracting features from the matrix signal of the P wave to obtain traditional features and deep features of the P wave signal;
[0065] fusing and classifying the traditional features and the deep features of the P-wave signal to obtain feature classification of the P-wave signal;
[0066] The atrial fibrillation alarm result is determined based on the feature classification of the RR interval signal and the feature classification of the P wave signal.
[0067] The embodiments of the present invention have the following beneficial effects:
[0068] The present invention extracts features from matrix signals of two different domains, RR interval and P wave, respectively, to obtain respective traditional features and deep features; and fuses and classifies the traditional features and deep features of the matrix signals of the RR interval signal and the P wave, respectively, to obtain feature classification of the RR interval signal and feature classification of the P wave signal, and performs feature fusion and splicing by DCCA method to obtain the final feature vector, and then further performs judgment on atrial fibrillation detection, thereby enhancing the intra-class relationship and inter-class distance, making full use of the effective information provided in the electrocardiogram, enhancing the accuracy of atrial fibrillation discrimination, and fusing the two features to enhance the robustness of the algorithm, making it more suitable for the monitoring of high-field magnetic resonance imaging, and improving the detection accuracy and the accuracy of the detection results. BRIEF DESCRIPTION OF THE DRAWINGS
[0069] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0070] in:
[0071] Figure 1 is a schematic flow chart of a method for detecting atrial fibrillation under high-field magnetic resonance imaging in one embodiment;
[0072] Figure 2 is a structural diagram of an atrial fibrillation detection device under high field strength magnetic resonance in one embodiment;
[0073] Figure 3 is a schematic diagram of the structure of a computer device in one embodiment;
[0074] Figure 4 Schematic diagram of the structure of a computer-readable storage medium in one embodiment. DETAILED DESCRIPTION
[0075] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0076] To facilitate understanding, the relevant terms involved in this application are first introduced below.
[0077] (1) R wave: The R wave is one of the most prominent and easily recognizable waveforms in the electrocardiogram. It represents the depolarization process of the ventricles, that is, the process in which the ventricular myocytes begin to become excited and transmit electrical signals to the surrounding. On the electrocardiogram, the R wave appears as an upward spike or peak, and its amplitude and shape are of great value in diagnosing a variety of heart diseases. The height (or amplitude) of the R wave is often used to calculate other electrocardiogram parameters such as the QT interval.
[0078] (2) P wave: The P wave represents the depolarization process of the atrium, that is, the process in which the atrial myocytes begin to become excited and transmit electrical signals to the surrounding area. On the electrocardiogram, the P wave usually appears as a small positive waveform (in some cases it may be negative or bidirectional) and appears before the QRS complex. The morphology and duration of the P wave are of great significance for the diagnosis of atrial-related diseases such as atrial hypertrophy and atrial conduction block.
[0079] (3) F wave: F wave does not appear in all ECGs. It mainly appears in the ECG of patients with atrial fibrillation (AF). During atrial fibrillation, the electrical activity of the atria becomes very chaotic and rapid, making it impossible to clearly distinguish a single atrial depolarization waveform (P wave) on the ECG. Instead, it is replaced by a series of rapid, irregular waveforms with variable amplitudes and shapes, which are collectively referred to as F waves. The presence of F waves is an important diagnostic sign of atrial fibrillation.
[0080] (4) RR interval: The RR interval refers to the time interval between two consecutive R waves, which reflects the time between two ventricular contractions, that is, a complete cardiac cycle of the heart. The length of the RR interval is inversely proportional to the heart rate, that is, the shorter the RR interval, the faster the heart rate; conversely, the longer the RR interval, the slower the heart rate. The stability and variability of the RR interval (HRV) is also an important indicator for assessing heart health.
[0081] (5) HRV signal: HRV signal refers to the small difference between the RR intervals of successive heartbeats, which reflects the regulatory effect of the autonomic nervous system (especially the sympathetic nervous system and parasympathetic nervous system) on cardiac rhythm. The analysis of HRV signal is of great value in evaluating the autonomic nervous function of the cardiovascular system, predicting the risk of cardiovascular events, evaluating diabetic neuropathy, and guiding cardiac rehabilitation. High HRV is generally considered a sign of health, while low HRV may be associated with a variety of cardiovascular diseases and poor prognosis.
