Arrhythmia analysis method and device based on heartbeat classification, equipment and medium
By generating morphological amplitude encoding vectors and calculating inner product values from heartbeat segments of electrocardiogram signals, the problems of large computational load and ignored waveform differences in existing technologies are solved, achieving high efficiency and accuracy in arrhythmia analysis, which is suitable for real-time monitoring.
Patent Information
- Application Number
- CN202210383594.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-04-12
- Publication Date
- 2026-06-23
- Estimated Expiration
- 2042-04-12
AI Technical Summary
Existing methods for analyzing arrhythmias involve a large computational burden when calculating the similarity between a heartbeat template and a heartbeat segment, and they also ignore the amplitude differences of the waveforms and tend to overlook the morphological differences of the waveforms when the amplitude is low.
By acquiring the time series set of heartbeat segments from the target ECG signal, morphological amplitude encoding vectors are generated for the target heartbeat segments, and the inner product is calculated with each heartbeat template vector in the preset heartbeat template vector set to determine the target heartbeat type. Finally, arrhythmia analysis is performed.
It simplifies the calculation process, avoids the problems of large calculation volume and ignoring waveform differences, is suitable for real-time monitoring scenarios, and improves the accuracy of analysis results.
Smart Images
Figure CN116956101B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of electrocardiogram signal analysis technology, and in particular to a method, device, equipment and medium for analyzing arrhythmias based on heart rate classification. Background Technology
[0002] As society ages, the incidence and severity of cardiovascular diseases are constantly rising. Electrocardiogram (ECG), as a routine physical examination, plays a crucial role in the diagnosis and monitoring of cardiovascular diseases. Real-time arrhythmia analysis is an important function in clinical ECG monitoring, and current arrhythmia analysis methods mainly employ heartbeat template matching analysis. A key step in existing heartbeat template matching analysis methods is calculating the similarity between the heartbeat template and heartbeat segments. Methods for calculating this similarity include: Pearson correlation, Euclidean distance, and wavelet feature analysis. Pearson correlation is sensitive to morphological differences in heartbeat waveforms, but it ignores amplitude differences. Euclidean distance is sensitive to amplitude differences, but it tends to overlook morphological differences at low amplitudes. Wavelet feature analysis is highly sensitive to both morphological and amplitude differences, but it requires wavelet transform, resulting in significant computational overhead. Summary of the Invention
[0003] The main objective of this application is to provide a method, device, equipment, and medium for analyzing arrhythmias based on heartbeat classification. This aims to solve one of the technical problems of current arrhythmia analysis methods, namely, the large computational load, neglect of waveform amplitude differences, and easy neglect of waveform morphological differences when the amplitude is low.
[0004] To achieve the aforementioned objectives, this application proposes a method for analyzing cardiac arrhythmias based on heart rate classification, the method comprising:
[0005] Obtain the time sequence set of heartbeat segments corresponding to the target electrocardiogram signal;
[0006] A morphological amplitude encoding vector is generated for the target heartbeat segment, which serves as the target morphological amplitude encoding vector, wherein the target heartbeat segment is any heartbeat segment in the heartbeat segment time sequence set;
[0007] The inner product of the target shape amplitude encoding vector and each heartbeat template vector in the preset heartbeat template vector set is calculated to obtain the inner product value set.
[0008] The target heartbeat type corresponding to the target heartbeat segment is determined based on the inner product set;
[0009] Arrhythmia analysis is performed based on the target heartbeat type corresponding to the heartbeat segment time sequence set to obtain arrhythmia analysis results.
[0010] Furthermore, the step of obtaining the time series set of heartbeat segments corresponding to the target electrocardiogram signal includes:
[0011] Acquire initial electrocardiogram (ECG) signals;
[0012] The initial electrocardiogram (ECG) signal is bandpass filtered using a preset filtering range to obtain the target ECG signal.
[0013] The target electrocardiogram signal is subjected to heartbeat segment extraction to obtain the heartbeat segment time series.
[0014] Furthermore, the step of extracting the heartbeat segments from the target electrocardiogram signal to obtain the heartbeat segment time series set includes:
[0015] The target electrocardiogram signal is subjected to heartbeat location detection to obtain a heartbeat location set;
[0016] Using a preset time window and taking the target heartbeat position as the center position, the heartbeat segment corresponding to the target heartbeat position is extracted from the target electrocardiogram signal, wherein the target heartbeat position is any heartbeat position in the set of heartbeat positions;
[0017] The temporal set of the heartbeat segments is determined based on each heartbeat segment corresponding to the heartbeat position set.
[0018] Furthermore, the step of generating a morphological amplitude encoding vector for the target heartbeat fragment as the target morphological amplitude encoding vector includes:
[0019] The target heartbeat fragments are sampled at preset time intervals to obtain a sampling result set;
[0020] The average value of the sampled result set is calculated to obtain the target average value;
[0021] The target amplitude is obtained by calculating the amplitude based on the target average value and the sampling result set;
[0022] Based on the target average value and the target amplitude, the sampling result set is normalized to obtain the amplitude normalization result;
[0023] The target angle is obtained by mapping the target amplitude and the preset maximum amplitude of lossless encoding.
[0024] The target angle is calculated by performing target cosine and target sine values respectively;
[0025] The amplitude normalization result, the target cosine value, and the target sine value are concatenated into vectors to obtain the target shape amplitude encoding vector.
