Neural electrophysiological spike signal sorting method based on small batch k-means clustering

Abstract

The application discloses a kind of neural electrophysiological spike signal sorting methods based on small batch K-means clustering, comprising:1, acquisition electrophysiological data and pretreatment, to eliminate noise and enhance signal quality;2, using optimized clustering algorithm small batch K-means, obtain general template from data, based on general template, the nearest L channel of detected spike is carried out principal component feature extraction, and updated general template is obtained;3, based on graph clustering algorithm, the updated general template is accurately clustered, so as to obtain learning template;4, using learning template carries out spike detection and combines the graph clustering method of step 3 and outputs final spike clustering result.The application can significantly improve the processing speed and classification accuracy of electrophysiological data, realize the rapid and accurate identification of neuronal activity, provide a kind of efficient data analysis tool for neuroscience research, significantly reduce the time cost of data processing, and improve research efficiency.

Description

Technical Field

[0001] The present invention belongs to the field of neuroscience, and specifically is a method for sorting neural electrophysiological spike signals based on small-batch K-means clustering. Background Art

[0002] Neuroelectrophysiological signal analysis is an important area of ​​neuroscience research, involving the recording and sorting of electrical signals generated by neuronal activity. With the advancement of neuroscience and biomedical engineering, the demand for accurate sorting of neuroelectrophysiological signals is growing. However, existing technologies still have significant shortcomings in processing speed and classification accuracy, which limits their effectiveness in practical applications.

[0003] First, traditional electrophysiological signal processing methods often rely on manual manipulation and empirical judgment, which is not only time-consuming and labor-intensive but also susceptible to subjective operator influence, making it difficult to ensure the consistency and accuracy of the results. Especially when processing large amounts of electrophysiological data, manual analysis is extremely inefficient and cannot meet the high data processing speed requirements of modern neuroscience research. Second, existing signal classification algorithms often use traditional clustering methods, such as K-means clustering. When processing high-dimensional data, these methods often face problems such as inappropriate cluster center selection and local optimal solutions, which affect the accuracy of classification results. Furthermore, traditional clustering algorithms struggle to effectively distinguish different types of signals when faced with complex electrophysiological signals, especially when there is a high degree of similarity between signals or noise interference, resulting in suboptimal classification results. In terms of signal preprocessing, existing technologies often use simple filtering and denoising methods. These methods may not be able to effectively eliminate noise when processing complex signals, resulting in reduced accuracy of subsequent analysis. In particular, the presence of noise in neuroelectrophysiological signals can significantly affect the detection and classification of spikes, thereby compromising the understanding of neuronal activity. Furthermore, with advancements in electrophysiological data acquisition technology, data volumes have exploded. Traditional methods are struggling to process large amounts of data and are unable to quickly and accurately extract useful information. This not only increases the time required for data processing but also limits researchers' ability to deeply analyze and understand the data.

[0004] In summary, existing technologies have significant shortcomings in terms of processing speed and classification accuracy for electrophysiological signals. These issues have limited the advancement of neuroscience research and the development of clinical applications. Therefore, there is an urgent need to develop a new method for sorting neuroelectrophysiological spike signals to improve data processing speed and classification accuracy, enabling rapid and accurate identification of neuronal activity. Summary of the Invention

[0005] In order to address the shortcomings of the above-mentioned existing technologies, the present invention proposes a neural electrophysiological spike signal sorting method based on small-batch K-means clustering, in order to significantly improve the processing speed and classification accuracy of electrophysiological data, thereby realizing rapid and accurate identification of neuronal activity.

[0006] In order to achieve the above-mentioned object, the present invention adopts the following technical solutions:

[0007] The present invention provides a method for sorting neural electrophysiological spike signals based on small-batch K-means clustering, which comprises the following steps:

[0008] Step 1: Collect the electrophysiological signals of the mouse visual cortex and perform preprocessing, including: common average reference, time filtering, channel whitening and drift correction, to obtain the corrected electrophysiological signals ;

[0009] Step 2: Use the mini-batch K-means clustering algorithm to Process and obtain cluster centers and serves as a general template;

[0010] Step 3: Use a common template right Perform peak detection and obtain K peaks closest to the detected peaks The spike signals of each channel are extracted and the principal component features are extracted to obtain the corresponding The peak features of the channels are collected, and the peak feature with the largest amplitude is assigned to the corresponding cluster center, so as to obtain the updated general template ,in, Indicates the electrophysiological spike features, and , Indicates the The electrophysiological spike features are The value on each channel, Indicates the number of channels to which the electrophysiological spike signal is expanded;

[0011] Step 4: Based on graph clustering algorithm Process and obtain the learning template ;

[0012] Step 5: Based on the learning template right After spike detection, clustering is performed according to the process of step 4 to output the final electrophysiological spike signal cluster set. ,in, Indicates the electrophysiological spike signal clusters, , .

