Self-adaptive R-wave detection method and device based on multi-channel electrocardiosignal

Abstract

The invention provides a self-adaptive R-wave detection method and device based on multi-channel electrocardiosignals. The method comprises the steps that the multi-channel electrocardiosignals are obtained; performing sub-channel preprocessing on the multi-channel electrocardiogram signal to obtain a preprocessed multi-channel electrocardiogram signal; performing inter-channel time alignment on the preprocessed multi-channel electrocardiogram signal to obtain an aligned multi-channel electrocardiogram signal; based on signal quality evaluation, performing weighted fusion on the aligned multi-channel electrocardiogram signals to generate a fusion signal; carrying out adaptive R-wave detection on the fused signal to obtain a preliminary R-wave position sequence; and based on the preliminary R-wave position sequence, performing multi-channel fusion positioning verification on an original multi-channel signal, and outputting a final R-wave position sequence. According to the method, the robustness and the accuracy of R wave detection in the pathological electrocardiogram are improved through alignment, quality evaluation weighting and fusion of the multi-channel signals.

Description

Technical Field

[0001] This application relates to the field of electrocardiogram (ECG) signal detection technology, and in particular to an adaptive R-wave detection method and device based on multi-channel ECG signals. Background Technology

[0002] The R wave in an electrocardiogram (ECG) signal is a point change that occurs during ventricular depolarization. Its detection is crucial in ECG signal analysis, processing, and diagnosis, and it has significant diagnostic value, especially for various arrhythmias and atrial fibrillation.

[0003] Current ECG R-wave detection technologies typically involve: adaptive filtering and dynamic parameter optimization: removing baseline drift and high-frequency noise through median filtering and moving average filtering with a moving window frame, combined with dynamic threshold adjustment strategies to improve R-wave detection accuracy; wavelet transform and multi-scale analysis: using wavelet transform (such as the Mexican Hat wavelet) combined with modulus maxima to detect R-waves, while employing a fusion method of stationary wavelet transform and adaptive filtering to reduce waveform distortion; machine learning and multi-algorithm fusion: using deep learning models such as SE-ResNet and LSTM to improve R-wave recognition rate, or integrating traditional algorithms such as wavelet transform and difference methods to optimize detection stability.

[0004] However, the aforementioned existing technologies have several shortcomings in practical applications: Due to the diverse morphologies of actual human electrocardiogram (ECG) signals, many pathological ECGs are irregular, with potentially low R-wave amplitude, biphasic peaks, negative Q waves, S waves, or interference significantly affecting R-wave detection; most ECG devices use three channels or standard detection channels for R-wave localization, which can lead to deviations in the localization of pathological ECGs; some heart rate detection methods using the frequency domain cannot achieve detailed localization and QRS complex feature localization required for professional ECG analysis; secondly, to improve analysis efficiency during ECG calculation and analysis, channels with lower signal amplitudes are often discarded, and only pre-set standard analysis channels are selected, resulting in the loss of some effective information. Summary of the Invention

[0005] In view of this, this application proposes an adaptive R-wave detection method and device based on multi-channel electrocardiogram signals, which improves the robustness and accuracy of R-wave detection in pathological electrocardiograms by aligning, weighting and fusing multi-channel signals.

[0006] In a first aspect, this application provides an adaptive R-wave detection method based on multi-channel electrocardiogram signals, comprising:

[0007] Acquire multi-channel electrocardiogram signals;

[0008] The multi-channel electrocardiogram signal is preprocessed by channel to obtain the preprocessed multi-channel electrocardiogram signal;

[0009] The preprocessed multi-channel electrocardiogram (ECG) signals are time-aligned between channels to obtain aligned multi-channel ECG signals.

[0010] Based on signal quality assessment, the aligned multi-channel electrocardiogram signals are weighted and fused to generate a fused signal.

[0011] Adaptive R-wave detection is performed on the fused signal to obtain a preliminary R-wave position sequence;

[0012] Based on the preliminary R-wave position sequence, multi-channel fusion positioning verification is performed on the original multi-channel signal to output the final R-wave position sequence.

[0013] Therefore, the adaptive R-wave detection method based on multi-channel electrocardiogram (ECG) signals provided in this application firstly processes the multi-channel ECG signals separately through channel-specific preprocessing. This removes noise and other interference factors based on the characteristics of each channel signal, improving the signal quality of each channel. Then, the preprocessed multi-channel ECG signals are time-aligned between channels to ensure temporal consistency and avoid deviations in subsequent processing due to time differences. Based on signal quality assessment, the aligned multi-channel ECG signals are weighted and fused to generate a fused signal. Adaptive R-wave detection is then performed on the fused signal to obtain a preliminary R-wave position sequence. This sequence is then used for multi-channel fusion positioning verification on the original multi-channel signals. By using multi-channel information to mutually verify each other, the probability of false detection and false negative detection is effectively reduced, resulting in a more accurate final R-wave position sequence. This application solves the problem of insufficient detection accuracy in existing technologies when dealing with diverse morphologies, especially pathological irregular ECGs, improving the robustness and accuracy of R-wave detection in pathological ECGs.

[0014] Optionally, the channel-specific preprocessing includes performing at least one of the following independently on the signal of each channel:

[0015] Channel-by-channel denoising: The signal is decomposed using Gaussian wavelet transform, and the wavelet coefficients within a specified scale range are weighted and fused to reconstruct the signal, thereby enhancing the QRS group and suppressing noise.

[0016] Channel-specific normalization: Based on the standard deviation or absolute median difference of the signal within the sliding window, the signal amplitude is normalized to a uniform range;

[0017] Power frequency filtering: An adaptive notch filter based on the least mean square algorithm is used to filter out power frequency interference;

[0018] Baseline correction: Correcting the signal baseline by morphological filtering or by ignoring low-frequency approximation coefficients that contain baseline drift in wavelet reconstruction;

[0019] Gain adjustment: adaptively adjust the gain according to the dynamic range of the signal so that the amplitude of the QRS complex falls within a preset range.