[0026] Furthermore, the formula for calculating the target amplitude is amp. x for:
[0027]
[0028] The formula for calculating the i-th normalized result in the amplitude normalization results. for:
[0029]
[0030] The formula for calculating the target angle is θ. x for:
[0031]
[0032] Where N is the number of sampling results in the sampling result set, x i It is the i-th sampling result in the sampling result set. The target average value is eps, which is a preset constant to avoid division by zero, max() calculates the maximum value, and amp max is the maximum amplitude value of the lossless encoding, min() is used to calculate the minimum value, and π is the mathematical constant pi.
[0033] Furthermore, the step of determining the target heartbeat type corresponding to the target heartbeat segment based on the inner product set includes:
[0034] Find the inner product value with the largest value from the set of inner product values, and use it as the target inner product value;
[0035] The target inner product value is used as the target heartbeat template vector;
[0036] The heartbeat type corresponding to the target heartbeat template vector is taken as the target heartbeat type corresponding to the target heartbeat fragment.
[0037] Furthermore, the step of performing arrhythmia analysis based on each target heartbeat type corresponding to the heartbeat segment time series set to obtain arrhythmia analysis results includes:
[0038] Based on the target heartbeat type corresponding to the heartbeat segment time sequence set and the preset clinical rules, arrhythmia analysis is performed to obtain the arrhythmia analysis results.
[0039] This application also proposes a cardiac arrhythmia analysis device based on heartbeat classification, the device comprising:
[0040] The data acquisition module is used to acquire the time series set of heartbeat segments corresponding to the target electrocardiogram signal;
[0041] The target morphological amplitude encoding vector generation module is used to generate a morphological amplitude encoding vector for a target heartbeat segment, which serves as the target morphological amplitude encoding vector. The target heartbeat segment is any heartbeat segment in the heartbeat segment time sequence set.
[0042] The inner product value set determination module is used to perform inner product calculation on the target morphology amplitude encoding vector and each heartbeat template vector in the preset heartbeat template vector set to obtain the inner product value set.
[0043] The target heartbeat type determination module is used to determine the target heartbeat type corresponding to the target heartbeat segment based on the inner product set;
[0044] The arrhythmia analysis result determination module is used to perform arrhythmia analysis based on each target heartbeat type corresponding to the heartbeat segment time series set, and obtain arrhythmia analysis results.
[0045] This application also proposes a computer device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods described above.
[0046] This application also proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.
[0047] This application discloses a method, apparatus, device, and medium for arrhythmia analysis based on heartbeat classification. The method acquires a time series set of heartbeat segments corresponding to a target electrocardiogram signal; generates a morphological amplitude encoding vector for each target heartbeat segment as a target morphological amplitude encoding vector, wherein the target heartbeat segment is any heartbeat segment in the time series set of heartbeat segments; calculates the inner product between the target morphological amplitude encoding vector and each heartbeat template vector in a preset heartbeat template vector set to obtain an inner product value set; determines the target heartbeat type corresponding to the target heartbeat segment based on the inner product value set; and performs arrhythmia analysis based on each target heartbeat type corresponding to the heartbeat segment time series set to obtain the arrhythmia analysis result. By calculating the inner product between the morphological amplitude encoding vector and the heartbeat template vector, the similarity calculation is not required using Pearson correlation, Euclidean distance, or wavelet feature methods. This avoids the technical problems of large computational load, neglecting waveform amplitude differences, and easily ignoring waveform morphological differences when the amplitude is low. Moreover, the process of generating morphological amplitude encoding vectors for the target heartbeat segment and calculating the inner product between the morphological amplitude encoding vector and the heartbeat template vector is simple and computationally inefficient, making this application suitable for real-time monitoring scenarios. Attached Figure Description
[0048] Figure 1 This is a flowchart illustrating a cardiac arrhythmia analysis method based on heart rate classification according to an embodiment of this application.
[0049] Figure 2 This is a schematic block diagram of a cardiac arrhythmia analysis device based on heart rate classification according to an embodiment of this application;
[0050] Figure 3 This is a schematic block diagram of the structure of a computer device according to an embodiment of this application.
[0051] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0052] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0053] Reference Figure 1 This application provides a method for analyzing arrhythmias based on heart rate classification, the method comprising:
[0054] S1: Obtain the time sequence set of heartbeat segments corresponding to the target ECG signal;
[0055] S2: Generate a morphological amplitude encoding vector for the target heartbeat segment, which serves as the target morphological amplitude encoding vector, wherein the target heartbeat segment is any heartbeat segment in the heartbeat segment time sequence set;
[0056] S3: Calculate the inner product between the target shape amplitude encoding vector and each heartbeat template vector in the preset heartbeat template vector set to obtain the inner product value set;
[0057] S4: Determine the target heartbeat type corresponding to the target heartbeat segment based on the inner product set;
[0058] S5: Perform arrhythmia analysis based on the target heartbeat types corresponding to the heartbeat segment time sequence set, and obtain arrhythmia analysis results.
[0059] This embodiment calculates the inner product value by using the morphological amplitude encoding vector and the heartbeat template vector, eliminating the need for similarity calculations using Pearson correlation, Euclidean distance, and wavelet feature methods. This avoids the technical problems of high computational cost, neglecting waveform amplitude differences, and easily ignoring waveform morphological differences when the amplitude is low. Furthermore, generating morphological amplitude encoding vectors for the target heartbeat segment and calculating the inner product value based on the morphological amplitude encoding vector and the heartbeat template vector is a simple and computationally inefficient process, making this application suitable for real-time monitoring scenarios.
[0060] For S1, it can obtain the time series set of heartbeat segments corresponding to the target ECG signal input by the user, or obtain the time series set of heartbeat segments corresponding to the target ECG signal from the database, or obtain the time series set of heartbeat segments corresponding to the target ECG signal from a third-party application.
[0061] The target ECG signal is the ECG signal that needs to be classified into different heartbeats.