[0013] The neural electrophysiological spike signal sorting method based on small batch K-means clustering described in the present invention is also characterized in that step 1 includes the following steps:

[0014] Step 1.1: Acquire electrophysiological signals from the mouse visual cortex , ,in, Indicates the electrophysiological signals at each time point, and , , Indicates the The first time point Electrophysiological signals on channels, Represents the total duration of the electrophysiological signal, Indicates the total number of channels of electrophysiological signals, represents transpose;

[0015] Step 1.2: Calculate the average value of the electrophysiological signal of the i-th channel ;

[0016] Step 1.3: Construct an average reference signal ;

[0017] Step 1.4: Calculate the common-mean referenced electrophysiological signal ;

[0018] Step 1.5: Perform time filtering to obtain the filtered electrophysiological signal ;

[0019] Step 1.6: Perform channel whitening to obtain the whitened electrophysiological signal ;

[0020] Step 1.7: Perform drift correction to obtain the corrected electrophysiological signal .

[0021] Furthermore, step 2 includes the following steps:

[0022] Step 2.1: Setup is the number of cluster centers, and the cluster centers are initialized using the K-Means++ algorithm , ,in, Indicates the cluster centers, and ,in, Indicates the The cluster center is in the The value on the dimensional channel;

[0023] Step 2.2: From Randomly select a dimension of size A subset of electrophysiological signals , ,in, Indicates the The electrophysiological signal after correction at each time point, and , , Indicates the The first time point The electrophysiological signal after correction on each channel, represents the total duration of the electrophysiological signal subset Z, , represents the total number of channels of the electrophysiological signal subset Z, ;

[0024] Step 2.3: Calculation The distance from each cluster center and Assigned to the cluster center corresponding to the minimum distance, so that Each electrophysiological signal in is assigned to the corresponding cluster center;

[0025] Step 2.4: Update the kth cluster center using formula (1) , get the updated k-th cluster center :

[0026] (1)

[0027] In formula (1), yes The corresponding electrophysiological signal clustering, yes The number of electrophysiological signals in represents the adaptive learning rate;

[0028] Step 2.5: If and The difference between them meets the threshold , it means that the cluster center no longer changes, and the updated cluster center is output This is the general template for electrophysiological signal spike detection, where , Indicates the updated Cluster centers In the The value on the dimension channel; if the threshold is not met , then Assignment Then, return to step 2.2 and execute the sequence.

[0029] Furthermore, step 4 includes the following steps:

[0030] Step 4.1: From Randomly select a dimension of size A subset of electrophysiological spike features , ,in, Indicates the electrophysiological spike features, and , , Indicates the The electrophysiological spike features are The value on each channel, represents the number of spike features in the selected electrophysiological spike subset, , The number of channels representing the subset of electrophysiological spikes, ;

[0031] Step 4.2: The corresponding cluster set is recorded as ,in, for Clustering of spike signals;

[0032] Step 4.3: Calculate clusters based on the correlation coefficient method The similarity between any two electrophysiological spike signals in is calculated. If the similarity is greater than the set threshold, the corresponding two electrophysiological spike signals are connected by edges; otherwise, no edge is connected.

