Method and device for generating magnetoencephalogram evoked signal

Abstract

The invention belongs to the technical field of biological signal processing, and discloses a magnetoencephalogram induction signal generation method and device, and the magnetoencephalogram induction signal generation method comprises the steps: determining the alignment cost of all signals in magnetoencephalogram signals according to a minimum cost path between any two test signals in the magnetoencephalogram signals, the minimum cost path is used for representing the alignment cost of the minimum time between the signals of the two corresponding trials; determining a time similarity weight coefficient of the signal of each trial according to the alignment cost; and according to the signal of each trial and the corresponding time similarity weight coefficient, generating a magnetoencephalogram evoked signal. The method well solves the problems that in the prior art, interested signals in magnetoencephalogram signals cannot be effectively extracted, and the influence of signal jitter cannot be effectively reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of biological signal processing and magnetoencephalogram signal preprocessing, and specifically relates to a method and device for generating magnetoencephalogram induced signals. Background Art

[0002] In the prior art, arithmetic averaging is a classic method for removing noise from magnetoencephalogram signals. Although it has been widely used in auditory, visual and somatosensory experiments, it also has obvious limitations. When there is a large amount of noise or outliers (instantaneous high-intensity noise) in the experiment, the arithmetic averaging method can only rely on increasing the number of experimental trials to eliminate these noises as much as possible. Not only is it time-consuming, but as the experimental time increases, the fatigue of the subjects and the habit of repeated stimulation will cause the evoked response waveform to change. In addition, differences between different subjects and equipment may also lead to blurring of the average waveform, as well as changes in peak time and amplitude. These changes are called signal jitter. Therefore, how to reduce the impact of high-intensity noise and jitter is the focus of improving the averaging method. Summary of the invention

[0003] An object of the present invention is to provide a method for generating a magnetoencephalogram induced signal, which aims to solve the problems in the prior art that the signal of interest in the magnetoencephalogram signal cannot be effectively extracted and the influence of signal jitter cannot be effectively reduced.

[0004] Another object of the present invention is to provide a device for generating magnetoencephalogram induced signals. Another object of the present invention is to provide an electronic device, the electronic device comprising a memory and a processor, the memory storing a computer program, the processor implementing the steps of the method for generating magnetoencephalogram induced signals when executing the computer program. Another object of the present invention is to provide a readable medium, on which a computer program is stored, the computer program implementing the steps of the method for generating magnetoencephalogram induced signals when executed by the processor.

[0005] In order to solve the technical problems in the background technology of the present invention, the present invention provides the following technical solutions:

[0006] In a first aspect, the present invention provides a method for generating a magnetoencephalogram induced signal, comprising:

[0007] Determining the alignment cost of all signals in the magnetoencephalogram signal according to the minimum cost path between any two trial signals in the magnetoencephalogram signal, wherein the minimum cost path is used to represent the alignment cost of minimizing the time between the corresponding two trial signals;

[0008] Determining a temporal similarity weight coefficient of the signal of each trial according to the alignment cost;

[0009] Generate a magnetoencephalogram (MEG) evoked signal based on the signal of each trial and the corresponding temporal similarity weight coefficient.

[0010] In some embodiments of the present invention, a method for generating a MEG evoked signal further includes:

[0011] Generate a cost matrix for the signals of any two trials;

[0012] Determine the Euclidean distance between the signals of any two trials according to the cost matrix;

[0013] Determine the minimum cost path according to the Euclidean distance between the signals of any two trials.

[0014] In some embodiments of the present invention, determining the minimum cost path according to the Euclidean distance between the signals of any two trials includes:

[0015] Determine a cumulative distance matrix between the signals of any two trials according to the Euclidean distance;

[0016] Determine the minimum cost path according to the cumulative distance matrix.

[0017] In some embodiments of the present invention, the alignment cost is the average of the distance differences between the matching points on the minimum cost path, and the overall temporal difference degree between the signals of all trials in the MEG signal and the MEG evoked signal is less than a preset value.

[0018] In a second aspect, the present invention provides a device for generating a MEG evoked signal, the device includes:

[0019] An alignment cost determination module, configured to determine the alignment cost of all signals in the MEG signal according to the minimum cost path between the signals of any two trials in the MEG signal, where the minimum cost path is used to characterize the minimized temporal alignment cost between the signals of the corresponding two trials;

[0020] A temporal weight coefficient determination module, configured to determine the temporal similarity weight coefficient of the signal of each trial according to the alignment cost;

[0021] An evoked signal generation module, configured to generate a MEG evoked signal according to the signal of each trial and the corresponding temporal similarity weight coefficient.