[0062] Electrocardiogram (ECG) signals are the potential differences generated by the cell membranes of heart cells in the human heart. During rest intervals, the atria and ventricles are in a "polarized state" due to the large differences in the concentrations of ions (including K+, Na+, Ca2+, Cl-, etc.) inside and outside the cells.
[0063] The heartbeat segments in the heartbeat segment time sequence set are heartbeat segments extracted from the target electrocardiogram signal. The heartbeat segments in the heartbeat segment time sequence set are ordered from earliest to latest in chronological order.
[0064] A heartbeat segment is a signal segment that contains a heartbeat.
[0065] For S2, a morphological amplitude encoding vector is generated for the target heartbeat segment, and the generated morphological amplitude encoding vector is used as the target morphological amplitude encoding vector.
[0066] A morphological amplitude encoding vector encodes the morphological and amplitude information of a signal segment into a vector.
[0067] For S3, the inner product of the target shape amplitude encoding vector and each heartbeat template vector in the preset heartbeat template vector set is calculated, and the calculated inner product values are used as the inner product value set.
[0068] In other words, there is a one-to-one correspondence between the inner product values in the inner product value set and the heartbeat template vectors in the heartbeat template vector set.
[0069] The heartbeat template vector set includes: heartbeat template vector and heartbeat type. Based on waveform morphology and heartbeat interval, heartbeat segments are extracted from the electrocardiogram (ECG) signal. These extracted heartbeat segments are used as heartbeat templates. A morphological amplitude encoding vector is generated from the heartbeat template, and this generated morphological amplitude encoding vector is used as the heartbeat template vector. The heartbeat type has only one possible value. The range of heartbeat type values includes: normal, supraventricular abnormality, and ventricular abnormality.
[0070] Optionally, the step of calculating the inner product between the target morphology amplitude encoding vector and each heartbeat template vector in the preset heartbeat template vector set is to directly calculate the inner product between the target morphology amplitude encoding vector and each heartbeat template vector in the preset heartbeat template vector set.
[0071] Optionally, the step of calculating the inner product between the target morphology amplitude encoding vector and each heartbeat template vector in the preset heartbeat template vector set is to calculate the inner product between the transpose of the target morphology amplitude encoding vector and the transpose of each heartbeat template vector in the preset heartbeat template vector set.
[0072] For S4, the target inner product value is determined from the inner product value set, and the heartbeat type corresponding to the heartbeat template vector of the target inner product value in the heartbeat template vector set is taken as the target heartbeat type.
[0073] For S5, each target heartbeat type corresponding to the heartbeat segment time sequence set is taken as a heartbeat type combination, and then arrhythmia analysis is performed based on the heartbeat type combination. The data obtained from the analysis is taken as the arrhythmia analysis result.
[0074] In one embodiment, the step of obtaining the time series set of heartbeat segments corresponding to the target electrocardiogram signal includes:
[0075] S11: Acquire initial electrocardiogram signal;
[0076] S12: Using a preset filtering range, bandpass filter is applied to the initial ECG signal to obtain the target ECG signal;
[0077] S13: Extract the heartbeat segments from the target electrocardiogram signal to obtain the heartbeat segment time series.
[0078] This embodiment extracts heartbeat segments from the ECG signal obtained by bandpass filtering, thereby removing noise and improving the accuracy of the heartbeat segment time series set, thus improving the accuracy of the determined heartbeat classification results.
[0079] For S11, it can obtain the initial ECG signal input by the user, or obtain the initial ECG signal from the database, or obtain the initial ECG signal from a third-party application.
[0080] The initial electrocardiogram (ECG) signal is the ECG signal detected by the monitoring equipment.
[0081] For S12, a preset filtering range is used to perform bandpass filtering on the initial ECG signal, and the signal obtained by bandpass filtering is used as the target ECG signal.
[0082] For example, if the preset filtering range is 0.5Hz to 45Hz, then a signal of 0.5Hz to 45Hz is extracted from the initial ECG signal, and the extracted signal is used as the target ECG signal. At this time, the frequency range of the target ECG signal is greater than or equal to 0.5Hz and less than or equal to 45Hz.
[0083] In another embodiment of this application, the initial electrocardiogram signal is directly used as the target electrocardiogram signal.
[0084] For S13, extract signal segments containing heartbeats from the target electrocardiogram signal, take each extracted signal segment as a heartbeat segment, and determine the time series set of the heartbeat segments based on each heartbeat segment.
[0085] In one embodiment, the step of extracting the heartbeat segments from the target electrocardiogram signal to obtain the heartbeat segment time series set includes:
[0086] S131: The Pan-Tompkins algorithm is used to detect the heartbeat position of the target electrocardiogram signal to obtain a heartbeat position set;
[0087] S132: Using a preset time window and a method that takes the target heartbeat position as the center position, extract the heartbeat segment corresponding to the target heartbeat position from the target electrocardiogram signal, wherein the target heartbeat position is any one of the heartbeat positions in the set of heartbeat positions;
[0088] S133: Determine the temporal set of the heartbeat segments based on each heartbeat segment corresponding to the heartbeat position set.
[0089] In this embodiment, the heartbeat location is used as the center of a preset time window to extract heartbeat segments from the target electrocardiogram signal, thereby quickly extracting signal segments including heartbeats and providing a basis for classifying heartbeat segments.
[0090] For S131, the target electrocardiogram signal is subjected to heartbeat position detection, and each detected heartbeat position is used as a heartbeat position set.
[0091] Heartbeat position is the location of the heartbeat in the electrocardiogram (ECG) signal.