[0033] Step 4.4: Use Equation (2) to construct a modular cost function :

[0034] (2)

[0035] In formula (2), It is clustering The total number of edges connecting all pairs of spike signals in , It is clustering The total number of edges connecting pairs of inner spikes, It is clustering The sum of the number of edges corresponding to each electrophysiological spike signal in is a parameter that controls the number of clusters;

[0036] Step 4.5: Calculate electrophysiological spike characteristics based on the correlation coefficient method Clustering with electrophysiological spike signals The number of edges connected to the rest of the spike signals Hedi Electrophysiological spike features Clustering of spike signals The number of edges connected to the rest of the spike signals ;

[0037] like Make , then the electrophysiological spike signal Join Otherwise, remain unchanged, thus obtaining the updated electrophysiological signal clustering set ,in, Indicates the An updated electrophysiological signal cluster, Indicates the An updated electrophysiological signal cluster;

[0038] Step 4.6: Calculate according to formula (3) and Modularity cost function value before merging :

[0039] (3)

[0040] In formula (3), It is clustering The total number of edges connecting pairs of inner spikes, express The total number of edges connecting pairs of inner spikes, yes The sum of the number of edges corresponding to each electrophysiological spike signal in yes The sum of the number of edges corresponding to each electrophysiological spike signal in Indicates a fixed constant;

[0041] Step 4.7: Calculate according to formula (4) and The modular cost function value after merging :

[0042] (4)

[0043] In formula (4), It is clustering and the total number of edges connecting pairs of inner spikes;

[0044] Step 4.8: Calculate the difference before and after merging according to formula (5) :

[0045] (5)

[0046] Step 4.9: When When and Merge into a new cluster;

[0047] Step 4.10: Follow the process from step 4.1 to 4.9 to The clusters where the electrophysiological spike features are located are merged, so that the new clusters after the merger are recorded as ,in, is the new cluster after the mth merger, The corresponding cluster center is recorded as , thus The corresponding cluster center and serve as a learning template; , ; M represents the number of new clusters after merging.

[0048] The electronic device of the present invention includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the neural electrophysiological spike signal sorting method, and the processor is configured to execute the program stored in the memory.

[0049] The present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program executes the steps of the neural electrophysiological spike signal sorting method when the computer program is executed by a processor.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] 1. Step 2 of the present invention optimizes the electrophysiological signal data by adopting a small-batch K-means clustering algorithm. Compared with the traditional K-means clustering algorithm, the small-batch K-means clustering algorithm can process large-scale data sets more quickly because it only processes a subset of the data in each iteration, rather than the entire data set. This method reduces the demand for computing resources and increases the speed of data processing. At the same time, by introducing an adaptive learning rate, the present invention can more accurately update the cluster center, thereby improving the accuracy of electrophysiological signal classification. This method overcomes the problems of slow processing speed and insufficient classification accuracy in the prior art, and significantly improves the processing efficiency of electrophysiological data and the reliability of the results.

[0052] 2. Step three of the present invention further optimizes the sorting process of electrophysiological spike signals by combining graph clustering algorithms and principal component feature extraction technology. A universal template is used for spike detection, and principal component feature extraction is performed on the detected spike signal. This step helps to extract more representative signal features, enhance the signal discrimination, and update the universal template. A graph clustering algorithm is used to accurately cluster the updated universal template to obtain a learning template. This method effectively solves the problem of difficulty in distinguishing highly similar signals in the prior art, realizes rapid and accurate identification of neuronal activity, and provides an efficient data analysis tool for neuroscience research. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a flow chart of a neural electrophysiological spike signal sorting method based on small batch K-means clustering of the present invention. DETAILED DESCRIPTION

[0054] In this embodiment, a neural electrophysiological spike signal sorting method based on small batch K-means clustering is described. Figure 1 As shown, the following steps are included:

[0055] Step 1: Collect the electrophysiological signals of the mouse visual cortex and perform preprocessing, including: common average reference, time filtering, channel whitening and drift correction, to obtain the corrected electrophysiological signals :

[0056] Step 1.1: Acquire electrophysiological signals from the mouse visual cortex , ,in, Indicates the electrophysiological signals at each time point, and , , Indicates the The first time point Electrophysiological signals on channels, Represents the total duration of the electrophysiological signal, Indicates the total number of channels of electrophysiological signals, Indicates transpose.

[0057] Step 1.2: Calculate the average value of the electrophysiological signal of the i-th channel ;

[0058] Step 1.3: Construct an average reference signal ;

[0059] Step 1.4: Calculate the common-mean referenced electrophysiological signal ;

[0060] Step 1.5: Compute the impulse response of the second order Butterworth filter using formula (1) for its transfer function , thus the filtered signal is computed by a convolution operation of formula (2) :

[0061] (1)

[0062] In formula (1), denotes the cut-off frequency, denotes the unit step function.