[0022] In some embodiments of the present invention, a device for generating a MEG evoked signal further includes:

[0023] A cost matrix generation module, configured to generate a cost matrix for the signals of any two trials;

[0024] An Euclidean distance determination module, configured to determine the Euclidean distance between the signals of any two trials according to the cost matrix;

[0025] A minimum cost path determination module, configured to determine the minimum cost path according to the Euclidean distance between the signals of any two trials.

[0026] In some embodiments of the present invention, the minimum cost path determination module includes:

[0027] An accumulated distance matrix determination unit, configured to determine an accumulated distance matrix between the signals of any two trials according to the Euclidean distance;

[0028] A minimum cost path determination unit, configured to determine the minimum cost path according to the accumulated distance matrix.

[0029] In some embodiments of the present invention, the alignment cost is the average value of the distance differences between the matching points on the minimum cost path, and the overall time difference degree between the signals of all trials in the magnetoencephalogram signal and the magnetoencephalogram evoked signal is less than a preset value.

[0030] In a third aspect, the present invention provides a computer program product, including a computer program / instructions, and when the computer program / instructions are executed by a processor, the steps of a method for generating a magnetoencephalogram evoked signal are implemented.

[0031] In a fourth aspect, the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the program, the steps of a method for generating a magnetoencephalogram evoked signal are implemented.

[0032] In a fifth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of a method for generating a magnetoencephalogram evoked signal are implemented.

[0033] As can be seen from the above description, embodiments of the present invention provide a method and device for generating a magnetoencephalogram evoked signal. The corresponding method includes: First, determine the alignment cost of all signals in the magnetoencephalogram signal according to the minimum cost path between the signals of any two trials in the magnetoencephalogram signal, where the minimum cost path is used to represent the minimum alignment cost of time between the signals of the corresponding two trials; Then, determine the time similarity weight coefficient of the signal of each trial according to the alignment cost; Finally, generate a magnetoencephalogram evoked signal according to the signal of each trial and the corresponding time similarity weight coefficient.

[0034] The present invention proposes an improved robust weighted average method to overcome the limitations of existing average methods. By calculating the alignment cost between the trial signal and the average signal and using it as the time similarity weight coefficient, and then minimizing the criterion function through iteration to obtain the optimal weight for each trial, the negative impacts brought by outliers and jitters are reduced. In addition, the present invention proves through simulation and real data experiments that the method provided by the present invention is superior to several existing benchmark average techniques in terms of signal-to-noise ratio, root mean square error, and waveform recovery. In addition, this method can also be easily extended to other biological signal measurements. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other accompanying drawings can be obtained based on these drawings without creative efforts.

[0036] Figure 1 Flow schematic of a method for generating magnetoencephalogram evoked signals in an embodiment of the present invention Figure 1 ;

[0037] Figure 2 Flow schematic of a method for generating magnetoencephalogram evoked signals in an embodiment of the present invention Figure 2 ;

[0038] Figure 3 Flow schematic of step 600 in an embodiment of the present invention;

[0039] Figure 4 Flow schematic of a method for generating magnetoencephalogram evoked signals in a specific embodiment of the present invention;

[0040] Figure 5 Schematic diagram of the SNR evaluation index results of each method in the simulation experiment in a specific embodiment of the present invention;

[0041] Figure 6 Schematic diagram of the RMSE evaluation index results of each method in the simulation experiment in a specific embodiment of the present invention;

[0042] Figure 7 Schematic diagram of the waveform results of different methods in the simulation experiment in a specific embodiment of the present invention;

[0043] Figure 8 Schematic diagram of the waveform results of different methods in the real auditory experiment in a specific embodiment of the present invention;

[0044] Figure 9Block diagram of a device for generating magnetoencephalogram evoked signals in an embodiment of the present invention Figure 1 ;

[0045] Figure 10 Block diagram of a device for generating magnetoencephalogram evoked signals in an embodiment of the present invention Figure 1 ;

[0046] Figure 11 Block diagram of the minimum cost path determination module 60 in an embodiment of the present invention;

[0047] Figure 12 Schematic structural diagram of an electronic device in an embodiment of the present invention. Detailed implementation manners

[0048] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0049] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0050] It should be noted that the terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned accompanying drawings are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices. Without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0051] It can be understood that the evoked response in magnetoencephalography (MEG) is of great significance for neuroscience analysis and clinical research, and it is usually obtained from the average value of many trials. However, the traditional arithmetic average method has inherent disadvantages, such as being sensitive to strong abnormal noise, which can only be reduced by increasing the number of experimental trials. In addition, signal jitter caused by factors such as the subject or the device system will also affect the amplitude and time of the average signal peak. Based on this and in order to solve at least some of the technical problems in the prior art, an embodiment of the present invention provides a specific implementation manner of a method for generating magnetoencephalogram evoked signals. See Figure 1 , and the method specifically includes the following contents:

[0052] Step 100: Determine the alignment cost of all signals in the magnetoencephalogram signal according to the minimum cost path between the signals of any two trials in the magnetoencephalogram signal, where the minimum cost path is used to represent the minimized time alignment cost between the signals of the corresponding two trials;

[0053] Step 200: Determine the time similarity weight coefficient of the signal of each trial according to the alignment cost;

[0054] Step 300: Generate a magnetoencephalogram evoked signal according to the signal of each trial and the corresponding time similarity weight coefficient.