[0082] In the embodiment of the present application, a method for detecting atrial fibrillation under high field strength magnetic resonance is provided. Figure 1 , Figure 1 1 is a flow chart of a method for detecting atrial fibrillation under high-field magnetic resonance in an embodiment; the method for detecting atrial fibrillation under high-field magnetic resonance includes steps S1 to S8.
[0083] Step S1, obtaining an electrocardiogram signal;
[0084] In some embodiments, the electrocardiogram signals are acquired using a high field strength magnetic monitor.
[0085] Step S2, locating feature points of the electrocardiogram signal to obtain R waves and P waves;
[0086] In some implementations, the step of locating feature points of the electrocardiogram signal to obtain R waves and P waves specifically includes steps S21 to S24:
[0087] Step S21, preprocessing the electrocardiogram signal;
[0088] In some embodiments, the preprocessing of the electrocardiogram signal specifically includes:
[0089] Performing baseline processing on the electrocardiogram signal to obtain an electrocardiogram signal baseline,
[0090] The electrocardiogram signal baseline is removed from the original electrocardiogram signal, and then the high-frequency power frequency noise is removed to obtain a clean electrocardiogram signal, and the clean electrocardiogram signal is the preprocessed electrocardiogram signal.
[0091] Specifically, baseline processing is performed on the electrocardiogram signal. Using a 600 ms median filter and a 200 ms median filter, the electrocardiogram signal baseline is obtained. Then, from the original electrocardiogram signal, that is, the electrocardiogram signal before baseline processing, the baseline is subtracted, and the signal without the baseline is obtained. Further, high-frequency power frequency noise is removed to obtain a clean electrocardiogram signal, that is, the preprocessed electrocardiogram signal. This way of preprocessing the electrocardiogram signal can prevent noise from interfering with the feature point positioning, making the subsequent atrial fibrillation detection more accurate.
[0092] Step S22, use the PT algorithm to locate the preprocessed electrocardiogram signal to obtain the position set of the R wave in the electrocardiogram signal;
[0093] Step S23, obtain the HRV signal of the RR interval according to the R wave position set;
[0094] Specifically, use the PT algorithm to locate the electrocardiogram signal to obtain the accurate position set R(i) of the R wave, where i < size(ECG), that is, the number of the position set is less than the length of the electrocardiogram signal. Extract the R wave position information from the R wave position set R(i), and obtain the HRV signal, that is, the RR interval signal, according to RR(i) = R(i) - R(i - 1).
[0095] Step S24, determine the P wave according to the HRV signal of the RR interval;
[0096] Specifically, estimate the P wave range according to the RR interval.
[0097] Step S3, obtain the matrix signal of the RR interval and the matrix signal of the P wave according to the R wave and the P wave respectively;
[0098] Step S4, extract features from the matrix signal of the RR interval to obtain the traditional features and deep features of the RR interval signal;
[0099] In some embodiments, the extracting features from the matrix signal of the RR interval to obtain the traditional features and deep features of the RR interval signal specifically includes:
[0100] Extract features from the matrix signal of the RR interval to obtain ratio features, Poincaré plot features, and entropy features in the traditional features;
[0101] Specifically, extracting the ratio features in the matrix signal of the RR interval can indirectly reflect the irregularity of the RR interval; extracting the Poincare map features can reflect the patient's physical signs. Under normal circumstances, the Poincare map features extracted from the RR interval are relatively concentrated, and when atrial fibrillation occurs, the Poincare map features extracted from the RR interval are relatively scattered; extracting the entropy features can also reflect the patient's physical signs. Under normal circumstances, the entropy value of the RR interval is relatively low, while the entropy value of the RR interval with atrial fibrillation is relatively high.
[0102] The matrix signal of the RR interval is extracted with a neural network to obtain deep features.