[0063] (2)

[0064] In formula (2), denotes the convolution operation.

[0065] Step 1.6: Channel whitening of results in a whitened electrophysiological signal ;

[0066] Step 1.6.1: Compute the covariance matrix of the filtered signal according to formula (3) :

[0067] (3)

[0068] In formula (3), is the transpose of .

[0069] Step 1.6.2: Construct the diagonal matrix according to formula (4)

[0070] (4)

[0071] In formula (4), denotes the eigenvalues of the covariance matrix .

[0072] Step 1.6.3: Compute the inverse square root of each eigenvalue in the diagonal matrix according to formula (5) and construct a new diagonal matrix :

[0073] (5)

[0074] Step 1.6.4: Compute the whitening matrix according to formula (6)

[0075] (6)

[0076] In formula (4), Represents the covariance matrix The eigenvector of .

[0077] Step 1.6.5: Calculate the whitened electrophysiological signal according to formula (7) :

[0078] (7)

[0079] Step 1.7: Perform drift correction to obtain the corrected electrophysiological signal .

[0080] Step 2: Use the mini-batch K-means clustering algorithm to Process and obtain cluster centers And as a general template:

[0081] Step 2.1: Setup is the number of cluster centers, and the cluster centers are initialized using the K-Means++ algorithm , ,in, Indicates the cluster centers, and ,in, Indicates the The cluster center is in the The value on the dimensional channel.

[0082] Step 2.2: From Randomly select a dimension of size A subset of electrophysiological signals , ,in, Indicates the The electrophysiological signal after correction at each time point, and , , Indicates the The first time point The electrophysiological signal after correction on each channel, represents the total duration of the electrophysiological signal subset Z, , represents the total number of channels of the electrophysiological signal subset Z, .

[0083] Step 2.3: Calculation The distance from each cluster center and Assigned to the cluster center corresponding to the minimum distance, so that Each electrophysiological signal in is assigned to the corresponding cluster center;

[0084] Step 2.4: Update the kth cluster center using formula (8) , get the updated k-th cluster center :

[0085] (8)

[0086] In formula (8), yes The corresponding electrophysiological signal clustering, yes The number of electrophysiological signals in represents the adaptive learning rate.

[0087] Step 2.5: If and The difference between them meets the threshold , it means that the cluster center no longer changes, and the updated cluster center is output This is the general template for electrophysiological signal spike detection, where , Indicates the updated Cluster centers In the The value on the dimension channel; if the threshold is not met , then Assignment Then, return to step 2.2 and execute the sequence.

[0088] Step 3: Use a common template right Perform peak detection and obtain K peaks closest to the detected peaks The spike signals of each channel are extracted and the principal component features are extracted to obtain the corresponding The peak features of the channels are collected, and the peak feature with the largest amplitude is assigned to the corresponding cluster center, so as to obtain the updated general template ,in, Indicates the electrophysiological spike features, and , Indicates the The electrophysiological spike features are The value on each channel, Indicates the number of channels to which the electrophysiological spike signal is expanded.

[0089] Step 4: Based on the graph clustering algorithm Processing, get learning template :

[0090] Step 4.1: From a dimension size of is randomly selected , wherein represents the th electrophysiological spike feature, and , , represents the value of the th electrophysiological spike feature on the th channel, represents the number of spike features in the selected electrophysiological spike subset, , represents the number of channels of the electrophysiological spike subset, .

[0091] Step 4.2: the cluster set corresponding to is recorded as , wherein is the cluster of the spike signal in which is located;

[0092] Step 4.3: according to the correlation coefficient method, the similarity of any two electrophysiological spike signals in the cluster is calculated, if the similarity is greater than the set threshold, then the edge between the corresponding two electrophysiological spike signals is connected; otherwise, the edge is not connected.

[0093] Step 4.4: a modular cost function is constructed using formula (9):

[0094] (9)

[0095] In formula (9), is the total number of edges connected between all spike signal pairs in the cluster , is the total number of edges connected between spike signal pairs within the cluster , is the sum of the number of edges corresponding to each electrophysiological spike signal within the cluster , and is a parameter for controlling the number of clusters. Step 4.5: according to the correlation coefficient method, the number of edges connected between the

[0096] th electrophysiological spike feature and the remaining spike signals in the cluster , and the number of edges connected between the th electrophysiological spike feature and the remaining spike signals in the cluster of the spike signal in which the th electrophysiological spike feature is located are calculated, respectively.;

[0097] like Make , then the electrophysiological spike signal Join Otherwise, it remains unchanged, thus obtaining the updated electrophysiological signal clustering set ,in, Indicates the An updated electrophysiological signal cluster, Indicates the An updated electrophysiological signal clustering.