[0055] As can be seen from the above description, an embodiment of the present invention provides a method for generating magnetoencephalogram evoked signals, including: First, determine the alignment cost of all signals in the magnetoencephalogram signal according to the minimum cost path between the signals of any two trials in the magnetoencephalogram signal, where the minimum cost path is used to represent the minimized time alignment cost between the signals of the corresponding two trials; Then, determine the time similarity weight coefficient of the signal of each trial according to the alignment cost; Finally, generate a magnetoencephalogram evoked signal according to the signal of each trial and the corresponding time similarity weight coefficient.

[0056] The present invention calculates the alignment cost between each trial and the average signal, converts it into a time similarity coefficient to update the weight, and then minimizes the criterion function through iteration to obtain the optimal weight of each trial. The robust weighted average method provided by the present invention is tested with simulation and real experimental data, and has better performance compared with other average methods.

[0057] It can be understood that the trial in step 100 can be understood from the following aspects: A trial refers to each complete stimulus-response process in a magnetoencephalogram experiment, which is a time slice of a single complete experimental unit under a specific experimental paradigm, including the pre-stimulus baseline period and the post-stimulus response period.

[0058] Technical parameter description: Time window: e.g., "Each trial includes the data segment from 500 ms before the stimulus to 1000 ms after the stimulus"; Number of sampling points: e.g., "Each trial includes 1500 sampling points". Baseline time: e.g., "Using 200 ms before the stimulus as the baseline period".

[0059] Experimental design description: Number of trials: e.g., "Each subject completes 300 valid trials"; Classification situation: e.g., "Including 150 target stimulus trials and 150 non-target stimulus trials". Inter-trial interval: e.g., "The inter-trial interval randomly varies around 100 ms".

[0060] Data processing description: Trial extraction: e.g., "Trial segmentation is performed based on the trigger signal"; Trial screening: e.g., "The valid trials retained after removing the trials contaminated by artifacts"; Superposition averaging: e.g., "Superposition averaging is performed on all valid trials".

[0061] Regarding step 200, different from the prior art which uses the absolute value as the cost function (insensitive to the jitter between signals), here the time difference degree between the signals of two trials is calculated to characterize the alignment cost of all signals in the magnetoencephalogram signal. Specifically, the non-linear time mapping between the two signals can be found through dynamic programming to minimize the cumulative distance. First, a distance matrix (such as the Euclidean distance) is constructed, and then the minimum cumulative path from the starting point to the ending point is searched. The slope of the path reflects the time difference degree. Among them, the minimum cumulative distance (difference degree) and the time warping path.

[0062] Regarding step 300, in actual tests, it is not necessary to calculate the alignment cost of the entire trial, because such calculation not only increases the computational cost, but also there is mainly some noise that is not of interest in some time periods. Therefore, a time window of the signal of interest is set, for example, the moment where the auditory P100 is located (about 100 ms), and the alignment cost and the corresponding time similarity weight coefficient are calculated only within the time window.

[0063] In some embodiments of the present invention, refer to Figure 2 , a method for generating magnetoencephalogram evoked signals, further includes:

[0064] Step 400: Generate the cost matrix of the signals of any two trials;

[0065] Step 500: Determine the Euclidean distance between the signals of any two trials according to the cost matrix;

[0066] Each element in the cost matrix represents the Euclidean distance between the trial signals of the corresponding two sampling points.

[0067] Step 600: Determine the minimum cost path according to the Euclidean distance between the signals of any two trials.

[0068] In some embodiments of the present invention, referring to Figure 3 , step 600 includes:

[0069] Step 601: Determine a cumulative distance matrix between the signals of any two trials according to the Euclidean distance;

[0070] Step 602: Determine the minimum cost path according to the cumulative distance matrix.

[0071] Specifically, the Euclidean distance between the p-th sampling point of the signal of trial x and the q-th sampling point of the signal of trial v in the cost matrix is calculated by the following formula.

[0072] (1)

[0073] Subsequently, the cumulative distance matrix is obtained through recursive calculation , and the minimum cost path P between the two endpoints of the signal is calculated therefrom, which is also called the alignment path.