[0092] Specifically, the Pan-Tompkins algorithm is used to detect the heartbeat position of the target electrocardiogram signal. It is understood that a heartbeat detection algorithm can also be used to detect the heartbeat position.
[0093] The Pan-Tompkins algorithm, or PT algorithm for short, is an algorithm used to detect QRS complex information in electrocardiograms.
[0094] The QRS complex reflects the changes in depolarization potential and time between the left and right ventricles. The first downward wave is the Q wave, the upward wave is the R wave, and the next downward wave is the S wave. The time from the beginning to the end of the QRS complex is the QRS duration.
[0095] For S132, a method using a preset time window and centering on the target heartbeat position is employed. Specifically, the preset time window is used as the extraction length, and the target heartbeat position is used as the center of the extracted signal segment. A signal segment is extracted from the target ECG signal, and this extracted signal segment is taken as the heartbeat segment corresponding to the target heartbeat position. The duration of the heartbeat segment is the same as the duration of the preset time window.
[0096] Optionally, the duration of the preset time window is greater than 0.5 times the heartbeat interval and less than the heartbeat interval.
[0097] The heartbeat interval is the duration between two heartbeats.
[0098] For S133, the heartbeat segments corresponding to the heartbeat position set are sorted in chronological order, and the sorted heartbeat segments are used as the heartbeat segment time sequence set.
[0099] In one embodiment, the step of generating a morphological amplitude encoding vector for the target heartbeat segment as the target morphological amplitude encoding vector includes:
[0100] S21: Sample the target heartbeat fragments at preset time intervals to obtain a sampling result set;
[0101] S22: Calculate the average value of the sampled result set to obtain the target average value;
[0102] S23: Calculate the amplitude based on the target average value and the sampling result set to obtain the target amplitude;
[0103] S24: Based on the target average value and the target amplitude, the sampling result set is normalized to obtain the amplitude normalization result;
[0104] S25: Perform angle mapping based on the target amplitude and the preset maximum amplitude of lossless encoding to obtain the target angle;
[0105] S26: Calculate the target cosine and target sine values for the target angle respectively;
[0106] S27: The amplitude normalization result, the target cosine value, and the target sine value are concatenated into vectors to obtain the target shape amplitude encoding vector.
[0107] This embodiment separates the morphological information (i.e., amplitude normalization result) and amplitude information (i.e., target amplitude) of the heartbeat segment, which facilitates the relatively independent evaluation of waveform morphological and amplitude differences. Using the target cosine and sine values of the rotation angle (i.e., target angle) to represent amplitude information facilitates subsequent calculation of amplitude similarity using vector dot products. By encoding the morphological and amplitude information into a target morphological amplitude encoding vector, subsequent calculation only requires the dot product of the heartbeat template's morphological amplitude encoding vector and the target morphological amplitude encoding vector to obtain the similarity between the target heartbeat segment and the heartbeat template. This eliminates the need for similarity calculations using Pearson correlation, Euclidean distance, and wavelet feature methods, thus avoiding the technical problems of high computational load, neglecting waveform amplitude differences, and easily ignoring waveform morphological differences when amplitude is low.
[0108] For S21, the target heartbeat segment is sampled at a preset time interval, and the signal obtained from each sampling point is taken as a sampling result, and the sampling results are taken as a sampling result set.
[0109] For S22, the average value of all sampling results in the sampling result set is calculated, and the calculated average value is used as the target average value.
[0110] Formula for calculating the target average for:
[0111]
[0112] Where N is the number of sampling results in the sampling result set, x i It is the i-th sampling result in the sampling result set.
[0113] For S23, the target average value and the sampling result set are input into a preset amplitude algorithm to calculate the amplitude, and the calculated amplitude is used as the target amplitude.
[0114] For S24, the target average value, the target amplitude, and each sampling result in the sampling result set are input into a preset amplitude normalization function, each calculated data is taken as a normalization result, and each normalization result is taken as the amplitude normalization result.
[0115] In other words, the standardized results in the amplitude standardization results correspond one-to-one with the sampling results in the sampling result set.
[0116] For S25, based on the maximum amplitude of the lossless encoding, the target amplitude is rotated and mapped, and the mapped angle is used as the target angle.
[0117] The maximum amplitude value for lossless encoding refers to the maximum amplitude value that can be used for lossless encoding. The maximum amplitude value for lossless encoding is a pre-specified constant.
[0118] For S26, the cosine value of the target angle is calculated, and the calculated cosine value is used as the target cosine value; the sine value of the target angle is calculated, and the calculated sine value is used as the target sine value.
[0119] For S27, the amplitude normalization result, the target cosine value, and the target sine value are sequentially concatenated, and the concatenated vector is used as the target shape amplitude encoding vector.
[0120] Optionally, the target shape amplitude encoding vector is in, It is the Nth standardized result in the amplitude normalization results, cosθ x It is the target cosine value, sinθ x It is the target sine value.
[0121] In one embodiment, the formula for calculating the target amplitude is amp. x for:
[0122]
[0123] The formula for calculating the i-th normalized result in the amplitude normalization results. for:
[0124]
[0125] The formula for calculating the target angle is θ. x for:
[0126]
[0127] Where N is the number of sampling results in the sampling result set, x i It is the i-th sampling result in the sampling result set. The target average value is eps, which is a preset constant to avoid division by zero, max() calculates the maximum value, and amp max is the maximum amplitude value of the lossless encoding, min() is used to calculate the minimum value, and π is the mathematical constant pi.