[0020] As described above, the present application can perform channel-by-channel preprocessing on multi-channel electrocardiogram signals through methods such as channel-by-channel denoising, channel-by-channel normalization, power frequency filtering, baseline correction, and gain adjustment. Channel-by-channel denoising uses Gaussian wavelet transform to decompose the signal, weights and fuses the wavelet coefficients at specified scales for reconstruction, effectively enhancing the QRS complex and suppressing various noises, and improving the signal-to-noise ratio; Channel-by-channel normalization is based on the standard deviation or absolute median difference of the signals within a sliding window, unifies the amplitudes of the signals in different channels into the same range, eliminates amplitude differences, enhances the comparability of the signals in each channel, and facilitates subsequent fusion analysis; Power frequency filtering uses an adaptive notch filter based on the least mean square algorithm, which can accurately and adaptively filter out power frequency interference and improve the signal purity; Baseline correction effectively eliminates the influence of baseline drift on the signal through morphological filtering or ignoring the low-frequency approximation coefficients containing baseline drift in wavelet reconstruction, making the signal more truly reflect the cardiac electrical activity; Gain adjustment is adaptively adjusted according to the signal dynamic range, so that the amplitude of the QRS complex is within a suitable preset range, optimizes the amplitude of the QRS complex, improves the performance of the subsequent adaptive R-wave detection algorithm, reduces the probability of missed detection and false detection, and overall provides a high-quality, reliable and easy-to-process signal basis for the subsequent accurate detection of the R wave.

[0021] Optionally, the channel-by-channel denoising uses Gaussian wavelets to decompose the signal into N layers, selects the wavelet coefficients from the Kth layer to the Mth layer for weighted fusion to reconstruct the signal, where N, K, and M are natural numbers, and 3 ≤ K < M ≤ N, and the Kth layer to the Mth layer cover the main energy frequency band of the QRS complex.

[0022] As described above, the channel-by-channel denoising scheme of the present application decomposes the signal into N layers using Gaussian wavelets, and accurately selects the detail coefficients from the Kth layer to the Mth layer that cover the main energy frequency band of the QRS complex for weighted fusion reconstruction, achieving the technical effect of targeted denoising on the premise of maximizing the retention of the true QRS waveform characteristics. It can not only effectively filter out high-frequency noises such as electromyogram interference, but also simultaneously suppress baseline drift, and at the same time avoids the defects of traditional global filtering or fixed threshold denoising methods that are prone to cause blurring of the QRS complex edges or amplitude attenuation, significantly improving the signal-to-noise ratio and feature clarity of the signal, and providing a crucial high-quality signal basis for the subsequent accurate detection of the R wave.

[0023] Optionally, the time alignment includes:

[0024] 5]Select a channel with the highest signal-to-noise ratio from the preprocessed multi-channel electrocardiogram signals as the reference channel; [[ID=]6]

[0025] Calculate the cross-correlation function between the remaining channels and the reference channel, and determine the delay of each channel relative to the reference channel by locating the peak value of the cross-correlation function;

[0026] Based on the aforementioned delay, phase shifting is performed on the remaining channels using linear interpolation to achieve sampling point-level time alignment for all channels.

[0027] As described above, this time alignment scheme uses the channel with the highest signal-to-noise ratio as the precise timing benchmark. It objectively quantifies the relative delay between channels by detecting the peak value of the cross-correlation function and uses linear interpolation for lossless phase shifting. This eliminates time misalignment caused by differences in acquisition hardware or transmission paths at the molecular level, ensuring strict alignment of multi-channel signals from different spatial perspectives in the heartbeat cycle. This lays a precise timing foundation for subsequent effective signal fusion, energy superposition, and cross-channel consistency verification, fundamentally avoiding fuzzy fused signals, feature dilution, and positioning errors caused by time misalignment.

[0028] Optionally, the signal-to-noise ratio is calculated as follows:

[0029] Calculate the ratio of the total energy of the coefficients of the signal at the wavelet scale representing the QRS group energy to the total energy of the coefficients at the wavelet scale representing the noise energy.

[0030] As described above, by calculating the ratio of coefficient energy at specific wavelet scales (such as layers D3-D4) representing QRS group energy to that at specific scales (such as layers D1 and D10) representing noise energy, this method can intelligently distinguish and quantify useful physiological information and irrelevant noise interference in the signal, thereby achieving an essential, feature-based assessment of signal quality. This calculation method, which is closely related to the physical characteristics of QRS groups, can more accurately and stably reflect the actual value of each channel in the R-wave detection task compared to traditional full-band or fixed-band signal-to-noise ratio estimation. It provides an objective and reliable decision-making basis for subsequent selection of the optimal reference channel for time alignment and execution of high-quality weighted fusion, effectively avoiding systematic errors introduced by the misselection of low-quality channels as the benchmark.

[0031] Optionally, the weighted fusion includes:

[0032] Determine candidate channels: Calculate the cross-correlation coefficient between each channel and a reference channel, and determine the channels with cross-correlation coefficients higher than a preset threshold as candidate channels;

[0033] Weighting: Based on the QRS group energy of each candidate channel within the local window, a fusion weight is assigned to each candidate channel, where the channel with higher QRS group energy is assigned a higher weight.

[0034] Data weighted fusion: The signal sampling point values ​​of each candidate channel are multiplied by their corresponding fusion weights and then summed to generate the fused signal.

[0035] As described above, the weighted fusion scheme of this application first filters out candidate channels with waveform morphology consistent with the reference channel by using a cross-correlation coefficient threshold, effectively eliminating signal distortion channels caused by local interference or poor electrode contact, and ensuring the reliability of the signal sources participating in the fusion. Then, it assigns weights based on the QRS group energy magnitude of each candidate channel within a local window, so that the channel with more significant R-wave characteristics and higher signal-to-noise ratio dominates in the fused signal, which is equivalent to constructing a signal synthesizer that can adaptively focus on the clearest cardiac beat characteristics. Finally, through data-level weighted fusion, the QRS group energy of each high-quality channel is directionally superimposed, thereby generating a fused signal with significantly improved signal-to-noise ratio and greatly enhanced R-wave characteristics. This "super signal" provides an ideal data foundation for subsequent high-precision adaptive R-wave detection, fundamentally improving the detection capability and robustness of the entire system for weak or abnormal R-waves.