[0074]

[0075] The alignment path describes the cost of minimizing the temporal alignment between two signals.

[0076] In some embodiments of the present invention, the alignment cost is the average value of the distance differences between the matching points on the minimum cost path, and the overall temporal difference degree between the signals of all trials in the magnetoencephalogram signal and the magnetoencephalogram evoked signal is less than a preset value.

[0077] Preferably, the overall temporal difference degree between the signals of all trials in the magnetoencephalogram signal and the magnetoencephalogram evoked signal is as small as possible.

[0078] As can be seen from the above description, the embodiments of the present invention provide a method for generating a magnetoencephalogram evoked signal, including: First, determine the alignment cost of all signals in the magnetoencephalogram signal according to the minimum cost path between the signals of any two trials in the magnetoencephalogram signal, and the minimum cost path is used to characterize the cost of minimizing the temporal alignment between the corresponding two trial signals; Then, determine the temporal similarity weight coefficient of the signal of each trial according to the alignment cost; Finally, generate the magnetoencephalogram evoked signal according to the signal of each trial and the corresponding temporal similarity weight coefficient.

[0079] The averaging technique is an important step in removing noise from MEG data and obtaining a stable waveform. Aiming at the problem that the arithmetic averaging method in the existing technology is difficult to effectively cope with high-intensity noise and signal jitter, the present invention improves the weighted averaging method of minimizing the accurate function, adds the time similarity weight coefficient of the alignment cost to the weight iteration process, and enables each trial to obtain an optimal weight. Finally, simulation and real auditory evoked experiment data are used to verify that the method proposed by the present invention is superior to a variety of existing averaging methods. Therefore, the method provided by the present invention can be regarded as an alternative to traditional averaging, and the obtained results can improve the accuracy of subsequent research.

[0080] To further illustrate the solution, the present invention also provides a specific implementation manner of a method for generating MEG evoked signals, which specifically includes the following content.

[0081] As a neuroimaging technique with high spatio-temporal resolution, magnetoencephalography (MEG) can use non-invasive sensors to measure the weak magnetic fields generated by the activities of intracerebral neural current elements. Similar to the event-related potential (ERP) in electroencephalography (EEG), the event-related field (ERF) in MEG is the evoked response of the brain magnetic field generated by a stimulus. Therefore, ERF has broad prospects in neuroscience and clinical applications. However, the MEG signal is much smaller than the noise interference. Although there are currently various denoising methods, averaging the trial responses is still a necessary step to obtain a stable ERF waveform.

[0082] It can be understood that using the median to replace the arithmetic mean can avoid the influence of extreme values on the results, but it will contain quite strong high-frequency noise, and the median averaging will cause the results to be too robust and thus lose some useful information. As an improvement, trimmed averaging suppresses the extreme values at both ends of the threshold by setting a truncation threshold and assigning smaller weights to them, but it is necessary to select an appropriate threshold according to the data characteristics to achieve a balance between suppressing extreme values and retaining useful information as much as possible.

[0083] In addition to the above methods, weighted averaging is another type of improvement method. Usually, each trial is assigned a weight according to its noise level, which requires the assumption that the noise is stable within a trial, but this may not be satisfied in actual experiments. Based on this, see Figure 4 , a specific implementation manner of a method for generating MEG evoked signals includes the following steps:

[0084] S1: Define a scalar criterion function for the optimal weight.

[0085] Based on the idea of weighted averaging of the criterion function, the weighted averaging can be expressed as:

[0086]

[0087] Where N is the number of trials, k is the sample sampling point, is the weight of the i-th trial, is the sample of the i-th trial, is the average signal. The scalar criterion function for finding the optimal weight is defined as: (6)

[0088] Where is a measure of the dissimilarity of vector parameters, is the weighted exponential parameter. Equation (6) above can be interpreted as a measure of the overall dissimilarity between V and the trial X i Therefore, the aim is to obtain the best average signal by finding the optimal weight vector . Since V is fixed, the minimization of I m is regarded as a constrained optimization problem.

[0089] S2: Determine the weight vector of the signal for each trial.

[0090] Constraining Equation (6) by Equation (5) and setting the Lagrangian gradient to zero, the vector is finally obtained: (7)

[0091] By iteratively and alternately updating the weight W and the average signal , until W converges at the l -th iteration: , is a very small constant preset. Therefore, (7) can be updated as: (8)

[0092] S3: Determine the final optimal average signal.

[0093] Here, the cost function of the absolute value is used, that is . The final optimal average signal can be expressed as: (9)

[0094] S4: Determine the alignment cost of all signals in the magnetoencephalogram signal.

[0095] Specifically, see Formulas (1) to (3).