[0103] Specifically, the matrix signal of the RR interval is input into the convolution layer, and the features of the convolution layer are subjected to the relu activation function. This layer performs operations based on the nonlinear decision mapping function, which enhances the model's ability to express features. At this time, there will be multiple feature representations after the convolution layer, and the pooling layer can be further used to reduce the amount of parameters and reduce the pressure of calculation. The technical solution of the present application utilizes maximum pooling, continues to perform convolution layer operations, extracts activation functions, and reduces parameters in the pooling layer; continues to perform activation functions, enters the convolution layer, and obtains the final parameter features in the fully connected layer, that is, the deep features of the matrix signal of the RR interval. The fully connected layer plays the role of integrating features and can stably map the features extracted by the neurons in the hidden layer to the label space of the data.
[0104] Step S5, fusing and classifying the traditional features and deep features of the RR interval signal to obtain feature classification of the RR interval signal;
[0105] In some embodiments, the fusing and classifying the traditional features and the deep features of the RR interval signal to obtain the feature classification of the RR interval signal specifically includes:
[0106] Step S51, connecting the traditional features and deep features of the RR interval signal in series to form a first feature matrix;
[0107] Specifically, the traditional features of the RR interval: ratio feature, Poincare map feature, entropy feature and depth feature are prepared, and two feature matrices are connected in series to form a feature matrix, which is the first feature matrix.
[0108] Step S52, obtaining a first optimal projection matrix of the first feature matrix according to a class representativeness correlation analysis (DCCA) algorithm;
[0109] Step S53, obtaining a first feature projection on the X-axis and the Y-axis according to the first optimal projection matrix;
[0110] Step S54, splicing the first feature projections on the X-axis and the Y-axis to obtain the fusion feature of the RR interval signal;
[0111] Step S55, putting the fusion features of the RR interval signal into the SVM classifier for classification, and obtaining the atrial fibrillation warning information and non-atrial fibrillation warning information of the RR interval signal.
[0112] Specifically, according to the fusion characteristics of the RR interval signal, an SVM classifier is used for classification, and the classification result is obtained as early warning information of atrial fibrillation or non-atrial fibrillation.
[0113] Step S6, extracting features from the matrix signal of the P wave to obtain traditional features and depth features of the P wave signal;
[0114] In some embodiments, extracting features from the matrix signal of the P wave to obtain traditional features and depth features of the P wave signal specifically includes:
[0115] Extracting features from the matrix signal of the P wave to obtain the width of the P wave detection range and the maximum value within the P wave detection range in traditional features;
[0116] Specifically, whether there is a P wave can be determined based on the maximum value within the P wave detection range. When the maximum value within the P wave detection range is relatively large, there is a P wave. When the maximum value within the P wave detection range is relatively small, there is no P wave. Specifically, it can be calculated based on the class typicality correlation analysis (DCCA) algorithm. If the P wave exists, atrial fibrillation has not occurred. If the P wave does not exist, atrial fibrillation has occurred.
[0117] The deep features are obtained by extracting features from the matrix signal of the P wave through a neural network.
[0118] Specifically, the matrix signal of the P wave is input into the convolution layer, and the features of the convolution layer are subjected to the relu activation function. This layer performs operations based on the nonlinear decision mapping function, which enhances the model's ability to express features. At this time, there will be multiple feature representations after the convolution layer, and the pooling layer can be further used to reduce the amount of parameters and reduce the pressure of calculation. The technical solution of the present application utilizes maximum pooling, continues to perform convolution layer operations, extracts activation functions, and reduces parameters in the pooling layer; continues to perform activation functions, enters the convolution layer, and obtains the final parameter features in the fully connected layer, that is, the deep features of the matrix signal of the P wave. The fully connected layer plays the role of integrating features, and can stably map the features extracted by the neurons in the hidden layer to the label space of the data.