[0128] This embodiment separates the morphological information (i.e., amplitude normalization result) and amplitude information (i.e., target amplitude) of the heartbeat fragment. The amplitude information is represented by the target cosine and target sine values of the rotation angle (i.e., target angle). This provides a foundation for subsequently calculating the inner product of the morphological amplitude encoding vector of the heartbeat template and the target morphological amplitude encoding vector to obtain the similarity between the target heartbeat fragment and the heartbeat template. By using a preset constant to avoid division by zero, division errors are avoided, and the stability of operation is improved.
[0129] eps is a very small value. For example, eps can be set to 0.00001.
[0130] In one embodiment, the step of determining the target heartbeat type corresponding to the target heartbeat segment based on the inner product set includes:
[0131] S41: Find the inner product value with the largest value from the set of inner product values, and use it as the target inner product value;
[0132] S42: Use the heartbeat template vector corresponding to the target inner product value as the target heartbeat template vector;
[0133] S43: Use the heartbeat type corresponding to the target heartbeat template vector as the target heartbeat type corresponding to the target heartbeat segment.
[0134] This embodiment uses the heartbeat type corresponding to the heartbeat template vector with the largest inner product value as the target heartbeat type corresponding to the target heartbeat segment, thereby realizing the determination of heartbeat type based on inner product value. It does not require similarity calculation using Pearson correlation method, Euclidean distance method and wavelet feature method, thus avoiding the technical problems of large amount of calculation, ignoring the amplitude difference of waveform and the easy to ignore the morphological difference of waveform when the amplitude is low.
[0135] For S41, find the inner product value with the largest value from the set of inner product values, and use the found inner product value as the target inner product value.
[0136] Optionally, the inner product value with the largest value is found from the set of inner product values, and this found inner product value is used as a candidate inner product value; if the candidate inner product value is greater than a preset inner product value threshold, then the candidate inner product value is used as the target inner product value. This further improves the accuracy of the determined target inner product value.
[0137] For S42, the target inner product value corresponding to the heartbeat template vector is used as the target heartbeat template vector, thereby finding the morphological amplitude encoding vector of the most similar heartbeat template.
[0138] For S43, the heartbeat type corresponding to the target heartbeat template vector in the heartbeat template vector set is taken as the target heartbeat type corresponding to the target heartbeat segment.
[0139] In one embodiment, the step of performing arrhythmia analysis based on each target heartbeat type corresponding to the heartbeat segment time series set to obtain arrhythmia analysis results includes:
[0140] S51: Perform arrhythmia analysis based on the target heartbeat types corresponding to the heartbeat segment time sequence set and the preset clinical rules to obtain the arrhythmia analysis results.
[0141] This embodiment performs arrhythmia analysis based on the target heartbeat types corresponding to the heartbeat segment time series set, thereby achieving accurate heartbeat type application for arrhythmia analysis and improving the accuracy of arrhythmia analysis results.
[0142] For S51, the target heartbeat types corresponding to the heartbeat segment time series are taken as heartbeat type combinations, and the heartbeat type combinations are analyzed and matched in a preset clinical rule mapping table. The matched data is taken as the arrhythmia analysis result.
[0143] Reference Figure 2 This application also proposes a cardiac arrhythmia analysis device based on heartbeat classification, the device comprising:
[0144] The data acquisition module 100 is used to acquire the time series set of heartbeat segments corresponding to the target electrocardiogram signal;
[0145] The target morphological amplitude encoding vector generation module 200 is used to generate a morphological amplitude encoding vector for a target heartbeat segment, which serves as the target morphological amplitude encoding vector. The target heartbeat segment is any heartbeat segment in the heartbeat segment time sequence set.
[0146] The inner product value set determination module 300 is used to perform inner product calculation on the target morphology amplitude encoding vector and each heartbeat template vector in the preset heartbeat template vector set to obtain the inner product value set.
[0147] The target heartbeat type determination module 400 is used to determine the target heartbeat type corresponding to the target heartbeat segment based on the inner product set;
[0148] The arrhythmia analysis result determination module 500 is used to perform arrhythmia analysis based on each target heartbeat type corresponding to the heartbeat segment time sequence set, and obtain arrhythmia analysis results.
[0149] This embodiment calculates the inner product value by using the morphological amplitude encoding vector and the heartbeat template vector, eliminating the need for similarity calculations using Pearson correlation, Euclidean distance, and wavelet feature methods. This avoids the technical problems of high computational cost, neglecting waveform amplitude differences, and easily ignoring waveform morphological differences when the amplitude is low. Furthermore, generating morphological amplitude encoding vectors for the target heartbeat segment and calculating the inner product value based on the morphological amplitude encoding vector and the heartbeat template vector is a simple and computationally inefficient process, making this application suitable for real-time monitoring scenarios.
[0150] In one embodiment, the data acquisition module 100 includes: a target electrocardiogram signal determination submodule and a heartbeat segment time series determination submodule;
[0151] The target ECG signal determination submodule is used to acquire an initial ECG signal, and perform bandpass filtering on the initial ECG signal using a preset filtering range to obtain the target ECG signal;
[0152] The cardiac segment timing set determination submodule is used to extract cardiac segments from the target electrocardiogram signal to obtain the cardiac segment timing set.
[0153] In one embodiment, the above-mentioned cardiac segment timing set determination submodule includes: a cardiac position set determination unit, a cardiac segment extraction unit, and a cardiac segment timing set determination unit;
[0154] The heartbeat location set determination unit is used to perform heartbeat location detection on the target electrocardiogram signal using the Pan-Tompkins algorithm to obtain a heartbeat location set.
[0155] The heartbeat segment extraction unit is used to extract the heartbeat segment corresponding to the target heartbeat position from the target electrocardiogram signal by using a preset time window and taking the target heartbeat position as the center position, wherein the target heartbeat position is any one of the heartbeat positions in the set of heartbeat positions;
[0156] The heartbeat segment timing set determination unit is used to determine the heartbeat segment timing set based on each heartbeat segment corresponding to the heartbeat position set.
