Brain electroencephalogram signal processing method, device, circuit, system and equipment
Through low-bit quantization technology and clustering method, the calculation delay problem of implantable brain-computer interface system when processing high-spatial-time resolution EEG signals is solved, real-time and low-power EEG signal decoding is realized, and data is reduced and transmission bandwidth consumption.
Patent Information
- Application Number
- CN202411081845.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-08
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-08-08
AI Technical Summary
The implantable brain-computer interface system has calculation delay problems when processing high-spatial-time resolution EEG signals, and is sensitive to power consumption, sampling rate and real-time, making it difficult to reduce calculation delay based on low power consumption.
Low-bit quantization technology is used to compress the EEG signal, and the discharge rate data of virtual neurons is obtained through clustering, reducing processing complexity and computing resource overhead, and real-time decoding of intention information is achieved.
It effectively reduces the calculation delay and data volume overhead of high-spatial-time resolution EEG signal processing, realizes real-time and low-power EEG signal decoding, reduces data storage and transmission requirements, is good scalable and has high adaptability.
Smart Images

Figure CN118605735B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of brain-computer interface technology, and in particular to an electroencephalogram (EEG) signal processing method, device, circuit, system and equipment. Background Art
[0002] Brain-Computer Interface (BCI) is a technology that directly connects the human brain to external devices, allowing computers or other mechanical systems to be controlled through brain activity, or information to be transmitted from the computer to the brain. The development of brain-computer interface systems has brought great potential and innovation to many fields such as medical treatment, sports rehabilitation, assistive technology, augmented reality and virtual reality.
[0003] Generally, brain-computer interface systems can include non-implantable brain-computer interface systems and implantable brain-computer interface systems. In related technologies, traditional non-implantable brain-computer interface systems can obtain control instructions or neural intentions corresponding to the above-mentioned EEG signals based on the collected single-modal or multi-modal EEG signals. However, due to the low spatiotemporal resolution of the EEG signals collected by the non-implantable brain-computer interface system, it is difficult for the non-implantable brain-computer interface system to handle more complex and sophisticated brain-computer interaction tasks.
[0004] The implantable brain-computer interface system can collect EEG signals with high spatiotemporal resolution, and then can process more complex and sophisticated brain-computer interaction tasks based on the above-mentioned high spatiotemporal resolution EEG signals. However, when the implantable brain-computer interface system processes the above-mentioned high spatiotemporal resolution EEG signals, it will generate a large amount of data overhead and transmission bandwidth consumption, which will easily introduce higher computing delays. In addition, the implantable brain-computer interface system is more sensitive to power consumption, sampling rate and real-time performance of EEG signals. Therefore, how to reduce the computing delay of the implantable brain-computer interface system when processing EEG signals on the basis of ensuring low power consumption of the implantable brain-computer interface system, high sampling rate of EEG signals and high number of channels is a technical problem that needs to be solved urgently in this field. Summary of the invention
[0005] The present invention provides an electroencephalogram (EEG) signal processing method, device, circuit, system and equipment to solve the defect of computational delay in the prior art when an implantable brain-computer interface system processes EEG signals with high temporal and spatial resolution, thereby reducing the computational delay when the implantable brain-computer interface system processes EEG signals while ensuring low power consumption, high sampling rate and high number of channels of the EEG signals.
[0006] The present invention provides an electroencephalogram signal processing method, comprising the following steps.
[0007] Using an implantable EEG signal acquisition device to acquire EEG signals in the cerebral cortex of a target subject at a predefined sampling rate;
[0008] For any channel of the implantable EEG signal acquisition device, when it is determined that the current sampling period ends, based on the EEG signal acquired by the implantable EEG signal acquisition device through the any channel in the current sampling period, a target EEG signal corresponding to the any channel in the current sampling period is acquired, and based on a low-bit quantization technology, the target EEG signal corresponding to the any channel in the current sampling period is compressed to acquire target compressed data corresponding to the any channel in the current sampling period, wherein the EEG signal includes an action potential signal;
[0009] Clustering the target compressed data corresponding to any channel in the current sampling period, and determining the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs;
[0010] When it is determined that the current processing cycle is over, based on the target compressed data included in each cluster corresponding to any channel in the current processing cycle and the target compressed data included in each cluster corresponding to any channel in the previous processing cycle, the discharge rate data of each virtual neuron corresponding to any channel in the current processing cycle is obtained, and any processing cycle includes multiple sampling cycles;
[0011] Based on the discharge rate data of each virtual neuron corresponding to each channel in the current processing cycle, the intention information of the target subject in the current processing cycle is calculated.
[0012] According to an EEG signal processing method provided by the present invention, based on the low-bit quantization technology, compressing the target EEG signal corresponding to any channel in the current sampling period, and obtaining the target compressed data corresponding to any channel in the current sampling period, includes:
[0013] Determine any two adjacent sampling points in the target EEG signal corresponding to any channel in the current sampling period as a sampling point pair, and for each sampling point pair in the target EEG signal corresponding to any channel in the current sampling period, encode the target EEG signal based on the voltage rise and fall relationship of the action potential signals of the two sampling points in each sampling point pair, and obtain the encoded data corresponding to each sampling point pair;
[0014] The encoding data corresponding to each sampling point pair in the target EEG signal corresponding to any channel in the current sampling period are arranged in time sequence to obtain the target compressed data corresponding to any channel in the current sampling period.