[0119] Step S7, fusing and classifying the traditional features and deep features of the P-wave signal to obtain feature classification of the P-wave signal;
[0120] In some embodiments, the traditional features and deep features of the P wave signal are fused and classified to obtain feature classification of the P wave signal, specifically including:
[0121] The traditional features and the deep features of the P-wave signal are connected in series to form a second feature matrix;
[0122] Obtaining a second optimal projection matrix of the second feature matrix according to a class representativeness correlation analysis (DCCA) algorithm;
[0123] Obtaining second feature projections on the X and Y axes according to the second optimal projection matrix;
[0124] splicing the second feature projections on the X and Y axes to obtain a fusion feature of the P wave signal;
[0125] The fusion features of the P wave signal are put into the SVM classifier for classification to obtain the atrial fibrillation warning information and non-atrial fibrillation warning information of the P wave signal.
[0126] Specifically, the fusion features of the P wave signal are put into the SVM classifier for classification, and the classification result is the warning information of atrial fibrillation or non-atrial fibrillation.
[0127] Step S8, determining an atrial fibrillation alarm result according to the feature classification of the RR interval signal and the feature classification of the P wave signal.
[0128] In some implementations, determining the atrial fibrillation alarm result according to the characteristic classification of the RR interval signal and the characteristic classification of the P wave signal specifically includes:
[0129] Combine the atrial fibrillation warning information and non-atrial fibrillation warning information of the RR interval and P wave to make atrial fibrillation alarm judgment;
[0130] When the classification information of the RR interval and the P wave both show atrial fibrillation, an atrial fibrillation alarm is triggered.
[0131] Specifically, for example, if the P wave exists and the RR interval is regular, atrial fibrillation does not occur; if the P wave does not exist and the RR interval is irregular, atrial fibrillation occurs, triggering an atrial fibrillation alarm.
[0132] By adopting the technical solution of this embodiment, features are extracted from matrix signals of two different domains, the RR interval and the P wave, respectively, to obtain their respective traditional features and deep features; and the traditional features and deep features of the matrix signals of the RR interval signal and the P wave are fused and classified respectively to obtain feature classification of the RR interval signal and feature classification of the P wave signal, and the final feature vector is obtained by feature fusion and splicing using the DCCA method, and then the judgment of atrial fibrillation detection is performed, which enhances the intra-class relationship and inter-class distance, makes full use of the effective information provided in the electrocardiogram, enhances the accuracy of atrial fibrillation discrimination, and the fusion of the two features enhances the robustness of the algorithm, is more suitable for the monitoring of high-field magnetic resonance imaging, and improves the detection accuracy and the accuracy of the detection results.
[0133] In the embodiment of the present application, a device for detecting atrial fibrillation under high field strength magnetic resonance is provided. Figure 2 , Figure 2 2 is a diagram of an atrial fibrillation detection device under high-field strength magnetic resonance in an embodiment, wherein the atrial fibrillation detection device under high-field strength magnetic resonance includes an electrocardiogram acquisition module 201, a waveform acquisition module 202, a matrix signal acquisition module 203, a feature extraction module 204, a classification module 205 and a judgment module 206.
[0134] The electrocardiogram acquisition module 201 is configured to acquire electrocardiogram signals;
[0135] The waveform acquisition module 202 is configured to locate feature points of the electrocardiogram signal to obtain R waves and P waves;
[0136] A matrix signal acquisition module 203 is configured to obtain a matrix signal of the RR interval and a matrix signal of the P wave according to the R wave and the P wave respectively;
[0137] The feature extraction module 204 is configured to extract features from the matrix signal of the RR interval to obtain traditional features and deep features of the RR interval signal;
[0138] The classification module 205 is configured to fuse and classify the traditional features and deep features of the RR interval signal to obtain feature classification of the RR interval signal;
[0139] The feature extraction module 204 is further configured to extract features from the matrix signal of the P wave to obtain traditional features and deep features of the P wave signal;
[0140] The classification module 205 is further configured to fuse and classify the traditional features and the deep features of the P-wave signal to obtain feature classification of the P-wave signal;
[0141] The judgment module 206 is configured to determine the atrial fibrillation alarm result according to the characteristic classification of the RR interval signal and the characteristic classification of the P wave signal.