[0157] In one embodiment, the target shape amplitude encoding vector generation module 200 includes: a sampling result set determination submodule, a target average value determination submodule, a target amplitude determination submodule, an amplitude normalization result determination submodule, a target angle determination submodule, and a target shape amplitude encoding vector determination submodule;
[0158] The sampling result set determination submodule is used to sample the target heartbeat fragment at preset time intervals to obtain a sampling result set;
[0159] The target average value determination submodule is used to calculate the average value of the sampling result set to obtain the target average value;
[0160] The target amplitude determination submodule is used to calculate the target amplitude based on the target average value and the sampling result set.
[0161] The amplitude standardization result determination submodule is used to perform amplitude standardization on the sampling result set according to the target average value and the target amplitude to obtain the amplitude standardization result;
[0162] The target angle determination submodule is used to perform angle mapping based on the target amplitude and the preset lossless encoding maximum amplitude to obtain the target angle;
[0163] The target shape amplitude encoding vector determination submodule is used to calculate the target cosine value and the target sine value for the target angle, and to concatenate the amplitude standardization result, the target cosine value and the target sine value into a vector to obtain the target shape amplitude encoding vector.
[0164] In one embodiment, the formula for calculating the target amplitude is amp. x for:
[0165]
[0166] The formula for calculating the i-th normalized result in the amplitude normalization results. for:
[0167]
[0168] The formula for calculating the target angle is θ. x for:
[0169]
[0170] Where N is the number of sampling results in the sampling result set, x i It is the i-th sampling result in the sampling result set, x is the target average value, eps is a preset constant used to avoid division by zero, max() is used to calculate the maximum value, and amp max is the maximum amplitude value of the lossless encoding, min() is used to calculate the minimum value, and π is the mathematical constant pi.
[0171] In one embodiment, the target heartbeat type determination module 400 includes: a target heartbeat template vector determination submodule and a target heartbeat type determination submodule;
[0172] The target heartbeat template vector determination submodule is used to find the inner product value with the largest value from the inner product value set, and use it as the target inner product value, and use the heartbeat template vector corresponding to the target inner product value as the target heartbeat template vector.
[0173] The target heartbeat type determination submodule is used to take the heartbeat type corresponding to the target heartbeat template vector as the target heartbeat type corresponding to the target heartbeat segment.
[0174] In one embodiment, the above-mentioned arrhythmia analysis result determination module 500 includes: an arrhythmia analysis result determination submodule;
[0175] The arrhythmia analysis result determination submodule is used to perform arrhythmia analysis based on the target heartbeat type corresponding to the heartbeat segment time sequence set and preset clinical rules to obtain the arrhythmia analysis result.
[0176] Reference Figure 3 This application also provides a computer device, which may be a server, and its internal structure may be as follows: Figure 3 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data such as a cardiac arrhythmia analysis method based on heartbeat classification. The network interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a cardiac arrhythmia analysis method based on heartbeat classification. The arrhythmia analysis method based on heartbeat classification includes: acquiring a time series set of heartbeat segments corresponding to a target electrocardiogram signal; generating a morphological amplitude encoding vector for the target heartbeat segment as a target morphological amplitude encoding vector, wherein the target heartbeat segment is any heartbeat segment in the heartbeat segment time series set; calculating the inner product between the target morphological amplitude encoding vector and each heartbeat template vector in a preset heartbeat template vector set to obtain an inner product value set; determining the target heartbeat type corresponding to the target heartbeat segment based on the inner product value set; and performing arrhythmia analysis based on each target heartbeat type corresponding to the heartbeat segment time series set to obtain the arrhythmia analysis result.
[0177] This embodiment calculates the inner product value by using the morphological amplitude encoding vector and the heartbeat template vector, eliminating the need for similarity calculations using Pearson correlation, Euclidean distance, and wavelet feature methods. This avoids the technical problems of high computational cost, neglecting waveform amplitude differences, and easily ignoring waveform morphological differences when the amplitude is low. Furthermore, generating morphological amplitude encoding vectors for the target heartbeat segment and calculating the inner product value based on the morphological amplitude encoding vector and the heartbeat template vector is a simple and computationally inefficient process, making this application suitable for real-time monitoring scenarios.
[0178] In one embodiment, the step of obtaining the time series set of heartbeat segments corresponding to the target electrocardiogram signal includes: obtaining an initial electrocardiogram signal; performing bandpass filtering on the initial electrocardiogram signal using a preset filtering range to obtain the target electrocardiogram signal; and extracting the heartbeat segments from the target electrocardiogram signal to obtain the time series set of heartbeat segments.
[0179] In one embodiment, the step of extracting the heartbeat segments from the target electrocardiogram signal to obtain the heartbeat segment time series set includes: using the Pan-Tompkins algorithm to detect the heartbeat position of the target electrocardiogram signal to obtain a heartbeat position set; using a preset time window and a method with the target heartbeat position as the center position to extract the heartbeat segments corresponding to the target heartbeat positions from the target electrocardiogram signal, wherein the target heartbeat position is any heartbeat position in the heartbeat position set; and determining the heartbeat segment time series set based on each heartbeat segment corresponding to the heartbeat position set.