[0015] According to an EEG signal processing method provided by the present invention, encoding the target EEG signal based on the voltage rise and fall relationship of the action potential signals of the two sampling points in each sampling point pair, and obtaining the coded data corresponding to each sampling point pair, including:
[0016] In the case where the voltage of the action potential signal of the sampling point preceding the timing in each sampling point pair is less than the voltage of the action potential signal of the sampling point following the timing, the coding data corresponding to each sampling point pair is determined to be 1; in the case where the voltage of the action potential signal of the sampling point preceding the timing in each sampling point pair is not less than the voltage of the action potential signal of the sampling point following the timing, the coding data corresponding to each sampling point pair is determined to be 0,
[0017] or,
[0018] When the voltage of the action potential signal of the sampling point preceding the timing in each sampling point pair is less than the voltage of the action potential signal of the sampling point following the timing, the encoding data corresponding to each sampling point pair is determined to be 0; when the voltage of the action potential signal of the sampling point preceding the timing in each sampling point pair is not less than the voltage of the action potential signal of the sampling point following the timing, the encoding data corresponding to each sampling point pair is determined to be 1.
[0019] According to an electroencephalogram signal processing method provided by the present invention, clustering the target compressed data corresponding to any channel in the current sampling period to determine the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs includes:
[0020] Calculating the Euclidean distance between the target compressed data corresponding to any channel in the current sampling period and the feature information of the cluster center of each cluster corresponding to any channel, and then determining the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs based on each Euclidean distance;
[0021] Based on the target compressed data corresponding to any one of the channels in the current sampling period, the feature information of the cluster center of the cluster to which the target compressed data corresponding to any one of the channels in the current sampling period belongs is updated.
[0022] According to an electroencephalogram signal processing method provided by the present invention, the Euclidean distance between the target compressed data corresponding to any channel in the current sampling period and the feature information of the cluster center of each cluster corresponding to any channel is calculated, and then based on each Euclidean distance, the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs is determined, including:
[0023] The specific calculation formula for calculating the Euclidean distance between the target compressed data corresponding to any channel in the current sampling period and the feature information of the cluster center of each cluster corresponding to any channel includes:
[0024]
[0025] in, Indicates the target compressed data corresponding to any channel in the current sampling period and the first j The Euclidean distance between the feature information of the cluster centers of clusters, j represents a positive integer; i Indicates less than n A non-negative integer; n Indicates the number of sampling points in the target EEG signal corresponding to any channel in the current sampling period, n Determined based on the sampling rate and the duration of the sampling period; m Indicates that the current sampling period is m Sampling period; Indicates the first j The characteristic information of the cluster center of the cluster i The value of the dimension; Indicates j The number of target compressed data in the clusters;
[0026] Will Recorded as after, ;
[0027] Compare and , and in comparison and When the equivalence is compared and ;
[0028] reserve and The smaller value corresponds to the cluster and ,and j After increasing by 1 Compare and keep the smaller one according to the comparison result repeatedly, and j After increasing by 1 The comparison steps are as follows j The maximum value is reached to achieve the clustering cluster with the smallest Euclidean distance between the characteristic information of the cluster center and the target compressed data corresponding to any channel in the current sampling period, and is determined as the clustering cluster to which the target compressed data corresponding to any channel in the current sampling period belongs.
[0029] According to an EEG signal processing method provided by the present invention, after clustering the target compressed data corresponding to any channel in the current sampling period, the method further includes:
[0030] When the preset time length is reached, the similarity or distance between the feature information of the cluster centers of any two clusters corresponding to any one of the channels is obtained;
[0031] If the similarity between the feature information of the cluster centers of any two clusters corresponding to any channel is higher than a similarity threshold or the distance is lower than a distance threshold, then the any two clusters corresponding to any channel are merged.
[0032] The present invention also provides an electroencephalogram signal processing device, comprising the following modules:
[0033] The signal acquisition module is used to collect the EEG signals in the cerebral cortex of the target subject at a predefined sampling rate using an implantable EEG signal acquisition device.
[0034] A signal detection module is used for obtaining, for any channel of the implantable EEG signal acquisition device, a target EEG signal corresponding to any channel in the current sampling period based on the EEG signal collected by the implantable EEG signal acquisition device through any channel in the current sampling period, when determining that the current sampling period has ended; based on the low-bit quantization technology, compressing the target EEG signal corresponding to any channel in the current sampling period, and obtaining target compressed data corresponding to any channel in the current sampling period, wherein the EEG signal includes an action potential signal.
[0035] The signal clustering module is used to cluster the target compressed data corresponding to any channel in the current sampling period, and determine the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs.
[0036] The data conversion module is used to obtain the discharge rate data of each virtual neuron corresponding to any channel in the current processing cycle based on the target compressed data included in each cluster cluster corresponding to any channel in the current processing cycle and the target compressed data included in each cluster cluster corresponding to any channel in the previous processing cycle when it is determined that the current processing cycle is over, and any processing cycle includes multiple sampling cycles.
[0037] The signal decoding module is used to calculate the intention information of the target subject in the current processing cycle based on the discharge rate data of each virtual neuron corresponding to each channel in the current processing cycle.
[0038] The present invention also provides a dedicated integrated circuit, which is equipped with the electroencephalogram signal processing device as described above.
[0039] The present invention also provides a brain-computer interface system, comprising: the above-mentioned dedicated integrated circuit and an implantable electroencephalogram signal acquisition device; the implantable electroencephalogram signal acquisition device is connected to the dedicated integrated circuit;
[0040] After being implanted in the brain of a target subject, the implantable EEG signal acquisition device is used to acquire EEG signals in the cerebral cortex of the target subject at a predefined sampling rate, and transmit the acquired EEG signals to the dedicated integrated circuit.
[0041] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, any of the above-mentioned methods for processing electroencephalogram signals is implemented.
[0042] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for processing electroencephalogram signals as described above is implemented.
[0043] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned methods for processing electroencephalogram signals.