[0096] S5: Determine the time similarity weight coefficient of the signal for each trial according to the alignment cost.

[0097] During the calculation, the higher the temporal similarity of the trial x to the average signal v, the greater the weight obtained, so as to reduce the influence of trials with greater temporal jitter on v. Therefore, the alignment cost is first calculated using equations (1) and (3). , which is expressed as the average of the distance costs of all matching points of the minimum cost path (the average of the distance differences between paired points on the optimal alignment path).

[0098] (10)

[0099] According to the corresponding temporal similarity weight coefficient is determined :

[0100]

[0101] It should be noted that in actual tests, it is not necessary to calculate the alignment cost for the entire trial, because this not only increases the computational cost, but also there is mainly some noise of no interest in some time periods. Therefore, a time window of the signal of interest is set, for example, the moment where the auditory P100 is located (at about 100 ms), and the alignment cost and the corresponding temporal similarity weight coefficient are only calculated within the time window. In addition, a weight threshold c is added during the calculation, and the weights in the weight W that are less than c after each iterative update are set to zero, so as to further reduce the interference that may be brought by strong outliers. Finally, the flow of a method for generating magnetoencephalogram evoked signals is as follows:

[0102] Step 1. Fix the weighted exponent parameter m, the weight screening threshold c, the iterative convergence constant and the time window of interest, the iteration index , and initialize as the average or median of all trials.

[0103] Step 2. Calculate using formula (7), calculate using formulas (10) to (12), and multiply the vector by each sample and the screening threshold c to update the vector .

[0104] Step 3. Update using formula (9).

[0105] Step 4. If , then and return to Step 2.

[0106] Experimental results:

[0107] The performance of the algorithm provided by the present invention was tested on simulation experiment data and real auditory evoked experiment data respectively. The signals in the simulation experiment consisted of simulated brain nerve signals and actual measured noise. Among them, the simulated nerve signals were used to simulate the M100 component of the auditory stimulus experiment, and were generated by a signal source placed inside a real head model. The signal was a sine wave with a peak of about 500 fT, a frequency of 10 Hz, and a peak time at 100 ms, and the signal was limited within a 50-ms window. A total of 300 trials were generated, and the duration of each trial was 1 s. At the same time, in order to simulate signal jitter, the signal peaks in each trial were randomly distributed between 90 ms and 110 ms. The corresponding waveforms were obtained by 26 extracranial magnetometers through a conduction model. The noise was the empty room noise measured by a sensor array with the same layout in a shielded room without subjects. 300 seconds were measured to correspond to the simulation signals.

[0108] The real auditory evoked experiment data used the publicly available OPM auditory dataset, which was the OPM pure tone stimulus experiment data of 85 channels.

[0109] See Figures 5 to 8 (Due to the limitation of the picture size, in Figures 5 to 8 the weighted average method based on criterion function minimization is abbreviated as the weighted average method), in the experiment, the algorithm proposed by the present invention (hereinafter referred to as the dtwWACFM method) was compared with several existing benchmark algorithms in terms of performance. These include the traditional arithmetic mean algorithm (Arithmetic Mean, AM), the median average algorithm (Median), the trimmed mean algorithm (Trimmed Mean, TM), and the weighted average method based on criterion function minimization (Weighted averaging Based on Criterion Function Minimization, WACFM).

[0110] Experimental settings: The data sampling rate was 1000 Hz. The empty room noise data and the OPM auditory data in the simulation experiment were both band-pass filtered from 1 Hz to 40 Hz. The simulation experiment was additionally denoised by the SSP algorithm after band-pass filtering. After removing the damaged trials (noise amplitude greater than 10000 fT), there were 250 trials in the simulation experiment for testing, and 500 trials in the real auditory experiment for testing.

[0111] In the simulation and real auditory experiments, the threshold for trimmed mean was set to 0.1. The weighted exponent parameter m in WACFM was 2, and the iterative convergence constant was 1e -5 . The weighted exponent parameter m in the method proposed by the present invention was 2, and the iterative convergence constant was 1e -3, the weight screening threshold c is 1e -3 , and the time window of interest is set to 90 ms - 110 ms.

[0112] Evaluation: To quantitatively analyze the performance of dtwWACFM and the benchmark algorithm, the signal-to-noise ratio (SNR) and root mean square error (RMSE) results of the algorithm averaged over different numbers of trials were tested in the simulation experiment.

[0113]

[0114] Among them, is the simulated neural signal, is 's power. is the reconstructed signal after denoising, is the noise power in the reconstructed signal. N is the number of sensors. In addition, since there is no accurate neural signal in the real experiment, the peak signal-to-noise ratio of the average result is calculated.

