A multi-channel joint time-space domain filtering method for magnetocardiography signals

Abstract

A kind of multi-channel joint space-time filtering method of magnetocardiogram signal belongs to magnetocardiogram signal processing field, first, multi-channel magnetocardiogram signal is collected, all channels are fitted to obtain the Hermite parameter of each channel at each time using Hermite fitting method;Magnetocardiogram image is divided into several blocks from space, and the magnetocardiogram image at each time is expressed using the Hermite parameter of corresponding channel;All magnetocardiogram image blocks at all times are clustered using clustering method to construct different characteristic waveform magnetocardiogram image block set;Within the range of this set, the magnetocardiogram signal is weighted and averaged to obtain the denoised magnetocardiogram signal.The present application utilizes the characteristics of high synchronism and high repeatability of multi-channel magnetocardiogram signal, denoises magnetocardiogram image in time domain and space domain, improves the non-Gaussian noise suppression effect of spectral overlap and the quality of two-dimensional magnetocardiogram image data, while retaining the time domain characteristics of magnetocardiogram signal, improves the signal-to-noise ratio of magnetocardiogram signal.

Description

Technical Field

[0001] This invention belongs to the field of magnetocardiographic signal processing technology, specifically relating to a multi-channel joint spatiotemporal filtering method for magnetocardiographic signals. Background Technology

[0002] Cardiovascular disease poses a persistent threat to human life and health, and more than four-fifths of premature heart disease and stroke risks can be avoided through appropriate means. Therefore, improving the convenience and accuracy of cardiovascular disease monitoring and diagnosis is of great significance. Magnetocardiography (MCG), as a non-invasive measurement of cardiac electrophysiological activity, can be used to study the state of the heart. Compared with traditional electrocardiogram (ECG) devices, extremely weak magnetocardiography (MCG) devices have advantages such as non-contact measurement, high sensitivity, and strong specificity. Preoperative MCG measurement can provide valuable information for medical intervention.

[0003] Magnetocardiogram (MCC) signals are extremely weak and easily affected by environmental noise, instrument noise, and noise from human eye movements and breathing during actual detection. In particular, non-Gaussian noise overlapping with the signal spectrum severely impacts the signal-to-noise ratio. As a spatial signal, the MCC signal inherently contains rich spatial information. Current MCC filtering methods primarily employ single-channel processing, neglecting the correlation between different channels, thus failing to effectively guarantee the quality of the MCC image and exhibiting poor suppression of non-Gaussian noise with overlapping spectrum. Therefore, how to utilize the temporal and spatial information of the MCC to improve its quality is of great significance in this field. Summary of the Invention

[0004] To address the technical problems of poor image quality and ineffective suppression of non-Gaussian noise with overlapping spectra in existing magnetocardiogram (MCC) filtering methods that primarily rely on single-channel processing, this invention provides a multi-channel joint spatiotemporal domain filtering method for MCC signals.

[0005] The technical solution adopted by this invention to solve the technical problem is as follows:

[0006] The present invention provides a multi-channel joint spatiotemporal filtering method for magnetocardiogram signals, comprising the following steps:

[0007] Step 1: Acquire multi-channel magnetic field signals using a magnetocardiograph;

[0008] Step 2: Based on the multi-channel magnetocardiogram signal, the Hermite fitting method is used to fit each channel to obtain the Hermite fitting parameters of each channel at each time step.

[0009] Step 3: Divide the magnetocardiogram into several n*n blocks in space, and use the Hermite fitting parameters of the corresponding channels to represent the magnetocardiogram of each block at each time point.

[0010] Step 4: Based on the Hermite fitting parameters, clustering methods are used to cluster all magnetocardiogram (MCC) image blocks at all times to construct a set of MCC image blocks with different characteristic waveforms.

[0011] Step 5: Within the set of magnetocardiogram (MCC) image blocks with different characteristic waveforms, calculate the weight of the MCC signal for each MCC image block based on the Hermite fitting parameters, and then perform a weighted average of the MCC signals for each MCC image block based on the calculated weight information to obtain the denoised MCC signal.

[0012] Furthermore, Hermite fitting parameters, which characterize waveform features, are used to represent each sampling point.

[0013] Furthermore, the Hermite fitting parameters are fitted using the least squares method, and the specific fitting formula is as follows:

[0014]

[0015] Where σ represents the time scale factor of the Hermite fit, c n Let x(t) represent the nth-order Hermite coefficients obtained by Hermite fitting of the data segment, x(t) represent the data segment to be fitted, t∈T represent the t-th sampling point of the data segment, and Φ n (t,σ) denotes the nth-order Hermite basis functions, H n (t / σ) represents an nth-order Hermite polynomial with t / σ as the independent variable; the Hermite polynomial is obtained through a recursive formula, in which H1(x) represents a first-order Hermite polynomial, H... n (x) represents the Nth order Hermite polynomial, H n-1 (x) represents the (n-1)th Hermite polynomial, H n-2 (x) represents the (n-2)th Hermite polynomial.