[0044] The EEG signal processing method, device, circuit, system and equipment provided by the present invention compress the target EEG signal based on low-bit quantization technology, cluster the target compressed data after obtaining the target compressed data, obtain the discharge rate data of each virtual neuron according to the clustering result, and then calculate the intention information of the target subject based on the discharge rate data of each virtual neuron. It can fundamentally reduce the processing complexity, computing resource overhead and memory access bandwidth requirements of processing EEG signals with high spatiotemporal resolution and high sampling rate, reduce the data volume overhead and transmission bandwidth consumption of the above-mentioned EEG signal processing, thereby reducing the computational delay of processing the above-mentioned EEG signal, and can realize real-time, low-power decoding of the intention information of the EEG signal, reduce the data storage and transmission requirements of the EEG interface system, have good scalability by channel expansion, and have high flexibility for different EEG signals. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0046] Figure 1 It is a flow chart of the EEG signal processing method provided by the present invention.
[0047] Figure 2 This is one of the data flow diagrams for compressing and clustering target EEG signals in the EEG signal processing method provided by the present invention.
[0048] Figure 3 This is the second data flow diagram for compressing and clustering target EEG signals in the EEG signal processing method provided by the present invention.
[0049] Figure 4 This is the third data flow diagram for compressing and clustering target EEG signals in the EEG signal processing method provided by the present invention.
[0050] Figure 5 This is one of the schematic diagrams of the principle of clustering target compressed data in the EEG signal processing method provided by the present invention.
[0051] Figure 6 This is the second schematic diagram of the principle of clustering target compressed data in the EEG signal processing method provided by the present invention.
[0052] Figure 7 It is a structural schematic diagram of the electroencephalogram signal processing device provided by the present invention.
[0053] Figure 8 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION
[0054] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0055] In the description of the present application, the terms "first", "second", etc. are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are usually a class, and the number of objects is not limited. For example, the first object can be one or more. In addition, in the description of the present application, "and / or" represents at least one of the connected objects, and the character " / " generally represents that the front and back associated objects are in an "or" relationship.
[0056] It should be noted that a non-implantable brain-computer interface system refers to a brain-computer interface system that does not require electrodes or other devices to be implanted in the user's scalp or brain. A non-implantable brain-computer interface system can usually use external sensors to capture single-modal or multi-modal EEG signals, and then extract and decode the features of the above EEG signals to obtain the control instructions or assumed neural intentions corresponding to the above EEG signals.
[0057] An implantable brain-computer interface system is a brain-computer interface system that directly implants electrodes or other sensors into the user's brain tissue to obtain and interpret brain activity.
[0058] Because the EEG signals collected by the non-implantable brain-computer interface system have low temporal and spatial resolution, the required sampling rate is low, and the non-implantable brain-computer interface system is not sensitive to system size, power consumption, etc., the non-implantable brain-computer interface system is better in processing simple brain-computer interaction tasks such as external device control.
[0059] However, for more complex and sophisticated brain-computer interaction tasks such as capturing and interpreting motor intentions or language generation, the non-implantable brain-computer interface systems have difficulty handling more complex and sophisticated brain-computer interaction tasks due to the low spatiotemporal resolution of the EEG signals collected by them.
[0060] The implantable brain-computer interface system can collect EEG signals with high temporal and spatial resolution, and then process the above-mentioned more complex and sophisticated brain-computer interaction tasks based on the above-mentioned high temporal and spatial resolution EEG signals.
[0061] However, when the implantable brain-computer interface system processes the above-mentioned high spatiotemporal resolution EEG signals, when there are many channels and high sampling rates, it will generate a large amount of data overhead and transmission bandwidth consumption, which will easily introduce higher computing delays. In addition, the implantable brain-computer interface system is more sensitive to power consumption, sampling rate and real-time performance of EEG signals. Therefore, how to reduce the computing delay of the implantable brain-computer interface system when processing EEG signals while ensuring low power consumption, high sampling rate and high number of channels of the implantable brain-computer interface system is a technical problem that needs to be solved urgently in this field.
[0062] Combine the following Figure 1-Figure 4 The electroencephalogram signal processing method of the present invention is described.
[0063] Figure 1 is a flow chart of the method for processing EEG signals provided by the present invention, such as Figure 1 As shown, the method includes the following: Step 101, using an implantable EEG signal acquisition device to acquire EEG signals in the cerebral cortex of a target subject at a predefined sampling rate, the EEG signals including action potential signals.
[0064] It should be noted that the executor of the embodiment of the present invention is an electroencephalogram signal processing device. The electroencephalogram signal processing device can be configured in an application-specific integrated circuit (ASIC); the electroencephalogram signal processing device can also be configured in other electronic devices.
[0065] ASIC refers to an integrated circuit customized for a specific application. Unlike general-purpose microprocessors or memories, ASIC can be designed to perform specific functions or tasks, usually with customized hardware functions and performance optimization.
[0066] In an embodiment of the present invention, when it is determined that the current sampling cycle has ended, the step of acquiring each target EEG signal of the current sampling cycle based on the EEG signal of any channel collected by the implantable EEG signal acquisition device in the current sampling cycle can be performed by the signal acquisition module in the EEG signal processing device.
[0067] It should be noted that the EEG signal in the cerebral cortex of the target subject in the embodiment of the present invention is the processing object of the EEG signal processing method provided by the present invention. The target subject in the embodiment of the present invention can be determined based on actual needs, and the target subject is not specifically limited in the embodiment of the present invention.
[0068] The implantable EEG signal acquisition device in the embodiment of the present invention is a device that can be used to acquire EEG signals in the cerebral cortex of a subject after being implanted in the subject's brain. The implantable EEG signal acquisition device may include but is not limited to an implantable brain electrode array, an implantable EEG probe, etc.
[0069] Among them, the EEG signals collected by the above-mentioned implantable EEG signal acquisition device may include nerve impulse signals, local field potentials (LFP) signals and intracortical electroencephalography (ECoG) signals. The nerve impulse signals are also called action potentials (AP) or spike signals.
[0070] It is understandable that the above implantable EEG signal acquisition device has been implanted in the brain of the target subject. The implantable EEG signal acquisition device implanted in the brain of the target subject can collect EEG signals in the cerebral cortex of the target subject at each sampling point.