[0115]

[0116] Among them, represents the signal peak value in the time period of interest after stimulation, represents the variance of the signal noise baseline segment. In addition to quantitative analysis, the waveform results of the simulation and real experiments were also analyzed.

[0117] Figure 5 And Figure 6 shows the SNR and RMSE results of different averaging methods under different numbers of trials in the simulation experiment. All results are the averages obtained from 20 random trials. It can be seen that all other methods have been improved compared to the arithmetic mean. Among them, the method proposed in the present invention (referred to as dtwWACFM here) has the best performance under various numbers of trials. Compared with the second-best WACFM method, the SNR of dtwWACFM has increased by up to 24% (250 trials), with an average increase of 16.2%. The RMSE has increased by up to 15% (250 trials), with an average increase of 10.4%.

[0118] Figure 7shows a representative channel waveform result. The black line is the simulated neural signal, and the red line is the waveform of dtwWACFM. At the evoked signal of 0.1 s, the peak amplitude of dtwWACFM is closest to the simulated signal compared with the results of other methods, indicating that dtwWACFM only has the least signal distortion to the greatest extent. In addition, in the remaining time periods without simulated signal activation, the waveform of dtwWACFM is also closer to zero, suppressing more noise compared with other methods.

[0119] Table 1. SNR_peak results of different methods in the real auditory experiment

[0120]

[0121] In the real auditory experiment, the SNR_peak results of different methods are shown in Table 1. Except that the median result is slightly lower than the arithmetic mean, the trends of other results are similar to those of the simulation experiment. The performance of the dtwWACFM method proposed by the present invention is optimal, at least 4.1% higher than other methods.

[0122] Figure 8 is the waveform result of a channel in the auditory experiment. It can be seen that dtwWACFM well preserves the M100 (near 0.1 s) component and there is no obvious distortion compared with other methods. In addition, in the remaining time periods, including the baseline segment and the non-evoked time periods, the waveform is also closer to zero, better suppressing the noise in the signal.

[0123] In summary, for MEG, EEG or other physiological detection technologies, it is necessary to average multiple trials to remove residual noise and obtain a stable and obvious evoked waveform. However, when the noise intensity is high or there is a certain jitter (not strictly aligned) in the evoked signal, the performance of the existing arithmetic mean method will decline. There are many reasons for signal jitter. Different subjects, or the subjects themselves due to fatigue from long-term repeated experiments or adaptation to stimuli, may cause changes in the evoked signal. In addition, different experimental devices may cause certain random systematic errors in the generation and recording of the stimulus signal. Although improved average methods such as median, trimmed mean, and weighted mean have been proposed, it is still difficult to solve these two problems simultaneously. Therefore, the present invention proposes a method for generating MEG evoked signals, which improves the existing weighted mean method by adding the calculated time similarity weight coefficient to the weight update process, thereby improving the ability of the algorithm to handle signal jitter.

[0124] During iteration, dtwWACFM updates the weights of each trial that can minimize the criterion function, and selects the absolute value of the difference between the trial signal and the average signal as the criterion function. Therefore, this iterative process can be expressed as trials that contribute more (are more similar) to the average signal obtaining higher weights, and conversely, trials with more noise obtaining smaller weights. However, compared with high-intensity noise, small fluctuations in the evoked signals between different trials have much less impact on the criterion function, but these fluctuations will cause changes in the time and amplitude of the evoked peaks in the final average result. The alignment cost can describe the temporal similarity between two signals. The trial signals with smaller fluctuations have a smaller alignment cost with the average signal. Therefore, the present invention further multiplies the weights obtained by the criterion function by the temporal similarity weight coefficient, so that trials with smaller fluctuations obtain larger weights. Based on the above process, dtwWACFM has better performance compared with other averaging methods.

[0125] The method provided by the present invention is verified to have better performance compared with other benchmark averaging methods through simulation and real auditory experiments. dtwWACFM not only has the best SNR and RMSE results, but also the waveform results are closer to the theoretical values or ideal values.

[0126] It should be noted that although the method provided by the present invention is verified on OPM-MEG data, dtwWACFM can be easily extended to the applications of EEG, SQUID-MEG or other physiological detection technologies.