[0016] Furthermore, in step three, the number of obtained magnetocardiogram image blocks is Ch / n*n*T, where Ch represents the number of channels and T represents the number of magnetocardiogram signal samples.

[0017] Furthermore, in step four, a clustering method is used to cluster all magnetocardiogram (MCC) image blocks at all times based on the low-order Hermite coefficients of the n*n blocks at each time step. By iteratively finding K cluster centers, all MCC image block data are divided into K sets.

[0018] Furthermore, in step four, the clustering method employs the K-means algorithm.

[0019] Furthermore, in step five, the weights of the magnetic field signals of each magnetic field image block are derived from the similarity coefficients between the magnetic field image blocks obtained by Hermite fitting.

[0020] Furthermore, in step five, the specific calculation formula for the weight of the magnetic field signal in each magnetic field image block is as follows:

[0021]

[0022] Where w(i, j) represents the weight of the j-th sampling point to the i-th sampling point, i.e., the similarity coefficient, and Z represents the normalization coefficient, ensuring that... c i,t c represents the Hermite coefficient of order t at the i-th sampling point. j,t h represents the Hermite coefficient of order t at the j-th sampling point. k The sum of variances of the Hermite coefficients of all sampling points in the k-th set obtained by clustering is the filter coefficient, M represents the total Hermite order, and n represents the side length of the magnetocardiogram image block.

[0023] Furthermore, the filter coefficient h k The specific calculation formula is as follows:

[0024]

[0025] Where N represents the number of elements in the set, M represents the total Hermite order, n represents the side length of the magnetocardiogram image block, and c n,m This represents the m-th order Hermite coefficient at the nth sampling point. This represents the mean of the m-th Hermite coefficients of all N elements in a cluster set.

[0026] The beneficial effects of this invention are:

[0027] This invention discloses a multi-channel joint spatiotemporal filtering method for magnetocardiogram (MCG) signals. First, multi-channel MCG signals are acquired using a magnetocardiograph. The Hermite fitting method is then applied to all channels to obtain the Hermite fitting parameters for each channel at each time step. Next, the MCG images are spatially divided into several blocks, and the Hermite fitting parameters for each corresponding channel are used to represent the MCG image at each time step. Then, all blocks at all time steps are clustered to construct a set of MCG image blocks with different characteristic waveforms. Finally, the MCG signals are weighted and averaged within the set of MCG image blocks with different characteristic waveforms to obtain the denoised MCG signal. The weights are derived from the similarity between the MCG image blocks calculated through Hermite fitting.

[0028] This invention discloses a multi-channel joint spatiotemporal domain filtering method for magnetocardiogram (MCC) signals. Utilizing the high synchronization and repeatability of multi-channel MCC signals, it performs noise reduction processing on MCC images in both the time and spatial domains. This method adaptively handles various types of non-Gaussian noise, improving the suppression effect on non-Gaussian noise with overlapping spectra. The combination of temporal and spatial denoising helps improve the quality of two-dimensional MCC image data while preserving the temporal characteristics of the MCC signal, thus increasing the signal-to-noise ratio (SNR) of the MCC signal. Furthermore, it protects the spatial characteristics of the MCC image while improving the SNR of a single channel signal. Attached Figure Description

[0029] Figure 1 This is a flowchart illustrating a multi-channel joint spatiotemporal filtering method for magnetocardiogram signals provided by the present invention. Detailed Implementation

[0030] This invention provides a multi-channel joint spatiotemporal domain filtering method for magnetocardiogram (MCC) signals, which reduces noise in the time and spatial domains of the MCC signal and improves the quality of the MCC signal by utilizing the temporal and spatial information of the MCC signal.

[0031] To achieve the above objectives, this invention provides a multi-channel joint spatiotemporal filtering method for magnetocardiogram signals. For example... Figure 1 The present invention specifically includes the following steps:

[0032] Step 1: Collect multi-channel magnetocardiogram (MCG) signals from the test subjects using a magnetocardiogram (MCG) instrument, check the data quality, remove invalid or abnormal channels, and record the number of channels as Ch and the number of MCG signal samples as T, finally obtaining the noisy multi-channel MCG signal.

[0033] The preferred channel number Ch is 64, the preferred sampling rate is 1000Hz, and the preferred acquisition time is 60 seconds. During the acquisition process, the test subject must not carry any metal objects to avoid interference with the magnetocardiogram (MCG) equipment.

[0034] Step 2: Based on the multi-channel magnetocardiogram signal, the Hermite fitting method is used to fit each channel to obtain the Hermite fitting parameters of each channel at each time point. The Hermite fitting parameters, which characterize the waveform features, are used to represent each sampling point.