[0071] It should be noted that the implantable EEG signal acquisition device in the embodiment of the present invention can acquire EEG signals in the cerebral cortex of the target subject through a single channel or multiple channels. The EEG signals acquired by the implantable EEG signal acquisition device through any channel are time-series streaming data.
[0072] The sampling rate of the implantable EEG signal acquisition device in the embodiment of the present invention when acquiring EEG signals is predefined based on actual conditions and / or prior knowledge.
[0073] Step 102: for any channel of the implantable EEG signal acquisition device, when it is determined that the current sampling period ends, based on the EEG signal collected by the implantable EEG signal acquisition device through any channel during the current sampling period, obtain the target EEG signal corresponding to any channel of the current sampling period, and based on the low-bit quantization technology, compress the target EEG signal corresponding to any channel of the current sampling period to obtain the target compressed data corresponding to any channel of the current sampling period.
[0074] If it is determined that the first preset time has passed between the current moment and the start moment of the current sampling period, it can be determined that the current sampling period ends, and the end moment of the current sampling period can be determined as the start moment of the next sampling period.
[0075] It should be noted that the duration of a sampling period in the embodiment of the present invention is the first preset duration. The first preset duration may be determined based on prior knowledge and / or actual conditions. The specific value of the first preset duration is not specifically limited in the embodiment of the present invention.
[0076] It is understandable that, since the sampling rate of the implantable EEG signal acquisition device for sampling the EEG signal in the cerebral cortex of the target subject is predefined, and the duration of the sampling period is fixed, any target EEG signal corresponding to any sampling period includes EEG signals of the first preset number of sampling points, where the first preset number is the product of the sampling rate and the first preset duration.
[0077] In the embodiment of the present invention, p Identify any channel of the implantable EEG signal acquisition device, , P Indicates the number of channels of the implantable EEG signal acquisition device.
[0078] When it is determined that the current sampling period has ended, the implantable EEG signal acquisition device in the current sampling period can be p The EEG signals collected by the channels are preprocessed and the preprocessed EEG signals are determined as the first p The target EEG signal corresponding to each channel.
[0079] It should be noted that the data preprocessing in the embodiment of the present invention may include but is not limited to data preprocessing such as filtering, signal amplification and enhancement, and signal coupling.
[0080] Specifically, in the embodiments of the present invention, m Identifies the current sampling period. .
[0081] In the embodiment of the present invention, i Indicates the current sampling period. p Any sampling point in the target EEG signal corresponding to the channel, , n Indicates the current sampling period. p The total number of sampling points in the target EEG signal corresponding to the channel, that is n Indicates the first preset quantity.
[0082] It should be noted that the current sampling period is p The action potential signals of each sampling point in the target EEG signal corresponding to each channel are arranged in the order of the time sequence of collecting the action potential signals of each sampling point, that is, the current sampling period p The target EEG signal corresponding to the channel i The collection time of the action potential signal of the sampling point is earlier than the current sampling period. p The target EEG signal corresponding to the channel i +1 sampling point action potential signal acquisition time.
[0083] It can be understood that each sampling period p The target EEG signals corresponding to the channels are arranged in time sequence. m Sampling period p The end time of the target EEG signal corresponding to the channel is earlier than that of the m +1 sampling period p The starting time of the target EEG signal corresponding to each channel.
[0084] It should be noted that in the embodiment of the present invention, based on low-bit quantization technology, the target EEG signal corresponding to any channel in the current sampling period is compressed, and the step of obtaining the target compressed data corresponding to any channel in the current sampling period can be executed by the pulse recognition module in the EEG signal processing device.
[0085] Specifically, since the current sampling period is p The target EEG signal corresponding to each channel is obtained based on the EEG signal collected by the implanted EEG signal acquisition device. Therefore, the current sampling period is pThe target EEG signal corresponding to each channel may have the characteristics of high spatiotemporal resolution, high sampling rate and a large number of channels. If the above target EEG signal is processed based on the traditional EEG signal processing method, there is a defect of high calculation delay.
[0086] Therefore, in the embodiment of the present invention, the EEG signal collected by the implanted EEG signal has the characteristics of high temporal and spatial resolution, high sampling rate and large number of channels, and the low-bit quantization technology is used to quantize the current sampling period. p The target EEG signal corresponding to each channel is compressed and the current sampling period is p The target EEG signal corresponding to each channel is low-bit or even single-bit data, which can fundamentally reduce the processing p The processing complexity of the target EEG signal corresponding to each channel, the overhead of computing resources, and the memory bandwidth requirements, etc.
[0087] Among them, low-bit quantization technology is a data compression technology, the core idea of which is to minimize the number of bits required for data transmission or storage while maintaining sufficient data quality and feature distinguishability.
[0088] In the embodiment of the present invention, the bit width of the EEG signal of a single sampling point can be compressed, and low-bit quantization techniques such as compression of the duration of the EEG signal by waveform approximation or downsampling can be used to compress the bit width of the EEG signal of the current sampling period. p The target EEG signal corresponding to each channel is compressed to obtain the p The target compressed data corresponding to each channel.
[0089] As an optional embodiment, based on low-bit quantization technology, the target EEG signal corresponding to any channel in the current sampling period is compressed to obtain target compressed data corresponding to any channel in the current sampling period, including: determining any two adjacent sampling points in the target EEG signal corresponding to any channel in the current sampling period as a sampling point pair, and for each sampling point pair in the target EEG signal corresponding to any channel in the current sampling period, encoding the target EEG signal based on the voltage rise and fall relationship of the action potential signals of the two sampling points in each sampling point pair, and obtaining the encoded data corresponding to each sampling point pair.