[0036] Optionally, the weight allocation further includes:

[0037] Based on the pathological characteristics of ECG leads, higher baseline weights are pre-assigned to leads with more pronounced R-wave morphology characteristics under specific pathological conditions.

[0038] As described above, this optional weight allocation enhancement scheme, based on real-time QRS energy-based weight allocation, further incorporates pathological feature knowledge of ECG leads, pre-allocating higher base weights to leads known to exhibit more typical R-wave morphology in specific cardiac diseases. This strategy enables the fusion system to not only rely on the instantaneous quality of the signal but also possess pathology-oriented intelligence, thereby ensuring the fusion signal's ability to preserve and enhance the characteristics of pathological R waves from the source. This significantly improves the algorithm's detection accuracy and pathological identification specificity when facing abnormal ECGs with clear clinical features.

[0039] Optionally, the adaptive R-wave detection includes:

[0040] The fused signal is subjected to nonlinear transformation and differential operation to enhance the QRS group characteristics;

[0041] The enhanced signal is integrated by a local moving window to concentrate the enhanced QRS group energy into a smooth pulse peak;

[0042] A dynamically updated dual-threshold strategy is used on the integrated signal to identify R-wave candidate peaks;

[0043] The positions of the R-wave candidate peaks are mapped back to the local search window of the original fused signal. The precise R-wave position is determined by detecting the local extrema within the window. The detection supports the simultaneous detection of positive and negative R-waves.

[0044] As described above, this optional adaptive R-wave detection scheme first significantly sharpens the slope characteristics of the QRS group through the synergistic effect of nonlinear transformation and differential operation, while effectively suppressing interference from non-target components such as T-waves and P-waves, thus solving the problem of insignificant features when the wave group morphology varies in traditional methods. Subsequently, a local moving window integrator is used to converge the enhanced dispersed energy into a smooth and prominent pulse peak. This transformation greatly reduces the sensitivity of subsequent detection algorithms to noise and small fluctuations, creating ideal conditions for threshold decision-making. Finally, a dynamically updated dual-threshold strategy is adopted, using the signal threshold... The collaborative judgment with the noise threshold enables the system to adaptively track the dynamic changes in signal intensity, effectively avoiding continuous false detections or missed detections caused by fixed thresholds while ensuring a high detection rate. Finally, by mapping the candidate peak position back to the original signal and performing precise local extremum positioning that supports bidirectional polarity, the detection accuracy is not only improved to the sampling point level, but more importantly, it breaks through the limitation of traditional methods that can only detect positive R waves, enabling it to accurately capture negative R waves that appear under pathological conditions (such as right ventricular hypertrophy, myocardial infarction, etc.), thereby significantly improving the universality and diagnostic value of pathological electrocardiograms.

[0045] Optionally, the multi-channel fusion positioning verification includes:

[0046] Channel-by-channel verification: The preliminary R-wave position sequence is mapped to each original preprocessed channel signal, and local extremum detection is performed independently on each channel to verify the R-wave position;

[0047] Consistency check and conflict resolution: Calculate the proportion of channels that pass the verification. If the proportion exceeds the preset threshold, the R wave is confirmed; otherwise, the conflict resolution mechanism is activated, and the verification results of the high-weight channels are adopted first.

[0048] Final positioning output: Allows different polarities of the R-wave in different channels, and confirms the true R-wave only based on the consistency of the time point of the R-wave appearance.

[0049] As described above, this optional multi-channel fusion localization verification scheme first back-maps the preliminary detection results to each original channel for independent verification through channel-specific verification. This effectively avoids localization deviations caused by minor distortions or feature smoothing in the fused signal, ensuring local optimality in each reliable channel. Then, through consistency checks based on preset thresholds and a conflict resolution mechanism that prioritizes high-weight channels, it effectively eliminates false positive peaks caused by noise and supplements the true R waves that were missed in the fused signal. Finally, by allowing different channels to have different R wave polarities and using only temporal consistency as the judgment criterion, it fundamentally acknowledges and adapts to the physiological diversity of ECG signals under different lead spatial projections, ensuring accurate identification and inclusion of pathological morphologies such as negative R waves in lead V1. Ultimately, it outputs a robust and clinically accurate R wave position sequence that has undergone cross-validation of multi-source information.

[0050] Secondly, this application provides an adaptive R-wave detection device based on multi-channel electrocardiogram signals, comprising:

[0051] The acquisition module is used to acquire multi-channel electrocardiogram signals;

[0052] The preprocessing module is used to perform channel-by-channel preprocessing on the multi-channel electrocardiogram signal to obtain the preprocessed multi-channel electrocardiogram signal.

[0053] The time alignment module is used to perform time alignment between channels of the preprocessed multi-channel electrocardiogram signal to obtain an aligned multi-channel electrocardiogram signal.

[0054] The weighted fusion module is used to perform weighted fusion on the aligned multi-channel electrocardiogram signals based on signal quality assessment to generate a fused signal.

[0055] The detection module is used to perform adaptive R-wave detection on the fused signal to obtain a preliminary R-wave position sequence;

[0056] The verification module is used to perform multi-channel fusion positioning verification on the original multi-channel signal based on the preliminary R-wave position sequence, and output the final R-wave position sequence.

[0057] Thirdly, this application provides a computing device, the computing device comprising:

[0058] processor;

[0059] Memory, used to store one or more programs;

[0060] When the processor executes one or more programs, it enables the processor to implement the above-described adaptive R-wave detection method based on multi-channel electrocardiogram signals.