[0098] Step 4.6: Calculate according to formula (10) and Modularity cost function value before merging :

[0099] (10)

[0100] In formula (10), It is clustering The total number of edges connecting pairs of inner spikes, express The total number of edges connecting pairs of inner spikes, yes The sum of the number of edges corresponding to each electrophysiological spike signal in yes The sum of the number of edges corresponding to each electrophysiological spike signal in Indicates a fixed constant.

[0101] Step 4.7: Calculate according to formula (11) and The modular cost function value after merging :

[0102] (11)

[0103] In formula (11), It is clustering and The total number of edges connecting pairs of inner spikes.

[0104] Step 4.8: Calculate the difference before and after merging according to formula (12) :

[0105] (12)

[0106] Step 4.9: When When and merge into a new cluster;

[0107] Step 4.10: merge the clusters in which the electrophysiological spike features are located according to the process of steps 4.1 to 4.9, thereby obtaining a new merged cluster, denoted as , wherein is the mth new merged cluster, and M represents the number of new merged clusters; and the cluster center corresponding to is recorded as , thereby obtaining the cluster center corresponding to ; and as a learning template; , ;

[0108] Step 5: based on the learning template , perform spike detection on , and then perform clustering according to the process of step 4, thereby outputting a final electrophysiological spike signal cluster set , wherein represents the i th electrophysiological spike signal cluster, , , .

[0109] In this embodiment, an electronic device includes a memory for storing a program supporting a processor to execute the above method, and the processor is configured to execute the program stored in the memory.

[0110] In this embodiment, a computer readable storage medium has a computer program stored thereon, and the computer program is executed by a processor to perform the steps of the above method.​

Claims

1. A neural electrophysiological spike signal sorting method based on small batch K-means clustering, characterized in that: The steps include: Step 1: Collect the electrophysiological signals of the mouse visual cortex and perform preprocessing, including: common average reference, time filtering, channel whitening and drift correction, to obtain the corrected electrophysiological signals ; Step 2: Use the mini-batch K-means clustering algorithm to Process and obtain cluster centers and serves as a general template; Step 3: Use a common template right Perform peak detection and obtain K peaks closest to the detected peaks The spike signals of each channel are extracted and the principal component features are extracted to obtain the corresponding The peak features of the channels are collected, and the peak feature with the largest amplitude is assigned to the corresponding cluster center, so as to obtain the updated general template ,in, Indicates the electrophysiological spike features, and , Indicates the The electrophysiological spike features are The value on each channel, Indicates the number of channels to which the electrophysiological spike signal is expanded; Step 4: Based on graph clustering algorithm Process and obtain the learning template ; Step 5: Based on the learning template right After spike detection, clustering is performed according to the process in step 4 to output the final electrophysiological spike signal cluster set. ,in, Indicates the electrophysiological spike signal clusters, , .

2. A neural electrophysiological spike signal sorting method based on small batch K-means clustering according to claim 1, characterized in that: Step 1 includes the following steps: Step 1.1: Acquire electrophysiological signals from the mouse visual cortex , ,in, Indicates the electrophysiological signals at each time point, and , , Indicates the The first time point Electrophysiological signals on channels, Represents the total duration of the electrophysiological signal, Indicates the total number of channels of electrophysiological signals, represents transpose; Step 1.2: Calculate the average value of the electrophysiological signal of the i-th channel ; Step 1.3: Construct an average reference signal ; Step 1.4: Calculate the common-mean referenced electrophysiological signal ; Step 1.5: Perform time filtering to obtain the filtered electrophysiological signal ; Step 1.6: Perform channel whitening to obtain the whitened electrophysiological signal ; Step 1.7: Perform drift correction to obtain the corrected electrophysiological signal .