[0090] It is understandable that the current sampling period p The first sampling point pair in the target EEG signal corresponding to the channel includes the first sampling point pair in the current sampling period. p The first sampling point and the second sampling point in the target EEG signal corresponding to the channel, the current sampling period p The second sampling point pair in the target EEG signal corresponding to the channel includes the current sampling period pThe second and third sampling points in the target EEG signal corresponding to the channel, and so on, the current sampling period p The target EEG signal corresponding to the channel i The sampling point pair includes the current sampling period p The first channel of the target EEG signal i sampling points and i +1 sampling point, the current sampling period p The target EEG signal corresponding to the channel n -1 sampling point pair includes the current sampling period p The first channel of the target EEG signal n -1 sampling point and n sampling point. p The total number of sampling point pairs in the target EEG signal corresponding to the channel is n -1 pc.
[0091] In the embodiment of the present invention, Indicates the first m The target EEG signal i The voltage of the action potential signal at each sampling point.
[0092] Based on the corresponding m The target EEG signal i The voltage of the action potential signal at each sampling point and i +1 sampling point action potential signal voltage , we can determine the first m The target EEG signal i The action potential signal of the sampling point and the i The voltage rise and fall relationship of the action potential signal of the +1 sampling point can be used to calculate the voltage rise and fall relationship of the action potential signal of any channel corresponding to the m The target EEG signal i The action potential signal of the sampling point and the i +1 sampling point action potential signal is encoded, and then the corresponding m Target EEG signal i The encoded data corresponding to the sampling point pairs.
[0093] As an optional embodiment, based on the voltage rise and fall relationship of the action potential signals of the two sampling points in each sampling point pair, the target EEG signal is encoded to obtain the encoded data corresponding to each sampling point pair, including: when the voltage of the action potential signal of the sampling point in the previous sampling point in each sampling point pair is less than the voltage of the action potential signal of the sampling point in the next sampling point in time, the encoded data corresponding to each sampling point pair is determined to be 1; when the voltage of the action potential signal of the sampling point in the previous sampling point in each sampling point pair is not less than the voltage of the action potential signal of the sampling point in the next sampling point in time, the encoded data corresponding to each sampling point pair is determined to be 0; or, when the voltage of the action potential signal of the sampling point in the previous sampling point in each sampling point pair is less than the voltage of the action potential signal of the sampling point in the next sampling point in time, the encoded data corresponding to each sampling point pair is determined to be 0; when the voltage of the action potential signal of the sampling point in the previous sampling point in each sampling point pair is not less than the voltage of the action potential signal of the sampling point in the next sampling point in time, the encoded data corresponding to each sampling point pair is determined to be 1.
[0094] Specifically, in the embodiments of the present invention, CS (compressed spike) represents the coded data obtained by compressing the target EEG signal, which can be used Indicates the first i The encoded data corresponding to the sampling point pairs.
[0095] Figure 2 This is one of the data flow diagrams for compressing and clustering target EEG signals in the EEG signal processing method provided by the present invention. Figure 3 This is the second data flow diagram for compressing and clustering target EEG signals in the EEG signal processing method provided by the present invention. Figure 4 This is the third data flow diagram of compressing and clustering the target EEG signal in the EEG signal processing method provided by the present invention. Figures 2 to 4 As shown in sequence, for the target EEG signal corresponding to any channel, the embodiment of the present invention can be used Indicates the coded data corresponding to the target EEG signal corresponding to any channel. In the target EEG signal corresponding to any channel, n In the case of sampling points, for n -1-dimensional vector and each dimension is a single-bit number.
[0096] In the embodiment of the present invention, Indicates the first m The target EEG signal i The encoded data corresponding to the sampling point pairs.
[0097] If any channel corresponds to mThe target EEG signal i The voltage of the action potential signal at each sampling point Lower than the corresponding m The target EEG signal i +1 sampling point action potential signal voltage , we can determine the first m The target EEG signal i sampling points and i The voltage rise and fall relationship of the action potential signal of +1 sampling point is that the voltage rises, and then the voltage corresponding to any channel can be m The target EEG signal i The coded data corresponding to the sampling point Determine to be 1; if any channel corresponds to m The target EEG signal i The voltage of the action potential signal at each sampling point Not less than the corresponding m The target EEG signal i +1 sampling point action potential signal voltage , we can determine the first m The target EEG signal i sampling points and i The voltage rise and fall relationship of the action potential signal at +1 sampling point is a voltage decrease, and thus the voltage corresponding to any channel can be m The first i The coded data corresponding to the sampling point Determined to be 0.
[0098] Or, if any channel corresponds to m The target EEG signal i The voltage of the action potential signal at each sampling point Lower than the corresponding m The target EEG signal i +1 sampling point action potential signal voltage , we can determine the first m The target EEG signal i sampling points and i The voltage rise and fall relationship of the action potential signal of +1 sampling point is that the voltage rises, and then the voltage corresponding to any channel can be m The target EEG signal i The coded data corresponding to the sampling points Determine to be 0; if any channel corresponds to m The target EEG signal iThe voltage of the action potential signal at each sampling point Not less than the corresponding m The target EEG signal i +1 sampling point action potential signal voltage , we can determine the first m The target EEG signal i sampling points and i The voltage rise and fall relationship of the action potential signal at +1 sampling point is a voltage decrease, and thus the voltage corresponding to any channel can be m The first i The coded data corresponding to the sampling point Determined to be 1.
[0099] Any channel corresponds to m The target EEG signal i The coded data corresponding to the sampling point It can be expressed by the following formula:
[0100] .
[0101] The encoding data corresponding to each sampling point pair in the target EEG signal corresponding to any channel in the current sampling period are arranged in time sequence to obtain the target compressed data corresponding to any channel in the current sampling period.
[0102] Specifically, get the current sampling period p After each sampling point pair of the target EEG signal corresponding to the channel is coded, the current sampling period is arranged in chronological order. p The corresponding coded data of each sampling point pair in the target EEG signal corresponding to each channel can be obtained. p The target compressed data corresponding to each channel.