[0061] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a computer, implements the aforementioned adaptive R-wave detection method based on multi-channel electrocardiogram signals.

[0062] These and other aspects of this application will become more apparent in the description of the following embodiments(s). Attached Figure Description

[0063] Figure 1 A flowchart illustrating an adaptive R-wave detection method based on multi-channel electrocardiogram signals, provided for embodiments of this application;

[0064] Figure 2 A structural diagram of an adaptive R-wave detection device based on multi-channel electrocardiogram signals provided in this application embodiment;

[0065] Figure 3 This is a structural diagram of a computing device provided in an embodiment of this application.

[0066] It should be understood that the dimensions and shapes of the block diagrams in the above structural diagrams are for reference only and should not constitute an exclusive interpretation of the embodiments of this application. The relative positions and inclusion relationships between the block diagrams presented in the structural diagrams are only schematic representations of the structural relationships between the block diagrams, and are not intended to limit the physical connection methods of the embodiments of this application. Detailed Implementation

[0067] The technical solutions provided in this application will be further described below with reference to the accompanying drawings and embodiments. It should be understood that the system architecture and business scenarios provided in the embodiments of this application are mainly for illustrating possible implementations of the technical solutions of this application and should not be construed as the sole limitation on the technical solutions of this application. Those skilled in the art will recognize that the technical solutions provided in this application are equally applicable to similar technical problems as system architectures evolve and new business scenarios emerge.

[0068] Before providing a detailed description of the specific embodiments of this application, the technical terms used in the embodiments of this application will be explained. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. In case of any inconsistency, the meaning described in this specification or derived from the content recorded in this specification shall prevail. Furthermore, the terminology used herein is for the purpose of describing the embodiments of this application only and is not intended to limit this application.

[0069] 1) QRS complex: This is a set of waveforms on an electrocardiogram (ECG) representing ventricular depolarization (i.e., the activation of ventricular muscle cells by electrical signals, initiating contraction). It is the most prominent and sharpest part of the entire ECG cycle. Among them:

[0070] Q wave: The first downward-pointing negative wave in the QRS complex (if present).

[0071] R wave: The first upward positive wave in the QRS complex (usually the highest wave).

[0072] S-wave: A downward-sloping negative wave following the R-wave (if it exists).

[0073] 2) Multichannel electrocardiogram (ECG) signals: These refer to ECG signals simultaneously acquired from multiple electrode leads in the human body. Each lead records the projection of cardiac electrical activity in different spatial directions. These signals together constitute a multi-angle, three-dimensional observation of cardiac electrophysiological activity. In ECG equipment, one channel typically corresponds to one lead, that is, the potential difference signal between a pair of electrodes. For example, a 3-lead ECG contains signals from 3 channels, and a 12-lead ECG contains signals from 12 channels.

[0074] 3) Electrocardiogram rhythm signal: This is a commonly used term in electrocardiogram (ECG / EKG) analysis. It usually refers to one or more segments of electrocardiogram signal used to assess cardiac rhythm (i.e., heartbeat regularity). It focuses on the regularity of the RR interval (i.e., the time sequence of R wave appearance) and can be used to analyze heart rate changes and arrhythmias.

[0075] The solutions provided in this application will now be described in detail with reference to the accompanying drawings and embodiments.

[0076] This application proposes an adaptive R-wave detection method and apparatus based on multi-channel electrocardiogram (ECG) signals. By aligning, weighting, and fusing multi-channel signals, the robustness and accuracy of R-wave detection in pathological ECGs are improved. The following is a detailed description of this application.

[0077] like Figure 1 The image shows an adaptive R-wave detection method based on multi-channel electrocardiogram signals provided in an embodiment of this application. (Refer to...) Figure 1 As shown, the method includes:

[0078] S110: Acquire multi-channel electrocardiogram signals.

[0079] In this embodiment, the human body's multi-channel (e.g., 12-lead) electrocardiogram signals can be acquired in real time through the Bluetooth interface of an electrocardiogram acquisition device (e.g., a portable multi-lead electrocardiogram monitor), with each lead (e.g., I, II, III, V1~V6, aVR, aVL, aVF) as an independent channel.

[0080] S120: Perform channel-specific preprocessing on the multi-channel electrocardiogram signal to obtain the preprocessed multi-channel electrocardiogram signal.

[0081] Since the raw ECG signals acquired in step S110 typically contain baseline drift, electromyographic noise (high frequency), and power line interference, this step preprocesses each channel signal input in step S110. This preprocessing includes channel-specific denoising, channel-specific normalization, power line filtering, baseline correction, gain adjustment, and timestamp synchronization. Specifically, channel-specific denoising utilizes Gaussian wavelet transform to decompose the signal and weights and fuses wavelet coefficients at a specified scale to reconstruct the signal, effectively enhancing the QRS complex and suppressing various types of noise, thus improving the signal-to-noise ratio. Channel-specific normalization, based on the standard deviation or absolute median difference within a sliding window, unifies the amplitudes of different channels to the same range, eliminating amplitude differences. To enhance the comparability of signals from different channels and facilitate subsequent fusion analysis, the power frequency filtering employs an adaptive notch filter based on the least mean square algorithm, which can accurately and adaptively filter out power frequency interference and improve signal purity. Baseline correction effectively eliminates the influence of baseline drift on the signal by using morphological filtering or ignoring low-frequency approximation coefficients containing baseline drift in wavelet reconstruction, making the signal more realistically reflect cardiac electrical activity. Gain adjustment is adaptively adjusted according to the signal dynamic range, keeping the QRS complex amplitude within a suitable preset range, optimizing the QRS complex amplitude, improving the performance of the subsequent adaptive R-wave detection algorithm, reducing the probability of false positives and false negatives, and providing a high-quality, reliable, and easy-to-process signal foundation for accurate R-wave detection.