3. The method for sorting neural electrophysiological spike signals based on small batch K-means clustering according to claim 2, characterized in that: Step 2 includes the following steps: Step 2.1: Setup is the number of cluster centers, and the cluster centers are initialized using the K-Means++ algorithm , ,in, Indicates the cluster centers, and ,in, Indicates the The cluster center is in the The value on the dimensional channel; Step 2.2: From Randomly select a dimension of size A subset of electrophysiological signals , ,in, Indicates the The electrophysiological signal after correction at each time point, and , , Indicates the The first time point The electrophysiological signal after correction on each channel, represents the total duration of the electrophysiological signal subset Z, , represents the total number of channels of the electrophysiological signal subset Z, ; Step 2.3: Calculation The distance from each cluster center and Assigned to the cluster center corresponding to the minimum distance, so that Each electrophysiological signal in is assigned to the corresponding cluster center; Step 2.4: Update the kth cluster center using formula (1) , get the updated k-th cluster center : (1) In formula (1), yes The corresponding electrophysiological signal clustering, yes The number of electrophysiological signals in represents the adaptive learning rate; Step 2.5: If and The difference between them meets the threshold , it means that the cluster center no longer changes, and the updated cluster center is output This is the general template for electrophysiological signal spike detection, where , Indicates the updated Cluster centers In the The value on the dimension channel; if the threshold is not met , then Assignment Then, return to step 2.2 and execute the sequence.

4. The method for sorting neural electrophysiological spike signals based on small-batch K-means clustering according to claim 3, characterized in that: Step 4 includes the following steps: Step 4.1: From Randomly select a dimension of size A subset of electrophysiological spike features , ,in, Indicates the electrophysiological spike features, and , , Indicates the The electrophysiological spike features are The value on each channel, represents the number of spike features in the selected electrophysiological spike subset, , The number of channels representing the subset of electrophysiological spikes, ; Step 4.2: The corresponding cluster set is recorded as ,in, for Clustering of spike signals; Step 4.3: Calculate clusters based on the correlation coefficient method The similarity between any two electrophysiological spike signals in is calculated. If the similarity is greater than the set threshold, the corresponding two electrophysiological spike signals are connected by edges; otherwise, no edge is connected. Step 4.4: Use Equation (2) to construct a modular cost function : (2) In formula (2), It is clustering The total number of edges connecting all pairs of spike signals in , It is clustering The total number of edges connecting pairs of inner spikes, It is clustering The sum of the number of edges corresponding to each electrophysiological spike signal in is a parameter that controls the number of clusters; Step 4.5: Calculate electrophysiological spike characteristics based on the correlation coefficient method Clustered with electrophysiological spike signals The number of edges connected to the rest of the spike signals Hedi Electrophysiological spike features Clustering of spike signals The number of edges connected to the rest of the spike signals ; like Make , then the electrophysiological spike signal Join Otherwise, it remains unchanged, thus obtaining the updated electrophysiological signal clustering set ,in, Indicates the An updated electrophysiological signal cluster, Indicates the An updated electrophysiological signal cluster; Step 4.6: Calculate according to formula (3) and Modularity cost function value before merging : (3) In formula (3), It is clustering The total number of edges connecting pairs of inner spikes, express The total number of edges connecting pairs of inner spikes, yes The sum of the number of edges corresponding to each electrophysiological spike signal in yes The sum of the number of edges corresponding to each electrophysiological spike signal in Indicates a fixed constant; Step 4.7: Calculate according to formula (4) and The modular cost function value after merging : (4) In formula (4), It is clustering and the total number of edges connecting pairs of inner spikes; Step 4.8: Calculate the difference before and after merging according to formula (5) : (5) Step 4.9: When When and Merge into a new cluster; Step 4.10: Follow the process from step 4.1 to 4.9 to The clusters where the electrophysiological spike features are located are merged, so that the new clusters after the merger are recorded as ,in, is the new cluster after the mth merger, The corresponding cluster center is recorded as , thus The corresponding cluster center and serve as a learning template; , ; M represents the number of new clusters after merging.

5. An electronic device comprising a memory and a processor, characterized in that: The memory is used to store a program that supports the processor to execute the neural electrophysiological spike signal sorting method according to any one of claims 1 to 4, and the processor is configured to execute the program stored in the memory.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the neural electrophysiological spike signal sorting method according to any one of claims 1 to 4 are executed.

Images