[0103] In the embodiment of the present invention, the target EEG signal is encoded based on the voltage rise and fall relationship of the action potential signals of two sampling points in each sampling point pair in the target EEG signal to obtain target compressed data. The multi-bit action signal of any sampling point in the target EEG signal can be quantized into a single-bit action signal, thereby compressing the target EEG signal into target compressed data arranged in a single-bit data sequence. This can fundamentally reduce the processing complexity, computing resource overhead and memory access bandwidth requirements for processing EEG signals with high spatiotemporal resolution and high sampling rate, and can reduce the data volume overhead and transmission bandwidth consumption of the above-mentioned EEG signal processing, thereby reducing the computational delay in processing the above-mentioned EEG signal.
[0104] Step 103: cluster the target compressed data corresponding to any channel in the current sampling period, and determine the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs.
[0105] It should be noted that, in the embodiment of the present invention, the step of clustering the target compressed data corresponding to any channel in the current sampling period and determining the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs can be executed by the pulse clustering module in the EEG signal processing device.
[0106] Specifically, get the current sampling period p After the target compression data corresponding to each channel is obtained, the current sampling period can be calculated by numerical calculation. p The target compressed data corresponding to each channel is clustered to determine the current sampling period. p The cluster to which the target compressed data corresponding to each channel belongs.
[0107] It should be noted that in the embodiment of the present invention, the clusters correspond to the virtual neurons one by one, that is, the number of cluster centers is the same as the number of virtual neurons.
[0108] It should be noted that the number of virtual neurons corresponding to any channel in the embodiment of the present invention (ie, the number of clusters) is predefined according to actual conditions and / or prior knowledge. The number of virtual neurons in the embodiment of the present invention is not specifically limited.
[0109] As an optional embodiment, the target compressed data corresponding to any channel of the current sampling period is clustered to determine the cluster cluster to which the target compressed data corresponding to any channel of the current sampling period belongs, including: calculating the Euclidean distance between the target compressed data corresponding to any channel of the current sampling period and the feature information of the cluster center of each cluster cluster corresponding to any channel, and then determining the cluster cluster to which the target compressed data corresponding to any channel of the current sampling period belongs based on each Euclidean distance.
[0110] It should be noted that the EEG signals collected by implantable EEG signal processing not only have the characteristics of high spatiotemporal resolution, high sampling rate and a large number of channels, but also have the characteristics of low-dimensional manifold embedding space and higher sensitivity to the number of channels than to signal quality.
[0111] Therefore, in the embodiment of the present invention, when the current sampling period is p When clustering the target compressed data corresponding to each channel, the clustering of the first channel in the current sampling period is completed according to the evaluation methods such as Euclidean distance measurement or waveform correlation. p Clustering of target compressed data corresponding to channels.
[0112] It should be noted that the number of clusters in the embodiment of the present invention is the second preset number, which can be k Indicates that the second preset quantity k It can be predefined based on prior knowledge and / or actual situation, for example, the second preset number k It can be determined based on the number of clusters that any channel can accommodate. k There is no specific limitation on the specific value of .
[0113] In the embodiment of the present invention, j Indicates p Any cluster corresponding to the channels.
[0114] Figure 5 This is one of the schematic diagrams of the principle of clustering target compressed data in the EEG signal processing method provided by the present invention. Figure 6 This is the second schematic diagram of the principle of clustering target compressed data in the EEG signal processing method provided by the present invention. Figures 2 to 6 For the current sampling period p The process of clustering the target compressed data corresponding to each channel is described below.
[0115] It should be noted that Figures 2 to 6 Indicates the p The channel corresponds to j The characteristic data of clusters include and , ACS It is the abbreviation of accumulated center of spikes. Indicates p The channel corresponds to j The characteristic information of the cluster centers of clusters, for n dimensional vector, Indicates p The channel corresponds to j The number of target compressed data in the clusters, j The index representing the cluster. express Middle i The value of the dimension.
[0116] In the current sampling cycle p When clustering the target compressed data corresponding to the channels, if m ≤ k , then based on the current sampling period ( m sampling period) p The target compressed data corresponding to the channel is obtained.p The channel corresponds to m The characteristic information of the cluster center of each cluster.
[0117] In the current sampling cycle p When clustering the target compressed data corresponding to the channels, if m > k , then the current sampling period can be calculated by numerical calculation. p The target compressed data corresponding to the channel is p The Euclidean distance between the feature information of the cluster center of each cluster corresponding to the channels.
[0118] Current sampling period p The target compressed data corresponding to the channel is p The Euclidean distance between the feature information of the cluster center of each cluster corresponding to the channel , can be calculated by the following formula:
[0119]
[0120] Among them, Recorded as , the above formula can be transformed into , and due to is a single-bit data number, so it is the same as The multiplication operation can be replaced by a multiplexer.
[0121] In the comparison target compressed data i When calculating the Euclidean distance between the encoded data corresponding to the sampling point pairs and the cluster centers of different clusters, the traditional comparison formula can be used: ,Right now To avoid division operations, the formula can be equivalently transformed into .
[0122] Compare and , and in comparison and When the equivalence is compared and .
[0123] reserve and The smaller one and the corresponding cluster and ,and j After increasing by 1 Compare and keep the smaller one according to the comparison result repeatedly, and j After increasing by 1 The comparison steps are as follows j The maximum value is reached to realize the cluster with the smallest Euclidean distance between the characteristic information of the cluster center and the target compressed data corresponding to any channel of the current sampling period, and determine it as the cluster to which the target compressed data corresponding to any channel of the current sampling period belongs.
[0124] Therefore, when designing the comparator circuit, only the smaller group of and The feature is input to the next comparator.
[0125] For ease of description, and The comparator involved in the comparison j Output Recorded as ,Will and The comparator involved in the comparison j Output Recorded as Until the last level of comparator is obtained k-1 Output and .
[0126] Based on the target compressed data corresponding to any channel in the current sampling period, the feature information of the cluster center of the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs is updated.