[0082] The following is a detailed explanation of the specific implementation methods for channel-specific denoising, channel-specific normalization, power frequency filtering, baseline correction, gain adjustment, and timestamp synchronization:

[0083] Channel-by-channel denoising: A threshold denoising algorithm based on wavelet transform is employed. A Gaussian wavelet basis is used to decompose the ECG signal of each channel. The detail coefficients are then reconstructed after soft thresholding, effectively smoothing noise while preserving the steep edges of the QRS complex. For example, for the original signal of each channel (such as lead II), a 10-level continuous wavelet transform (CWT) is performed using the Gaussian wavelet basis to obtain 10 different scales of detail coefficients (D1-D10) and approximation coefficients (A10). Taking a sampling frequency of 500Hz and a 10-layer decomposition as an example, the frequency distribution of each layer is roughly as follows: the frequency range of the first layer detail coefficient (D1) is approximately 125Hz~250Hz, the frequency range of the second layer detail coefficient (D2) is approximately 62.5Hz~125Hz, the frequency range of the third layer detail coefficient (D3) is approximately 31.25Hz~62.5Hz, the frequency range of the fourth layer detail coefficient (D4) is approximately 15.6Hz~31.25Hz, and the frequency range of the fifth layer detail coefficient (D5) is approximately 7.8Hz~15.6Hz. Among these, most of the energy of electromyographic interference is concentrated in layers D1 and D2, because these two layers cover their core frequency band. Some lower frequency electromyographic interference may extend to layer D3. Analysis showed that the QRS complex response was strongest in the main energy layers 3 to 7. Based on this, soft thresholding (setting small coefficients to zero and retaining large coefficients) can be applied to the detail coefficients D1-D2. The coefficients of layers D3-D7 are then weighted and fused according to their energy proportions (e.g., D3 weight 0.3, D4 weight 0.25, and D5-D7 weights 0.15 each). The fused coefficients are then reconstructed to obtain a denoised Lead II signal with enhanced QRS characteristics. In this fused signal, the QRS complex is greatly enhanced while low-frequency noise is relatively suppressed.

[0084] Channel-specific normalization: Based on the standard deviation or absolute median difference of the signal within a sliding window, the signal amplitude is normalized to a uniform range. For example, for each denoised channel (such as lead V5), a sliding window is taken (the window length is set to 2 seconds, i.e., 1000 data points), the standard deviation of the signal within the window is calculated, and all data points within the window are divided by the standard deviation (or absolute median difference) to achieve amplitude normalization [-1,1], giving it a uniform scale.

[0085] Power frequency filtering: An adaptive notch filter based on the least mean square algorithm is used to filter out power frequency interference. Specifically, an adaptive notch filter is used to generate a sine-cosine reference pair [sin(2π*50n), cos(2π*50n)] with the same frequency and phase as the power frequency interference (n is the sampling point number, 50Hz is the initial power frequency). Then, the least mean square (LMS) algorithm is used to adjust the weight of the reference signal in real time so that the filter output approximates the actual power frequency interference in the signal. The interference is then subtracted from the normalized channel signal to obtain a signal without power frequency noise. Compared with traditional fixed parameter filters, this embodiment significantly reduces the phase distortion introduced by filtering while ensuring the filtering effect, and better preserves the original morphological characteristics of the QRS group.

[0086] Baseline correction: The signal baseline is corrected by morphological filtering or by ignoring low-frequency approximation coefficients containing baseline drift during wavelet reconstruction. Specifically, in the preprocessing stage, preliminary baseline correction is performed using morphological filtering (FIR) and a mean filter. During wavelet reconstruction, the approximation coefficients A10 (containing residual baseline drift) from the 10-layer decomposition are discarded, and the signal is reconstructed using only layers D3-D5 (QRS main energy layers) to further eliminate baseline shift, resulting in a signal with a stable baseline and enhanced QRS characteristics.

[0087] Gain Adjustment: The gain is adaptively adjusted based on the dynamic range of the signal to ensure that the QRS group amplitude falls within a preset range. Specifically, the dynamic range (maximum value - minimum value) of the normalized signal for each channel is calculated. If the range is less than 0.8, it is multiplied by a gain factor (e.g., 1.2); if the range is greater than 1.2, it is multiplied by 0.9. This adjusts the normalized signal amplitude to the preset optimal range (e.g., 0.8-1.2), ensuring that the QRS group is easier to identify in subsequent detection.

[0088] Timestamp synchronization: Using the unified timestamp of the acquisition device as a reference, the time offset of each channel's data is calculated (e.g., V5 is delayed by 2ms compared to the reference). The missing data in the first 2ms of V5 is supplemented by linear interpolation, or the redundant data in the last 2ms of lead II is deleted, thereby eliminating the time deviation caused by the acquisition / transmission delay of each channel (e.g., lead II is delayed by 2ms compared to V5), and ensuring that all channel data are aligned on the time axis.

[0089] S130: Perform time alignment between channels on the preprocessed multi-channel electrocardiogram signal to obtain the aligned multi-channel electrocardiogram signal.

[0090] In this step, during time alignment, the channel with the highest signal-to-noise ratio is first selected from the preprocessed multi-channel ECG signal as the reference channel. Then, the cross-correlation function between the remaining channels and the reference channel is calculated. The peak value of the cross-correlation function is used to determine the delay of each channel relative to the reference channel. Based on this delay, phase shifting is performed on the remaining channels using linear interpolation to achieve sampling point-level time alignment for all channels. Example:

[0091] Reference channel selection: Select a data window of 4 heart rate cycles in a sliding evaluation window, calculate the signal-to-noise ratio of each channel signal. Calculation method: Based on the energy ratio of wavelets, calculate the ratio of the total energy of the coefficients of the wavelet scale (D3, D4) where the main energy of the QRS group is located to the total energy of the coefficients of the scale (D1, D10) where the noise is located, and select the channel with the highest signal-to-noise ratio as the reference benchmark for time alignment.