[0142] In some embodiments, the waveform acquisition module 202 is further configured to pre-process the electrocardiogram signal;
[0143] Positioning the preprocessed electrocardiogram signal by using a PT algorithm to obtain a position set of the R wave in the electrocardiogram signal;
[0144] Obtaining an HRV signal of the RR interval according to the R wave position set;
[0145] The P wave is determined based on the HRV signal of the RR interval.
[0146] In some embodiments, the atrial fibrillation detection device under high field strength magnetic resonance further includes a preprocessing module 207, wherein the preprocessing module 207 is used to perform baseline processing on the electrocardiogram signal to obtain an electrocardiogram signal baseline.
[0147] The electrocardiogram signal baseline is removed from the original electrocardiogram signal, and then the high-frequency power frequency noise is removed to obtain a clean electrocardiogram signal, and the clean electrocardiogram signal is the preprocessed electrocardiogram signal.
[0148] In some implementations, the feature extraction module 204 is further configured to extract features from the matrix signal of the RR interval to obtain ratio features, Poincare map features, and entropy features among traditional features;
[0149] The matrix signal of the RR interval is extracted with a neural network to obtain deep features.
[0150] In some embodiments, the classification module 205 is further configured to concatenate the traditional features and the deep features of the RR interval signal into a first feature matrix;
[0151] Obtaining a first optimal projection matrix of the first feature matrix according to a class representativeness correlation analysis (DCCA) algorithm;
[0152] Obtaining a first feature projection on the X-axis and the Y-axis according to the first optimal projection matrix;
[0153] Splicing the first feature projections on the X-axis and the Y-axis to obtain the fusion feature of the RR interval signal;
[0154] The fusion features of the RR interval signal are put into the SVM classifier for classification to obtain atrial fibrillation warning information and non-atrial fibrillation warning information of the RR interval signal.
[0155] In some embodiments, the feature extraction module 204 is further configured to extract features from the matrix signal of the P wave to obtain the width of the P wave detection range and the maximum value within the P wave detection range in traditional features;
[0156] The deep features are obtained by extracting features from the matrix signal of the P wave through a neural network.
[0157] In some embodiments, the classification module 205 is further configured to connect the traditional features and the deep features of the P-wave signal in series to form a second feature matrix;
[0158] Obtaining a second optimal projection matrix of the second feature matrix according to a class representativeness correlation analysis (DCCA) algorithm;
[0159] Obtaining second feature projections on the X and Y axes according to the second optimal projection matrix;
[0160] splicing the second feature projections on the X and Y axes to obtain a fusion feature of the P wave signal;
[0161] The fusion features of the P wave signal are put into the SVM classifier for classification to obtain the atrial fibrillation warning information and non-atrial fibrillation warning information of the P wave signal.
[0162] In some implementations, the determination module 206 is further configured to perform an atrial fibrillation alarm determination by combining the atrial fibrillation warning information and the non-atrial fibrillation warning information of the RR interval and the P wave;
[0163] When the classification information of the RR interval and the P wave both show atrial fibrillation, an atrial fibrillation alarm is triggered.
[0164] For other details about how each module in the atrial fibrillation detection device under high-field strength magnetic resonance implements the above technical solution, please refer to the description of the atrial fibrillation detection method under high-field strength magnetic resonance provided above, which will not be repeated here.
[0165] In an embodiment of the present application, a computer device is provided. Figure 3 , Figure 3 3 is a schematic diagram of the structure of a computer device in an embodiment. The device includes a memory 301 and a processor 302. The memory 301 stores a computer program. When the computer program is executed by the processor 302, the processor 302 performs the following steps:
[0166] Acquire electrocardiogram signals;
[0167] Positioning feature points of the electrocardiogram signal to obtain R waves and P waves;
[0168] According to the R wave and the P wave, a matrix signal of the RR interval and a matrix signal of the P wave are obtained respectively;
[0169] Extracting features from the matrix signal of the RR interval to obtain traditional features and deep features of the RR interval signal;
[0170] fusing and classifying the traditional features and deep features of the RR interval signal to obtain feature classification of the RR interval signal;
[0171] Extracting features from the matrix signal of the P wave to obtain traditional features and deep features of the P wave signal;
[0172] fusing and classifying the traditional features and the deep features of the P-wave signal to obtain feature classification of the P-wave signal;
[0173] The atrial fibrillation alarm result is determined based on the feature classification of the RR interval signal and the feature classification of the P wave signal.