[0180] In one embodiment, the step of generating a morphological amplitude encoding vector for a target heartbeat fragment as a target morphological amplitude encoding vector includes: sampling the target heartbeat fragment at preset time intervals to obtain a sampling result set; calculating the average value of the sampling result set to obtain a target average value; calculating the amplitude based on the target average value and the sampling result set to obtain a target amplitude; standardizing the amplitude of the sampling result set based on the target average value and the target amplitude to obtain an amplitude standardization result; mapping the target amplitude to a preset lossless encoding maximum amplitude to obtain a target angle; calculating the target cosine value and the target sine value for the target angle; and concatenating the amplitude standardization result, the target cosine value, and the target sine value into a vector to obtain the target morphological amplitude encoding vector.
[0181] In one embodiment, the formula for calculating the target amplitude is amp. x for:
[0182]
[0183] The formula for calculating the i-th normalized result in the amplitude normalization results. for:
[0184]
[0185] The formula for calculating the target angle is θ. x for:
[0186]
[0187] Where N is the number of sampling results in the sampling result set, x i It is the i-th sampling result in the sampling result set. The target average value is eps, which is a preset constant to avoid division by zero, max() calculates the maximum value, and amp max is the maximum amplitude value of the lossless encoding, min() is used to calculate the minimum value, and π is the mathematical constant pi.
[0188] In one embodiment, the step of determining the target heartbeat type corresponding to the target heartbeat segment based on the inner product set includes: finding the inner product value with the largest value from the inner product set as the target inner product value; using the heartbeat template vector corresponding to the target inner product value as the target heartbeat template vector; and using the heartbeat type corresponding to the target heartbeat template vector as the target heartbeat type corresponding to the target heartbeat segment.
[0189] In one embodiment, the step of performing arrhythmia analysis based on each of the target heartbeat types corresponding to the heartbeat segment time series set to obtain arrhythmia analysis results includes: performing arrhythmia analysis based on each of the target heartbeat types corresponding to the heartbeat segment time series set and preset clinical rules to obtain the arrhythmia analysis results.
[0190] One embodiment of this application also provides a computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements a cardiac arrhythmia analysis method based on heartbeat classification, including the following steps: obtaining a time series set of heartbeat segments corresponding to a target electrocardiogram signal; generating a morphological amplitude encoding vector for the target heartbeat segments as a target morphological amplitude encoding vector, wherein the target heartbeat segment is any heartbeat segment in the heartbeat segment time series set; calculating the inner product between the target morphological amplitude encoding vector and each heartbeat template vector in a preset heartbeat template vector set to obtain an inner product value set; determining the target heartbeat type corresponding to the target heartbeat segment based on the inner product value set; and performing arrhythmia analysis based on each target heartbeat type corresponding to the heartbeat segment time series set to obtain arrhythmia analysis results.
[0191] The aforementioned arrhythmia analysis method based on heartbeat classification calculates the inner product of the morphological amplitude encoding vector and the heartbeat template vector, eliminating the need for similarity calculations using Pearson correlation, Euclidean distance, and wavelet feature methods. This avoids the technical problems of high computational cost, neglecting waveform amplitude differences, and easily ignoring waveform morphological differences when the amplitude is low. Furthermore, the method generates morphological amplitude encoding vectors for the target heartbeat segment and calculates the inner product of the morphological amplitude encoding vector and the heartbeat template vector, resulting in a simple procedure with low computational cost, making this application suitable for real-time monitoring scenarios.
[0192] In one embodiment, the step of obtaining the time series set of heartbeat segments corresponding to the target electrocardiogram signal includes: obtaining an initial electrocardiogram signal; performing bandpass filtering on the initial electrocardiogram signal using a preset filtering range to obtain the target electrocardiogram signal; and extracting the heartbeat segments from the target electrocardiogram signal to obtain the time series set of heartbeat segments.
[0193] In one embodiment, the step of extracting the heartbeat segments from the target electrocardiogram signal to obtain the heartbeat segment time series set includes: using the Pan-Tompkins algorithm to detect the heartbeat position of the target electrocardiogram signal to obtain a heartbeat position set; using a preset time window and a method with the target heartbeat position as the center position to extract the heartbeat segments corresponding to the target heartbeat positions from the target electrocardiogram signal, wherein the target heartbeat position is any heartbeat position in the heartbeat position set; and determining the heartbeat segment time series set based on each heartbeat segment corresponding to the heartbeat position set.
[0194] In one embodiment, the step of generating a morphological amplitude encoding vector for a target heartbeat fragment as a target morphological amplitude encoding vector includes: sampling the target heartbeat fragment at preset time intervals to obtain a sampling result set; calculating the average value of the sampling result set to obtain a target average value; calculating the amplitude based on the target average value and the sampling result set to obtain a target amplitude; standardizing the amplitude of the sampling result set based on the target average value and the target amplitude to obtain an amplitude standardization result; mapping the target amplitude to a preset lossless encoding maximum amplitude to obtain a target angle; calculating the target cosine value and the target sine value for the target angle; and concatenating the amplitude standardization result, the target cosine value, and the target sine value into a vector to obtain the target morphological amplitude encoding vector.
[0195] In one embodiment, the formula for calculating the target amplitude is amp. x for:
[0196]
[0197] The formula for calculating the i-th normalized result in the amplitude normalization results. for:
[0198]
[0199] The formula for calculating the target angle is θ. x for:
[0200]
[0201] Where N is the number of sampling results in the sampling result set, x i It is the i-th sampling result in the sampling result set. The target average value is eps, which is a preset constant to avoid division by zero, max() calculates the maximum value, and amp max is the maximum amplitude value of the lossless encoding, min() is used to calculate the minimum value, and π is the mathematical constant pi.
[0202] In one embodiment, the step of determining the target heartbeat type corresponding to the target heartbeat segment based on the inner product set includes: finding the inner product value with the largest value from the inner product set as the target inner product value; using the heartbeat template vector corresponding to the target inner product value as the target heartbeat template vector; and using the heartbeat type corresponding to the target heartbeat template vector as the target heartbeat type corresponding to the target heartbeat segment.