[0092] Cross-correlation analysis and phase alignment: Taking the R-wave position of the second heartbeat in the reference channel as the center, a data region of [-1000, 1000] in its neighborhood is selected. The cross-correlation function of the other channels (the signal range selection method is the same) with this signal is calculated. The relative delay is determined by locating the peak position of the cross-correlation function (the peak position of the cross-correlation function indicates the relative time difference when the two signals are most similar). Then, the phase shift of the non-reference channels is performed by linear interpolation to complete the sampling point-level time alignment of all channels.

[0093] S140: Based on signal quality assessment, the aligned multi-channel electrocardiogram signals are weighted and fused to generate a fused signal.

[0094] In this step, the weighted fusion scheme first filters candidate channels with waveforms consistent with the reference channel by using a cross-correlation coefficient threshold. This effectively eliminates channels with signal distortion caused by local interference or poor electrode contact, ensuring the reliability of the signal sources involved in the fusion. Then, weights are allocated based on the QRS complex energy of each candidate channel within a local window, ensuring that channels with more pronounced R-wave characteristics and higher signal-to-noise ratio dominate the fused signal. This is equivalent to constructing a signal synthesizer that adaptively focuses on the clearest cardiac beat characteristics. Finally, through data-level weighted fusion, the QRS complex energy of each high-quality channel is directionally superimposed, generating a fused signal with significantly improved signal-to-noise ratio and greatly enhanced R-wave characteristics. The specific process is as follows:

[0095] Determine candidate channels: Calculate the cross-correlation coefficient between each channel and the reference channel, and set a threshold (e.g., 0.85). Channels with cross-correlation coefficients higher than the preset threshold are determined as candidate channels.

[0096] Weighting: Within a local window (the first heartbeat), calculate the QRS energy (e.g., sum of squares) of each candidate channel, sort them by magnitude, and apply a pre-defined nonlinear weighting scheme based on the sorting results.

[0097] Fusion signal = 0.3*II + 0.25*V5 + 0.15*(V2 + V3 + V4 + V6) + 0.1*I + 0.2*III;

[0098] In some embodiments, this weight allocation enhancement scheme, based on weight allocation according to real-time QRS energy, further incorporates the pathological characteristics of ECG leads, pre-allocating higher base weights to leads known to present more typical R-wave morphology in specific cardiac diseases. This ensures, from the source, the ability of the fused signal to preserve and enhance the characteristics of pathological R waves, significantly improving the detection accuracy and pathological identification specificity of the algorithm when faced with abnormal ECGs with clear clinical characteristics.

[0099] Data weighted fusion: The signal sampling point values ​​of each candidate channel are multiplied by the corresponding fusion weight and then summed to generate a fusion signal with a significantly improved signal-to-noise ratio, which serves as the intermediate signal for subsequent R-wave detection.

[0100] S150: Perform adaptive R-wave detection on the fused signal to obtain a preliminary R-wave position sequence.

[0101] In this step, the adaptive R-wave detection process is as follows:

[0102] Signal enhancement: The fused signal is subjected to nonlinear transformation and differential operation to enhance the QRS group characteristics;

[0103] Constructing a local integration window: By using a local moving window integrator, the enhanced signal is integrated through a local moving window to concentrate the enhanced QRS group energy into a smooth pulse peak;

[0104] Dynamic threshold detection: A dual threshold strategy (signal threshold and noise threshold) is adopted on the integrated signal, and the threshold is dynamically updated within a sliding window to reliably identify R-wave candidate peaks;

[0105] R-wave bidirectional localization: The candidate R-wave positions detected on the integrated signal are mapped back to a local search window of the original fused signal, and the local maxima and minima of the signal are detected in parallel. By finding the maximum and minimum points of the signal within this window, the maximum extreme points are detected normally. When a negative R-wave is determined, the detection target is automatically switched to finding the local minimum, and finally, a preliminary R-wave position sequence is output.

[0106] S160: Based on the preliminary R-wave position sequence, perform multi-channel fusion positioning verification on the original multi-channel signal and output the final R-wave position sequence.

[0107] In this step, the multi-channel fusion positioning verification process is as follows:

[0108] Channel-specific verification: Map the initial R-wave position to each original lead (e.g., if the initial R-wave is at the 500th sampling point, the corresponding point is the 500th point in each lead). For conventional leads (e.g., II, III, V5, etc.): detect local extrema within the [-50, 50] data point window. If the deviation of the extrema from the initial position is <10ms, the verification is successful. For special leads (e.g., aVL, V1, V2, which are prone to double R-waves in special cases): detect the highest voltage peak within the [-100, 100] data point window as the R-wave position of that lead. If the deviation from the initial position is <20ms, the verification is successful.

[0109] Consistency check and conflict resolution: Statistically calculate the proportion of channels that pass verification. If the proportion of channels that pass verification exceeds a preset threshold, the R wave is confirmed. Otherwise, activate the conflict resolution mechanism, prioritize the verification results of high-weight channels, and remove false positive R waves or supplement missed R waves accordingly.

[0110] Final localization output: During the multi-channel fusion localization stage, different polarities of the R-wave are allowed across different channels. During separate channel verification, the R-wave is independently verified on each channel according to its own polarity (or an adaptively determined polarity). For example, a negative peak is verified on channel V1, and a positive peak on channel V6. The consistency check focuses on whether the timing of the R-wave's appearance is consistent across multiple channels, not on whether its polarity is consistent. An R-wave that is negative in lead V1 and positive in lead V6, as long as the timing is aligned, is considered a projection of the same heartbeat and is thus confirmed as a true R-wave.

[0111] The final R-wave position sequence output after the multi-channel fusion localization verification can be used as the result of the entire detection process and can be used for subsequent applications such as heart rate calculation and arrhythmia analysis.