[0127] Specifically, the characteristic information of the cluster center is combined with the p The cluster with the smallest Euclidean distance between the target compressed data corresponding to the channels is determined as the cluster in the current sampling period. p After the target compressed data corresponding to the channel belongs to the cluster, the current sampling period can be updated. p The characteristics of the cluster to which the target compressed data corresponding to each channel belongs, and the characteristics of the updated cluster can be expressed by the following formula:
[0128]
[0129] Since the number of clusters accommodated by a single channel is usually small, that is, k It is generally in the single digit, so the comparison circuit will not be too large in size.
[0130] As an optional embodiment, after clustering the target compressed data corresponding to any channel in the current sampling period, the method further includes: obtaining the similarity or distance between the feature information of the cluster centers of any two clusters corresponding to any channel when a preset time length is reached;
[0131] If the similarity between the feature information of the cluster centers of any two clusters corresponding to any channel is higher than a similarity threshold or the distance is lower than a distance threshold, then the any two clusters corresponding to any channel are merged.
[0132] Specifically, the feature information of the cluster center of the merged cluster is updated as follows:
[0133]
[0134] Step 104, when it is determined that the current processing cycle is over, based on the target compressed data included in each cluster corresponding to any channel of the current processing cycle and the target compressed data included in each cluster corresponding to any channel of the previous processing cycle, the discharge rate data of each virtual neuron corresponding to any channel of the current processing cycle is obtained, and any processing cycle includes multiple sampling cycles.
[0135] It should be noted that, in the embodiment of the present invention, when it is determined that the current processing cycle has ended, the step of obtaining the discharge rate data of each virtual neuron corresponding to any channel of the current processing cycle based on the target compressed data included in each clustering cluster corresponding to any channel of the current processing cycle and the target compressed data included in each clustering cluster corresponding to any channel of the previous processing cycle can be performed by the data conversion module in the EEG signal processing device.
[0136] Specifically, if it is determined that the third preset time has passed between the current moment and the start moment of the current processing cycle, it can be determined that the current processing cycle ends, and the end moment of the current processing cycle can be determined as the start moment of the next processing cycle.
[0137] It should be noted that the duration of a processing cycle in the embodiment of the present invention is the third preset duration. The third preset duration may be determined based on prior knowledge and / or actual conditions. The specific value of the third preset duration is not specifically limited in the embodiment of the present invention.
[0138] It should be noted that the third preset duration in the embodiment of the present invention may include multiple first preset durations.
[0139] In the case of determining that the current processing cycle is over, the m ) p Each cluster corresponding to each channel includes the target compressed data and the last processing cycle (m -1) p Each cluster corresponding to each channel includes the target compressed data, and determines the p The number of target compressed data added in each cluster corresponding to the channel can be based on the number of p The number of target compressed data added in each cluster corresponding to the channel is calculated to obtain the p The discharge rate data of each virtual neuron corresponding to each channel.
[0140] The embodiment of the present invention determines the cluster to which the target compressed data corresponding to any channel of the current sampling period belongs by the Euclidean distance between the target compressed data corresponding to any channel of the current sampling period and the feature information of the cluster center of each cluster corresponding to any channel, and then calculates the discharge rate data of each virtual neuron corresponding to any channel based on the target compressed data included in each cluster corresponding to any channel of the current processing period and the target compressed data included in each cluster corresponding to any channel of the previous processing period. The neural signal characteristics of the virtual neurons can be extracted only by linear operations such as addition, subtraction and multiplication, and the clustering of EEG signals can be achieved by comparison without division operations. The data volume overhead and transmission bandwidth consumption of EEG signal processing with high spatiotemporal resolution and high sampling rate can be further reduced, thereby further reducing the computational delay of processing the above EEG signals.
[0141] Step 105: Based on the discharge rate data of each virtual neuron corresponding to each channel in the current processing cycle, the intention information of the target subject in the current processing cycle is calculated.
[0142] It should be noted that the step of calculating the intention information of the target subject in the current processing cycle based on the discharge rate data of each virtual neuron corresponding to each channel in the current processing cycle in the embodiment of the present invention can be executed by the signal decoding module in the EEG signal processing device.
[0143] Specifically, in an embodiment of the present invention, based on the discharge rate data of each corresponding virtual neuron in the current processing cycle, linear algorithm circuits such as support vector machines, Kalman filters, linear classifiers, and Bayesian estimation algorithms can be used to calculate the intention information of the target subject in the current processing cycle.
[0144] The embodiment of the present invention compresses the target EEG signal based on low-bit quantization technology, and after obtaining the target compressed data, clusters the target compressed data, obtains the discharge rate data of each virtual neuron according to the clustering result, and then calculates the intention information of the target subject based on the discharge rate data of each virtual neuron. This can fundamentally reduce the processing complexity, computing resource overhead and memory access bandwidth requirements of processing EEG signals with high spatiotemporal resolution and high sampling rate, and can reduce the data volume overhead and transmission bandwidth consumption of the above-mentioned EEG signal processing, thereby reducing the computational delay of processing the above-mentioned EEG signals, and can realize real-time, low-power decoding of the intention information of EEG signals, reduce the data storage and transmission requirements of the EEG interface system, have good scalability by channel expansion, and have high flexibility for different EEG signals.
[0145] As an optional embodiment, after processing the target EEG signal corresponding to any channel in the current processing cycle, the method further includes: obtaining the similarity between the central features of each cluster corresponding to any channel in the current processing cycle and the central features of each cluster corresponding to the channel in the previous processing cycle.
[0146] When the fluctuation value of the similarity obtained within the preset time length is less than the fluctuation threshold, the key parameters in the step of clustering the target compressed data corresponding to any channel in the current sampling period are updated.
[0147] It should be noted that the key parameters in the embodiment of the present invention include but are not limited to: a first preset time length and a cluster center distance threshold used when merging clusters.