[0174] Among them, the processor 302 can also be called a CPU (Central Processing Unit), and the processor 302 may be an integrated circuit chip with signal processing capabilities; the processor 302 can also be a general-purpose processor, DSP (Digital Signal Process), ASIC (Application Specific Integrated Circuit), FPGA (Field Programmable Gate Array) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, among which the general-purpose processor can be a microprocessor or the processor 302 can also be any conventional processor, etc.
[0175] In an embodiment of the present application, a computer readable storage medium is provided. Figure 4 , Figure 4 The structure diagram of a computer-readable storage medium in an embodiment is a schematic diagram, on which a readable computer program 401 is stored; wherein the computer program 401 may be stored in the above storage medium in the form of a software product, and includes a number of instructions for causing a computer device (which may be a personal computer, a server machine, or a network device, etc.) or a processor to perform the following steps:
[0176] Acquire electrocardiogram signals;
[0177] Positioning feature points of the electrocardiogram signal to obtain R waves and P waves;
[0178] According to the R wave and the P wave, a matrix signal of the RR interval and a matrix signal of the P wave are obtained respectively;
[0179] Extracting features from the matrix signal of the RR interval to obtain traditional features and deep features of the RR interval signal;
[0180] fusing and classifying the traditional features and deep features of the RR interval signal to obtain feature classification of the RR interval signal;
[0181] Extracting features from the matrix signal of the P wave to obtain traditional features and deep features of the P wave signal;
[0182] fusing and classifying the traditional features and deep features of the P-wave signal to obtain feature classification of the P-wave signal;
[0183] The atrial fibrillation alarm result is determined based on the feature classification of the RR interval signal and the feature classification of the P wave signal.
[0184] The aforementioned storage media include: U disk, mobile hard disk, magnetic disk or CD, ROM (Read-Only Memory), RAM (Random Access Memory) and other media that can store program code, or terminal devices such as computers, service machines, mobile phones, and tablets.
[0185] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0186] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0187] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
Claims
1. A method for detecting atrial fibrillation under high-field magnetic resonance imaging, the method comprising: Acquire electrocardiogram signals; Positioning feature points of the electrocardiogram signal to obtain R waves and P waves; According to the R wave and the P wave, a matrix signal of the RR interval and a matrix signal of the P wave are obtained respectively; Extracting features from the matrix signal of the RR interval to obtain traditional features and deep features of the RR interval signal; fusing and classifying the traditional features and deep features of the RR interval signal to obtain feature classification of the RR interval signal; Extracting features from the matrix signal of the P wave to obtain traditional features and deep features of the P wave signal; fusing and classifying the traditional features and the deep features of the P-wave signal to obtain feature classification of the P-wave signal; The atrial fibrillation alarm result is determined based on the feature classification of the RR interval signal and the feature classification of the P wave signal.
2. The method for detecting atrial fibrillation under high-field magnetic resonance according to claim 1, characterized in that: The performing of feature point positioning on the electrocardiogram signal to obtain the R wave and the P wave specifically includes: Preprocessing the electrocardiogram signal; Positioning the preprocessed electrocardiogram signal by using a PT algorithm to obtain a position set of the R wave in the electrocardiogram signal; Obtaining an HRV signal of the RR interval according to the R wave position set; The P wave is determined based on the HRV signal of the RR interval.
3. The method for detecting atrial fibrillation under high-field magnetic resonance according to claim 2, characterized in that: The preprocessing of the electrocardiogram signal specifically includes: Performing baseline processing on the electrocardiogram signal to obtain an electrocardiogram signal baseline, The electrocardiogram signal baseline is removed from the original electrocardiogram signal, and then the high-frequency power frequency noise is removed to obtain a clean electrocardiogram signal, and the clean electrocardiogram signal is the preprocessed electrocardiogram signal.