[0112] In summary, the adaptive R-wave detection method based on multi-channel ECG signals provided in this application firstly processes the multi-channel ECG signals separately through channel-specific preprocessing. This removes noise and other interference factors tailored to the characteristics of each channel signal, improving the signal quality of each channel. Then, the preprocessed multi-channel ECG signals are time-aligned between channels to ensure temporal consistency and avoid deviations in subsequent processing due to time differences. Based on signal quality assessment, the aligned multi-channel ECG signals are weighted and fused to generate a fused signal. Adaptive R-wave detection is then performed on the fused signal to obtain a preliminary R-wave position sequence. This sequence is then used for multi-channel fusion positioning verification on the original multi-channel signals. By using multi-channel information to mutually verify each other, the probability of false detection and false negatives is effectively reduced, resulting in a more accurate final R-wave position sequence. This application addresses the problem of insufficient detection accuracy in existing technologies when dealing with diverse morphologies, especially pathological irregular ECGs, improving the robustness and accuracy of R-wave detection in pathological ECGs.

[0113] like Figure 2 As shown, this application also provides an adaptive R-wave detection device based on multi-channel electrocardiogram signals. This device can be used to implement any step of the above-described adaptive R-wave detection method based on multi-channel electrocardiogram signals and its optional embodiments, as described above. Figure 2 As shown, the device includes an acquisition module 210, a preprocessing module 220, a time alignment module 230, a weighted fusion module 240, a detection module 250, and a verification module 260.

[0114] The acquisition module 210 is used to acquire multi-channel electrocardiogram (ECG) signals; the preprocessing module 220 is used to perform channel-by-channel preprocessing on the multi-channel ECG signals to obtain preprocessed multi-channel ECG signals; the time alignment module 230 is used to perform inter-channel time alignment on the preprocessed multi-channel ECG signals to obtain aligned multi-channel ECG signals; the weighted fusion module 240 is used to perform weighted fusion on the aligned multi-channel ECG signals based on signal quality assessment to generate a fused signal; the detection module 250 is used to perform adaptive R-wave detection on the fused signal to obtain a preliminary R-wave position sequence; and the verification module 260 is used to perform multi-channel fusion positioning verification on the original multi-channel signals based on the preliminary R-wave position sequence to output the final R-wave position sequence.

[0115] It should be understood that the apparatus or module in the embodiments of this application can be implemented by software, for example, by a computer program or instruction having the above-described functions. The corresponding computer program or instruction can be stored in the internal memory of the terminal, and the above functions can be implemented by the processor reading the corresponding computer program or instruction in the memory. Alternatively, the apparatus or module in the embodiments of this application can also be implemented by hardware. Or, the apparatus or module in the embodiments of this application can also be implemented by a combination of a processor and a software module.

[0116] It should be understood that the processing details of the apparatus or module in the embodiments of this application can be found by referring to... Figure 1 The descriptions of the embodiments and related extended embodiments shown will not be repeated in this application.

[0117] Figure 3 This is a structural diagram of a computing device 1000 provided in an embodiment of this application. The computing device 1000 includes: a processor 1010, a memory 1020, a communication interface 1030, and a bus 1040.

[0118] It should be understood that Figure 3 The communication interface 1030 in the computing device 1000 shown can be used to communicate with other devices.

[0119] The processor 1010 can be connected to the memory 1020. The memory 1020 can be used to store the program code and data. Therefore, the memory 1020 can be a storage unit inside the processor 1010, an external storage unit independent of the processor 1010, or a component that includes both the storage unit inside the processor 1010 and the external storage unit independent of the processor 1010.

[0120] Optionally, the computing device 1000 may also include a bus 1040. The memory 1020 and communication interface 1030 can be connected to the processor 1010 via the bus 1040. The bus 1040 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus 1040 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 The symbol is represented by only one line, but this does not mean that there is only one bus or one type of bus.

[0121] It should be understood that in the embodiments of this application, the processor 1010 may be a central processing unit (CPU). The processor may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor. Alternatively, the processor 1010 may employ one or more integrated circuits to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0122] The memory 1020 may include read-only memory and random access memory, and provides instructions and data to the processor 1010. A portion of the processor 1010 may also include non-volatile random access memory. For example, the processor 1010 may also store device type information.

[0123] When the computing device 1000 is running, the processor 1010 executes the computer execution instructions in the memory 1020 to perform the operation steps of the above method.

[0124] It should be understood that the computing device 1000 according to the embodiments of this application can correspond to the corresponding subject executing the methods according to the various embodiments of this application, and the other operations and / or functions of each module in the computing device 1000 are respectively for implementing the corresponding processes of the methods of this embodiment. For the sake of brevity, they will not be described in detail here.

[0125] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0126] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0127] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0128] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0129] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0130] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0131] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, performs the above-described method, which includes at least one of the schemes described in the above embodiments.

[0132] The computer storage medium in this application embodiment can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0133] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0134] The program code contained on a computer-readable medium may be transmitted using any suitable medium, including, but not limited to, wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0135] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0136] It should be noted that the embodiments described in this application are merely some embodiments, not all embodiments. The components of the embodiments of this application typically described and shown in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the above detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0137] The terms "first, second, third, etc." or similar terms such as module A, module B, module C, etc., used in the specification and claims are only used to distinguish similar objects and do not represent a specific ordering of objects. It is understood that a specific order or sequence may be interchanged where permitted so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0138] In the above description, the labels indicating the steps do not necessarily mean that the steps will be executed. They may also include intermediate steps or be replaced by other steps. Where permissible, the order of the steps may be interchanged or executed simultaneously.

[0139] The term "comprising" as used in the specification and claims should not be construed as limiting itself to what follows; it does not exclude other elements or steps. Therefore, it should be interpreted as specifying the presence of the mentioned feature, integral, step, or component, but does not exclude the presence or addition of one or more other features, integrals, steps, or components, or groups thereof. Thus, the statement "device comprising means A and B" should not be limited to a device consisting solely of components A and B.