[0127] Based on the same inventive concept, the embodiments of the present invention also provide a device for generating magnetoencephalogram evoked signals, which can be used to implement the method described in the above embodiments, as in the following embodiments. Since the principle of solving problems by the device for generating magnetoencephalogram evoked signals is similar to that of the method for generating magnetoencephalogram evoked signals, the implementation of the device for generating magnetoencephalogram evoked signals can refer to the implementation of the method for generating magnetoencephalogram evoked signals, and the repeated parts will not be described again. As used hereinafter, the term "unit" or "module" can be a combination of software and / or hardware that can implement a predetermined function. Although the systems described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0128] The embodiments of the present invention provide a specific implementation manner of a device for generating magnetoencephalogram evoked signals that can implement the method for generating magnetoencephalogram evoked signals. Refer to Figure 9 A device for generating magnetoencephalogram evoked signals specifically includes the following content:

[0129] An alignment cost determination module 10, configured to determine the alignment cost of all signals in the magnetoencephalogram (MEG) signal according to the minimum cost path between the signals of any two trials in the MEG signal, where the minimum cost path is used to characterize the alignment cost of minimizing time between the signals of the corresponding two trials;

[0130] A time weight coefficient determination module 20, configured to determine the time similarity weight coefficient of the signal of each trial according to the alignment cost;

[0131] An evoked signal generation module 30, configured to generate an MEG evoked signal according to the signal of each trial and the corresponding time similarity weight coefficient.

[0132] In some embodiments of the present invention, referring to Figure 10 , a device for generating an MEG evoked signal further includes:

[0133] A cost matrix generation module 40, configured to generate a cost matrix of the signals of any two trials;

[0134] An Euclidean distance determination module 50, configured to determine the Euclidean distance between the signals of any two trials according to the cost matrix;

[0135] A minimum cost path determination module 60, configured to determine the minimum cost path according to the Euclidean distance between the signals of any two trials.

[0136] In some embodiments of the present invention, referring to Figure 11 , the minimum cost path determination module 60 includes:

[0137] An accumulated distance matrix determination unit 60a, configured to determine an accumulated distance matrix between the signals of any two trials according to the Euclidean distance;

[0138] A minimum cost path determination unit 60b, configured to determine the minimum cost path according to the accumulated distance matrix.

[0139] In some embodiments of the present invention, the alignment cost is the average value of the distance differences between matching points on the minimum cost path, and the overall time difference degree between the signals of all trials in the MEG signal and the MEG evoked signal is less than a preset value.

[0140] Embodiments of the present invention further provide a specific implementation manner of an electronic device capable of implementing all steps in the method for generating an MEG evoked signal in the above embodiments. Referring to Figure 12 , the electronic device specifically includes the following contents:

[0141] A processor 1201, a memory 1202, a communications interface 1203, and a bus 1204;

[0142] Among them, the processor 1201, the memory 1202, and the communications interface 1203 communicate with each other through the bus 1204; the communications interface 1203 is used to implement information transmission between related devices such as server-side devices and client-side devices;

[0143] The processor 1201 is used to call a computer program in the memory 1202. When the processor executes the computer program, all steps in the method for generating magnetoencephalogram evoked signals in the above embodiments are implemented. For example, when the processor executes the computer program, the following steps are implemented:

[0144] Step 100: Determine the alignment cost of all signals in the magnetoencephalogram signal according to the minimum cost path between the signals of any two trials in the magnetoencephalogram signal. The minimum cost path is used to represent the alignment cost of minimizing the time between the signals of the corresponding two trials;

[0145] Step 200: Determine the time similarity weight coefficient of the signal of each trial according to the alignment cost;

[0146] Step 300: Generate a magnetoencephalogram evoked signal according to the signal of each trial and the corresponding time similarity weight coefficient.

[0147] An embodiment of the present invention also provides a computer-readable storage medium capable of implementing all steps in the method for generating magnetoencephalogram evoked signals in the above embodiments. A computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, all steps in the method for generating magnetoencephalogram evoked signals in the above embodiments are implemented. For example, when the processor executes the computer program, the following steps are implemented:

[0148] Step 100: Determine the alignment cost of all signals in the magnetoencephalogram signal according to the minimum cost path between the signals of any two trials in the magnetoencephalogram signal. The minimum cost path is used to represent the alignment cost of minimizing the time between the signals of the corresponding two trials;

[0149] Step 200: Determine the time similarity weight coefficient of the signal of each trial according to the alignment cost;

[0150] Step 300: Generate a magnetoencephalogram evoked signal according to the signal of each trial and the corresponding time similarity weight coefficient.

[0151] Each embodiment in this specification is described in a progressive manner. For the parts that are the same or similar among the embodiments, reference can be made to each other, and the key point of each embodiment is to illustrate the differences from other embodiments. In particular, for the hardware + program type of embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and for the relevant parts, reference can be made to the partial description of the method embodiments.

[0152] The above describes specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be executed in a different order from that in the embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0153] Although the present invention provides method operation steps such as in the embodiments or flowcharts, based on routine or non-creative labor, there can be more or fewer operation steps. The step order listed in the embodiments is only one way among the execution orders of numerous steps and does not represent the only execution order. When the actual device or client product is executed, it can be executed in the method order shown in the embodiments or the drawings or in parallel (such as in an environment of parallel processors or multi-threaded processing).