[0035] By fitting all channels using the Hermite fitting method, the Hermite fitting parameters for each channel at each time step can be obtained. The preferred data length is a 100ms data segment, and the preferred Hermite polynomial order is 6. The time-domain signal of each channel of the noisy magnetocardiogram signal at each time step is represented by low-order Hermite coefficients obtained based on the Hermite fitting, i.e., each 100ms data segment is compressed into 6th-order Hermite coefficients.

[0036] The Hermite fitting parameters can be specifically fitted using the least squares method, and the specific solution steps are as follows:

[0037] Extract a 100ms data segment near sampling point s, and fit the M-order Hermite coefficients within the vicinity of the single-channel time-domain signal using the following formula, compressing the sampling points within the 100ms segment into a vector of length M. The specific fitting formula is as follows:

[0038]

[0039] Where σ represents the time scale factor of the Hermite fit, c n Let x(t) represent the nth-order Hermite coefficients obtained by Hermite fitting of the data segment, x(t) represent the data segment to be fitted, t∈T represent the t-th sampling point of the data segment, and Φ n (t,σ) denotes the nth-order Hermite basis functions, H n (t / σ) represents an nth-order Hermite polynomial with t / σ as the independent variable; the Hermite polynomial can be obtained through a recursive formula, in which H1(x) represents a first-order Hermite polynomial, H... n (x) represents the Nth order Hermite polynomial, H n-1 (x) represents the (n-1)th Hermite polynomial, H n-2 (x) represents the (n-2)th Hermite polynomial;

[0040] Step 3: Divide the magnetocardiogram (MCC) image into several n*n blocks in space, and use the Hermite fitting parameters of the corresponding channels to represent the MCC image of each block at each time point. The number of MCC image blocks is Ch / n*n*T, where Ch represents the number of channels and T represents the number of MCC signal samples.

[0041] Step 4: Based on the Hermite fitting parameters, clustering methods are used to cluster all magnetocardiogram (MCC) image blocks at all times to construct a set of MCC image blocks with different characteristic waveforms.

[0042] Specifically, a clustering method is used to cluster all magnetocardiogram (MCC) image blocks at all times based on the low-order Hermite coefficients of the n*n blocks at each time step. Through iteration, K cluster centers are found, and all MCC image block data are divided into K sets.

[0043] Specifically, with n=2 and K=50, each frame of magnetocardiogram (MCC) image is spatially divided into several 2x2 blocks. The four 6th-order Hermite coefficients are merged, and each 2x2 block is represented as a 4x6 Hermite coefficient matrix. Based on the low-order Hermite coefficients of the 2x2 blocks at each time step, the K-means algorithm is used to cluster all MCC image blocks across all time steps. Through iteration, 50 cluster centers are found, dividing all MCC image block data into 50 sets.

[0044] It should be noted that the K-means algorithm is a well-known prior art and will not be described in detail here. For the clustering of magnetic heart image blocks, in addition to the maximum likelihood estimation method, any existing method can be used. This invention does not specifically limit the estimation method.

[0045] Step 5: Within the set of magnetocardiogram (MCC) image blocks with different characteristic waveforms, calculate the weight of the MCC signal for each MCC image block based on the Hermite fitting parameters, and then perform a weighted average of the MCC signals for each MCC image block based on the calculated weight information to obtain the denoised MCC signal. The weights are derived from the similarity coefficients between the MCC image blocks calculated through Hermite fitting.

[0046] The specific formula for calculating the similarity coefficient between different blocks of the magnetocardiogram image is as follows:

[0047]

[0048] Where ω(i,j) represents the weight of the j-th sampling point to the i-th sampling point, i.e., the similarity coefficient, and Z represents the normalization coefficient, ensuring that... c i,t c represents the Hermite coefficient of order t at the i-th sampling point. j,t h represents the Hermite coefficient of order t at the j-th sampling point. k The sum of variances of the Hermite coefficients of all sampling points in the k-th set obtained by clustering is the filter coefficient, M represents the total Hermite order, and n represents the side length of the magnetocardiogram image block.

[0049] The following formula can be used as the filter coefficient h. k Global adaptation:

[0050]

[0051] Where N represents the number of elements in the set, M represents the total Hermite order, n represents the side length of the magnetocardiogram image block, and c n,m This represents the m-th order Hermite coefficient at the nth sampling point. This represents the mean of the m-th Hermite coefficients of all N elements in a cluster set.