[0140] The terms "an embodiment" or "an embodiment" as used in this specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in at least one embodiment of this application. Therefore, the terms "in one embodiment" or "in an embodiment" appearing throughout this specification do not necessarily refer to the same embodiment, but may refer to the same embodiment. Furthermore, in the various embodiments of this application, unless otherwise specified or in case of logical conflict, the terminology and / or descriptions between different embodiments are consistent and can be mutually referenced. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0141] Note that the above are merely preferred embodiments and the technical principles employed in this application. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present application has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, all of which fall within the scope of protection of the present invention.

Claims

1. A method for adaptive R-wave detection based on multi-channel electrocardiogram signals, characterized in that, including: acquiring multi-channel electrocardiogram signals; performing channel-by-channel preprocessing on the multi-channel electrocardiogram signals to obtain preprocessed multi-channel electrocardiogram signals; performing time alignment between channels on the preprocessed multi-channel electrocardiogram signals to obtain aligned multi-channel electrocardiogram signals; performing weighted fusion on the aligned multi-channel electrocardiogram signals based on signal quality assessment to generate a single fused signal; performing adaptive R-wave detection on the fused signal to obtain a preliminary R-wave position sequence; performing multi-channel fusion positioning verification on the original multi-channel signals based on the preliminary R-wave position sequence and outputting a final R-wave position sequence.

2. The method according to claim 1, characterized in that, The channel-by-channel preprocessing includes independently performing at least one of the following on the signals of each channel: channel-by-channel denoising: decomposing the signal using Gaussian wavelet transform, performing weighted fusion on the wavelet coefficients within a specified scale range to reconstruct the signal, so as to enhance the QRS complex and suppress noise; channel-by-channel normalization: normalizing the signal amplitude to a unified interval based on the standard deviation or absolute median difference of the signal within a sliding window; power frequency filtering: using an adaptive notch filter based on the least mean square algorithm to filter out power frequency interference; baseline correction: correcting the signal baseline by morphological filtering or ignoring the low-frequency approximation coefficients containing baseline drift in wavelet reconstruction; gain adjustment: adaptively adjusting the gain according to the dynamic range of the signal to make the amplitude of the QRS complex fall within a preset interval.

3. The method according to claim 2, characterized in that, The channel-by-channel denoising uses Gaussian wavelets to decompose the signal into N layers, selects the wavelet coefficients from the Kth layer to the Mth layer for weighted fusion to reconstruct the signal, where N, K, and M are natural numbers, and 3 ≤ K < M ≤ N, and the Kth layer to the Mth layer cover the main energy frequency band of the QRS complex.

4. The method according to claim 1, characterized in that, The time alignment includes: selecting a channel with the highest signal-to-noise ratio from the preprocessed multi-channel electrocardiogram signals as a reference channel; calculating the cross-correlation function of each of the remaining channels with the reference channel, and determining the delay amount of each channel relative to the reference channel by locating the peak of the cross-correlation function; performing phase shift on each of the remaining channels by linear interpolation according to the delay amount to achieve time alignment at the sampling point level for all channels.

5. The method according to claim 4, characterized in that, The calculation method of the signal-to-noise ratio is: calculating the ratio of the total energy of the coefficients of the signal on the wavelet scale representing the energy of the QRS complex to the total energy of the coefficients on the wavelet scale representing the energy of the noise.

6. The method according to claim 1, characterized in that, The weighted fusion includes: determining candidate channels: calculating the cross-correlation coefficient of each channel with a reference channel, and determining the channels with the cross-correlation coefficient higher than a preset threshold as candidate channels; weight assignment: based on the QRS complex energy size of each candidate channel within a local window, assigning a fusion weight to each candidate channel, where the channel with higher QRS complex energy is assigned a higher weight; data weighted fusion: multiplying the signal sampling point values of each candidate channel by the corresponding fusion weight and then summing to generate the fused signal.

7. The method according to claim 6, characterized in that, The weight assignment further includes: combining the pathological characteristics of the electrocardiogram leads, and pre-assigning a higher basic weight to the leads with more significant R-wave morphological characteristics under specific pathological conditions.

8. The method according to claim 1, characterized in that, The adaptive R-wave detection includes: The fused signal is subjected to nonlinear transformation and differential operation to enhance the QRS group characteristics; The enhanced signal is integrated by a local moving window to concentrate the enhanced QRS group energy into a smooth pulse peak; A dynamically updated dual-threshold strategy is used on the integrated signal to identify R-wave candidate peaks; The positions of the R-wave candidate peaks are mapped back to the local search window of the original fused signal. The precise R-wave position is determined by detecting the local extrema within the window. The detection supports the simultaneous detection of positive and negative R-waves.

9. The method according to claim 1, characterized in that, The multi-channel fusion positioning verification includes: Channel-by-channel verification: The preliminary R-wave position sequence is mapped to each original preprocessed channel signal, and local extremum detection is performed independently on each channel to verify the R-wave position; Consistency check and conflict resolution: Calculate the proportion of channels that pass the verification. If the proportion exceeds the preset threshold, the R wave is confirmed; otherwise, the conflict resolution mechanism is activated, and the verification results of the high-weight channels are adopted first. Final positioning output: Allows different polarities of the R-wave in different channels, and confirms the true R-wave only based on the consistency of the time point of the R-wave appearance.

10. An adaptive R-wave detection device based on multi-channel electrocardiogram signals, characterized in that, include: The acquisition module is used to acquire multi-channel electrocardiogram signals; The preprocessing module is used to perform channel-by-channel preprocessing on the multi-channel electrocardiogram signal to obtain the preprocessed multi-channel electrocardiogram signal. The time alignment module is used to perform time alignment between channels of the preprocessed multi-channel electrocardiogram signal to obtain an aligned multi-channel electrocardiogram signal. The weighted fusion module is used to perform weighted fusion on the aligned multi-channel electrocardiogram signals based on signal quality assessment to generate a fused signal. The detection module is used to perform adaptive R-wave detection on the fused signal to obtain a preliminary R-wave position sequence; The verification module is used to perform multi-channel fusion positioning verification on the original multi-channel signal based on the preliminary R-wave position sequence, and output the final R-wave position sequence.

Images