[0154] For the convenience of description, when describing the above device, it is divided into various modules according to functions for separate description. Of course, when implementing the embodiments of this specification, the functions of each module can be realized in the same or multiple software and / or hardware, or the modules implementing the same function can be realized by a combination of multiple sub-modules or sub-units, etc. The device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the couplings or direct couplings or communication connections shown or discussed with each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical or other forms.

[0155] Those skilled in the art also know that, in addition to implementing the controller in the form of pure computer-readable program code, the method steps can be logically programmed to enable the controller to implement the same functions in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc. Therefore, such a controller can be considered a hardware component, and the devices included therein for implementing various functions can also be regarded as the structures within the hardware component. Or even, the devices for implementing various functions can be regarded as either software modules for implementing the method or structures within the hardware component.

[0156] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and a memory.

[0157] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM) and / or non-volatile memory such as read-only memory (ROM) or flash RAM). The memory is an example of computer-readable media.

[0158] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and the relevant parts can refer to the description of the method embodiments. In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the embodiments of this specification. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0159] The above is only the embodiments of this specification and is not used to limit the embodiments of this specification. For those skilled in the art, various changes and modifications can be made to the embodiments of this specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the embodiments of this specification shall be included within the scope of the claims of the embodiments of this specification.

Claims

1. A method for generating magnetoencephalogram evoked signals, characterized in that, Including: Determining the alignment cost of all signals in the magnetoencephalogram (MEG) signal based on the minimum cost path between the signals of any two trials in the MEG signal, where the minimum cost path is used to characterize the alignment cost of minimizing time between the signals of the corresponding two trials; Determining the time similarity weight coefficient of the signal of each trial according to the alignment cost; Generating a magnetoencephalogram evoked signal based on the signal of each trial and the corresponding time similarity weight coefficient.

2. The method for generating a magnetoencephalogram evoked signal according to claim 1, wherein Further including: Generating a cost matrix for the signals of any two trials; Determining the Euclidean distance between the signals of any two trials according to the cost matrix; Determining the minimum cost path according to the Euclidean distance between the signals of any two trials.

3. The method for generating magnetoencephalogram evoked signals according to claim 2, wherein Determining the minimum cost path according to the Euclidean distance between the signals of any two trials includes: Determining a cumulative distance matrix between the signals of any two trials according to the Euclidean distance; Determining the minimum cost path according to the cumulative distance matrix.

4. The method for generating magnetoencephalogram evoked signals according to claim 1, wherein The alignment cost is the average value of the distance differences between matching points on the minimum cost path, and the overall time difference degree between the signals of all trials in the MEG signal and the magnetoencephalogram evoked signal is less than a preset value.

5. A generating device for magnetoencephalogram evoked signals, characterized in that Including: An alignment cost determination module, configured to determine the alignment cost of all signals in the MEG signal based on the minimum cost path between the signals of any two trials in the MEG signal, where the minimum cost path is used to characterize the alignment cost of minimizing time between the signals of the corresponding two trials; A time weight coefficient determination module, configured to determine the time similarity weight coefficient of the signal of each trial according to the alignment cost; An evoked signal generation module, configured to generate a magnetoencephalogram evoked signal based on the signal of each trial and the corresponding time similarity weight coefficient.

6. The apparatus for generating magnetoencephalogram evoked signals according to claim 5, wherein Further including: A cost matrix generation module, configured to generate a cost matrix for the signals of any two trials; A Euclidean distance determination module, configured to determine the Euclidean distance between the signals of any two trials according to the cost matrix; A minimum cost path determination module, configured to determine the minimum cost path according to the Euclidean distance between the signals of any two trials.

7. The apparatus for generating magnetoencephalogram evoked signals according to claim 6, wherein The minimum cost path determination module includes: A cumulative distance matrix determination unit, configured to determine a cumulative distance matrix between the signals of any two trials according to the Euclidean distance; A minimum cost path determination unit, configured to determine the minimum cost path according to the cumulative distance matrix.

8. The apparatus for generating magnetoencephalogram evoked signals according to claim 5, wherein The alignment cost is the average value of the distance differences between matching points on the minimum cost path, and the overall time difference degree between the signals of all trials in the MEG signal and the magnetoencephalogram evoked signal is less than a preset value.

9. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by a processor, the steps of the method for generating a magnetoencephalogram evoked signal according to any one of claims 1 to 4 are implemented.

10. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, the steps of the method for generating a magnetoencephalogram evoked signal according to any one of claims 1 to 4 are implemented.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method for generating magnetoencephalogram evoked signals according to any one of claims 1 to 4.

Images