[0052] In summary, this invention provides a multi-channel joint spatiotemporal filtering method for magnetocardiogram (MCG) signals. It acquires multi-channel MCG signals using a magnetocardiograph, employs Hermite fitting to fit all channels, and obtains Hermite fitting parameters for each channel at each time step. The MCG image is spatially divided into several blocks, and the Hermite fitting parameters for each corresponding channel represent the MCG image at each time step. All blocks at all time steps are clustered to construct MCG image block sets with different characteristic waveforms. The MCG signal is then weighted and averaged within these sets, with the weights derived from the similarity between the MCG image blocks calculated through Hermite fitting. This invention utilizes the high synchronization and repeatability of multi-channel MCG signals to reduce noise in both the temporal and spatial domains, helping to improve the signal-to-noise ratio of single-channel signals while preserving the spatial characteristics of the MCG image.

[0053] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art to depart from the spirit and technical concept disclosed herein should still be covered by the protection scope of the present invention.

Claims

1. A multi-channel joint spatiotemporal domain filtering method for magnetocardiogram signals, characterized in that, Includes the following steps: Step 1: Acquire multi-channel magnetic field signals using a magnetocardiograph; Step 2: Based on the multi-channel magnetocardiogram signal, the Hermite fitting method is used to fit each channel to obtain the Hermite fitting parameters of each channel at each time step. Step 3: Divide the magnetocardiogram into several blocks of size a*a in space, and use the Hermite fitting parameters of the corresponding channel to represent the magnetocardiogram of each block at each time point. Step 4: Based on the Hermite fitting parameters, clustering methods are used to cluster all magnetocardiogram (MCC) image blocks at all times to construct a set of MCC image blocks with different characteristic waveforms. Step 5: Within the set of magnetocardiogram (MCC) image blocks with different characteristic waveforms, calculate the weight of the MCC signal for each MCC image block based on the Hermite fitting parameters, and then perform a weighted average of the MCC signals for each MCC image block based on the calculated weight information to obtain the denoised MCC signal. The specific formula for calculating the weight of the magnetic field signal in each magnetic field image block is as follows: ， ， in Z represents the weight of the j-th sampling point to the i-th sampling point, i.e., the similarity coefficient, and Z represents the normalization coefficient, ensuring that... c i,t c represents the Hermite coefficient of order t at the i-th sampling point. j,t h represents the Hermite coefficient of order t at the j-th sampling point. k The sum of variances of the Hermite coefficients of all sampling points in the k-th set obtained by clustering is the filter coefficient, M represents the total Hermite order, and β represents the side length of the magnetocardiogram image block.

2. The multi-channel joint spatiotemporal filtering method for magnetocardiogram signals according to claim 1, characterized in that, Each sampling point is represented by Hermite fitting parameters that characterize waveform features.

3. The multi-channel joint spatiotemporal filtering method for magnetocardiogram signals according to claim 1, characterized in that, The Hermite fitting parameters were fitted using the least squares method, and the specific fitting formula is as follows: in, c represents the time scale factor of the Hermite fit. n The denot represents the nth-order Hermite coefficients obtained by Hermite fitting of the data segment, and x(t) represents the data segment to be fitted. This represents the t-th sampling point in the data segment. H represents the nth-order Hermite basis function. n ( ) indicates with Let H1(x) be the nth-order Hermite polynomial of the independent variable; the Hermite polynomial is obtained through a recursive formula, in which H1(x) represents the first-order Hermite polynomial, H... n (x) represents the nth Hermite polynomial, H n-1 (x) represents the (n-1)th Hermite polynomial, H n-2 (x) represents the (n-2)th Hermite polynomial.

4. The multi-channel joint spatiotemporal filtering method for magnetocardiogram signals according to claim 1, characterized in that, In step three, the number of obtained magnetocardiogram image blocks is Ch / a*a*T, where Ch represents the number of channels and T represents the number of magnetocardiogram signal samples.

5. The multi-channel joint spatiotemporal filtering method for magnetocardiogram signals according to claim 1, characterized in that, In step four, a clustering method is used to cluster all magnetocardiogram (MCC) image blocks at all times based on the low-order Hermite coefficients of the a*a block at each time step. By iteratively finding K cluster centers, all MCC image block data are divided into K sets.

6. The multi-channel joint spatiotemporal domain filtering method for magnetocardiogram signals according to claim 1, characterized in that, In step four, the clustering method used is the K-means algorithm.

7. The multi-channel joint spatiotemporal filtering method for magnetocardiogram signals according to claim 1, characterized in that, In step five, the weights of the magnetic field signals of each magnetic field image block are derived from the similarity coefficients between the magnetic field image blocks obtained by Hermite fitting.

8. The multi-channel joint spatiotemporal filtering method for magnetocardiogram signals according to claim 1, characterized in that, The filter coefficient h k The specific calculation formula is as follows: ， Where N represents the number of elements in the set, M represents the total Hermite order, β represents the side length of the magnetocardiogram image block, and c η,m This represents the m-th order Hermite coefficient at the η-th sampling point. This represents the mean of the m-th Hermite coefficients of all N elements in a cluster set.

Images