4. The method for detecting atrial fibrillation under high-field magnetic resonance according to claim 3, characterized in that: The extracting features of the matrix signal of the RR interval to obtain the traditional features and deep features of the RR interval signal specifically includes: Extracting features from the matrix signal of the RR interval to obtain ratio features, Poincare map features, and entropy features among traditional features; The matrix signal of the RR interval is extracted with a neural network to obtain deep features.
5. The method for detecting atrial fibrillation under high field strength magnetic resonance according to claim 4, characterized in that: The fusing and classifying the traditional features and the deep features of the RR interval signal to obtain the feature classification of the RR interval signal specifically includes: The traditional features and deep features of the RR interval signal are connected in series to form a first feature matrix; Obtaining a first optimal projection matrix of the first feature matrix according to a class representativeness correlation analysis (DCCA) algorithm; Obtaining a first feature projection on the X-axis and the Y-axis according to the first optimal projection matrix; Splicing the first feature projections on the X-axis and the Y-axis to obtain the fusion feature of the RR interval signal; The fusion features of the RR interval signal are put into the SVM classifier for classification to obtain atrial fibrillation warning information and non-atrial fibrillation warning information of the RR interval signal.
6. The method for detecting atrial fibrillation under high-field magnetic resonance according to claim 5, characterized in that: The extracting features of the matrix signal of the P wave to obtain the traditional features and deep features of the P wave signal specifically includes: Extracting features from the matrix signal of the P wave to obtain the width of the P wave detection range and the maximum value within the P wave detection range in traditional features; The deep features are obtained by extracting features from the matrix signal of the P wave through a neural network.
7. The method for detecting atrial fibrillation under high-field magnetic resonance according to claim 6, characterized in that: The traditional features and deep features of the P wave signal are integrated and classified to obtain feature classification of the P wave signal, specifically including: The traditional features and the deep features of the P-wave signal are connected in series to form a second feature matrix; Obtaining a second optimal projection matrix of the second feature matrix according to a class representativeness correlation analysis (DCCA) algorithm; Obtaining second feature projections on the X and Y axes according to the second optimal projection matrix; splicing the second feature projections on the X and Y axes to obtain a fusion feature of the P wave signal; The fusion features of the P wave signal are put into the SVM classifier for classification to obtain the atrial fibrillation warning information and non-atrial fibrillation warning information of the P wave signal.
8. The method for detecting atrial fibrillation under high-field magnetic resonance according to claim 7, characterized in that: The determining of the atrial fibrillation alarm result according to the characteristic classification of the RR interval signal and the characteristic classification of the P wave signal specifically includes: Combine the atrial fibrillation warning information and non-atrial fibrillation warning information of the RR interval and P wave to make atrial fibrillation alarm judgment; When the classification information of the RR interval and the P wave both show atrial fibrillation, an atrial fibrillation alarm is triggered.
9. A device for detecting atrial fibrillation under high-field magnetic resonance, characterized in that: The device comprises: An electrocardiogram acquisition module, used for acquiring electrocardiogram signals; A waveform acquisition module, used to locate characteristic points of the electrocardiogram signal to obtain R waves and P waves; A matrix signal acquisition module, used to obtain a matrix signal of the RR interval and a matrix signal of the P wave according to the R wave and the P wave respectively; A feature extraction module, used to extract features from the matrix signal of the RR interval to obtain traditional features and deep features of the RR interval signal; A classification module, used for fusing and classifying the traditional features and deep features of the RR interval signal to obtain feature classification of the RR interval signal; The feature extraction module is also used to extract features from the matrix signal of the P wave to obtain traditional features and deep features of the P wave signal; The classification module is further used to fuse and classify the traditional features and deep features of the P-wave signal to obtain feature classification of the P-wave signal; A judgment module is used to determine the atrial fibrillation alarm result according to the characteristic classification of the RR interval signal and the characteristic classification of the P wave signal.
10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to any one of claims 1 to 8.
11. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 8.