[0203] In one embodiment, the step of performing arrhythmia analysis based on each of the target heartbeat types corresponding to the heartbeat segment time series set to obtain arrhythmia analysis results includes: performing arrhythmia analysis based on each of the target heartbeat types corresponding to the heartbeat segment time series set and preset clinical rules to obtain the arrhythmia analysis results.
[0204] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media provided in this application and in the embodiments may 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. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
[0205] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.
[0206] The above description is only a preferred embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural changes made based on the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.
Claims
1. A method for analyzing cardiac arrhythmias based on heart rate classification, characterized in that, The method includes: Obtain the time sequence set of heartbeat segments corresponding to the target electrocardiogram signal; A morphological amplitude encoding vector is generated for the target heartbeat segment, which serves as the target morphological amplitude encoding vector, wherein the target heartbeat segment is any heartbeat segment in the heartbeat segment time sequence set; The inner product of the target shape amplitude encoding vector and each heartbeat template vector in the preset heartbeat template vector set is calculated to obtain the inner product value set. The target heartbeat type corresponding to the target heartbeat segment is determined based on the inner product set; Arrhythmia analysis is performed based on the target heart beat types corresponding to the heart beat segment time series set to obtain arrhythmia analysis results. The step of generating a morphological amplitude encoding vector for the target heartbeat fragment, as the target morphological amplitude encoding vector, includes: The target heartbeat fragments are sampled at preset time intervals to obtain a sampling result set; The average value of the sampled result set is calculated to obtain the target average value; The target amplitude is obtained by calculating the amplitude based on the target average value and the sampling result set; Based on the target average value and the target amplitude, the sampling result set is normalized to obtain the amplitude normalization result; The target angle is obtained by mapping the target amplitude and the preset maximum amplitude of lossless encoding. The target angle is calculated by performing target cosine and target sine values respectively; The amplitude normalization result, the target cosine value, and the target sine value are concatenated into vectors to obtain the target shape amplitude encoding vector.
2. The arrhythmia analysis method based on heart rate classification according to claim 1, characterized in that, The step of obtaining the time series set of heartbeat segments corresponding to the target electrocardiogram signal includes: Acquire initial electrocardiogram (ECG) signals; The initial electrocardiogram (ECG) signal is bandpass filtered using a preset filtering range to obtain the target ECG signal. The target electrocardiogram signal is subjected to heartbeat segment extraction to obtain the heartbeat segment time series.
3. The arrhythmia analysis method based on heart rate classification according to claim 2, characterized in that, The step of extracting the heartbeat segments from the target electrocardiogram signal to obtain the heartbeat segment time series set includes: The target electrocardiogram signal is subjected to heartbeat location detection to obtain a heartbeat location set; Using a preset time window and taking the target heartbeat position as the center position, the heartbeat segment corresponding to the target heartbeat position is extracted from the target electrocardiogram signal, wherein the target heartbeat position is any heartbeat position in the set of heartbeat positions; The temporal set of the heartbeat segments is determined based on each heartbeat segment corresponding to the heartbeat position set.
4. The arrhythmia analysis method based on heart rate classification according to claim 1, characterized in that, The formula for calculating the target amplitude for: The formula for calculating the i-th normalized result in the amplitude normalization results. for: The formula for calculating the target angle for: Where N is the number of sampling results in the sampling result set. It is the i-th sampling result in the sampling result set. It is the target average value. It is a preset constant used to avoid division by zero, and max() calculates the maximum value. This is the maximum amplitude value of the lossless encoding, and min() calculates the minimum value. It is pi (π).
5. The arrhythmia analysis method based on heart rate classification according to claim 1, characterized in that, The step of determining the target heartbeat type corresponding to the target heartbeat segment based on the inner product set includes: Find the inner product value with the largest value from the set of inner product values, and use it as the target inner product value; The target inner product value is used as the target heartbeat template vector; The heartbeat type corresponding to the target heartbeat template vector is taken as the target heartbeat type corresponding to the target heartbeat fragment.
6. The arrhythmia analysis method based on heart rate classification according to claim 1, characterized in that, The step of performing arrhythmia analysis based on each target heartbeat type corresponding to the heartbeat segment time series set to obtain arrhythmia analysis results includes: Based on the target heartbeat type corresponding to the heartbeat segment time sequence set and the preset clinical rules, arrhythmia analysis is performed to obtain the arrhythmia analysis results.
7. A cardiac arrhythmia analysis device based on heartbeat classification, used to implement the cardiac arrhythmia analysis method based on heartbeat classification as described in any one of claims 1-6, characterized in that, The device includes: The data acquisition module is used to acquire the time series set of heartbeat segments corresponding to the target electrocardiogram signal; The target morphological amplitude encoding vector generation module is used to generate a morphological amplitude encoding vector for a target heartbeat segment, which serves as the target morphological amplitude encoding vector. The target heartbeat segment is any heartbeat segment in the heartbeat segment time sequence set. The inner product value set determination module is used to perform inner product calculation on the target morphology amplitude encoding vector and each heartbeat template vector in the preset heartbeat template vector set to obtain the inner product value set. The target heartbeat type determination module is used to determine the target heartbeat type corresponding to the target heartbeat segment based on the inner product set; The arrhythmia analysis result determination module is used to perform arrhythmia analysis based on each target heartbeat type corresponding to the heartbeat segment time series set, and obtain arrhythmia analysis results.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.
Citation Information
Patent Citations
Automatic electrocardioelectrode placement error detection method based on kernel function classification algorithm
CN104473629A