[0148] It should be noted that, since ASIC has a rigid structure and lacks flexibility, in the embodiment of the present invention, the key parameters that affect the algorithm operation results are set in the form of adaptive update thresholds.
[0149] Specifically, in the embodiment of the present invention, the conventional similarity calculation method can be used to calculate the current sampling period. p The target EEG signal corresponding to the channel is the same as the last sampling period. p The similarity of the target EEG signals corresponding to the channels.
[0150] It should be noted that the key parameters in the step of compressing the target EEG signal may include a detection threshold.
[0151] When the fluctuation value of the similarity obtained within the preset time period is less than the fluctuation threshold, the first preset time period and the cluster center distance threshold used when merging clusters may be updated.
[0152] It should be noted that when detecting target EEG signals such as spikes, the signal energy operator at each sampling moment is generally calculated based on the signal near the sampling point, such as the nonlinear energy operator NEO. The detection threshold is compared with the operator result. If the threshold is exceeded, it is considered that a spike signal appears at that moment.
[0153] It should be noted that when updating the key parameters in the clustering step of the target compressed data, the characteristic information of the cluster center of each cluster can be included. For example, after a long period of time, the ACS and num characteristic values of each cluster will be large. At this time, both are shifted left by the same number of bits to keep the division result of the two, that is, the cluster center, basically unchanged.
[0154] When the fluctuation value of the similarity obtained within the preset time is less than the fluctuation threshold, the EEG signal with the desired cluster center can be continuously input, that is, the EEG signal corresponding to the desired cluster center of the first cluster is first input, and maintained for a certain period of time. At this time, the cluster center of the EEG signal will be averaged according to the input EEG signal and the location information of the cluster center will be updated. The remaining distance clusters are processed in the same way. When the EEG signal stops being input or the type is changed, the consistency of the EEG signal will decrease rapidly, and the key parameter update phase ends.
[0155] The following is a description of the electroencephalogram signal processing device provided by the present invention. The electroencephalogram signal processing device described below and the electroencephalogram signal processing method described above can be referenced to each other. Figure 7 Schematic diagram of the structure of the electroencephalogram signal processing device provided by the present invention. Figure 7 As shown, the device includes: a signal acquisition module 701, a signal detection module 702, a signal clustering module 703, a data conversion module 704 and a signal decoding module 705.
[0156] The signal acquisition module 701 is used to collect the EEG signals in the cerebral cortex of the target subject at a predefined sampling rate using an implantable EEG signal acquisition device, wherein the EEG signals include action potential signals.
[0157] The signal detection module 702 is used for obtaining, for any channel of the implantable EEG signal acquisition device, a target EEG signal corresponding to any channel of the current sampling period based on the EEG signal collected by the implantable EEG signal acquisition device through any channel during the current sampling period, when determining that the current sampling period has ended, and compressing the target EEG signal corresponding to any channel of the current sampling period based on the low-bit quantization technology to obtain target compressed data corresponding to any channel of the current sampling period.
[0158] The signal clustering module 703 is used to cluster the target compressed data corresponding to any channel in the current sampling period, and determine the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs.
[0159] The data conversion module 704 is used to obtain the discharge rate data of each virtual neuron corresponding to any channel of the current processing cycle based on the target compressed data included in each cluster cluster corresponding to any channel of the current processing cycle and the target compressed data included in each cluster cluster corresponding to any channel of the previous processing cycle when it is determined that the current processing cycle is over. Any processing cycle includes multiple sampling cycles.
[0160] The signal decoding module 705 is used to calculate the intention information of the target subject in the current processing cycle based on the discharge rate data of each virtual neuron corresponding to each channel in the current processing cycle.
[0161] Specifically, the signal acquisition module 701 , the signal detection module 702 , the signal clustering module 703 , the data conversion module 704 and the signal decoding module 705 are electrically connected.
[0162] The above-mentioned EEG signal processing device can be configured in a dedicated integrated circuit; the above-mentioned EEG signal processing device can also be configured in other electronic devices.
[0163] The EEG signal processing device in the embodiment of the present invention compresses the target EEG signal based on low-bit quantization technology, and after obtaining the target compressed data, clusters the target compressed data, obtains the discharge rate data of each virtual neuron according to the clustering result, and then calculates the intention information of the target subject based on the discharge rate data of each virtual neuron. It can fundamentally reduce the processing complexity, computing resource overhead and memory access bandwidth requirements of processing EEG signals with high spatiotemporal resolution and high sampling rate, and can reduce the data volume overhead and transmission bandwidth consumption of the above-mentioned EEG signal processing, thereby reducing the computational delay of processing the above-mentioned EEG signal, and can realize real-time, low-power decoding of the intention information of the EEG signal, reduce the data storage and transmission requirements of the EEG interface system, and have good scalability by channel expansion and high flexibility for different EEG signals.
[0164] Based on the contents of the above embodiments, a dedicated integrated circuit is provided, wherein the dedicated integrated circuit is equipped with the electroencephalogram signal processing device as described above.
[0165] Based on the contents of the above embodiments, a brain-computer interface system includes: the dedicated integrated circuit and the implantable EEG signal acquisition device as described above; the implantable EEG signal acquisition device is electrically connected to the dedicated integrated circuit.
[0166] After being implanted in the brain of a target subject, the implantable EEG signal acquisition device is used to acquire EEG signals in the cerebral cortex of the target subject at a predefined sampling rate, and transmit the acquired EEG signals to the dedicated integrated circuit.
[0167] It should be noted that the implantable EEG signal acquisition device in the embodiment of the present invention may be multi-channel.
[0168] The EEG interface system in the embodiment of the present invention is directly connected to the brain nervous system of the subject through an implanted EEG signal acquisition device, and can directly interact with one or more independent neurons and directly contact the neurons in the brain, so that EEG signals with greater intensity and less noise can be acquired, and the acquired EEG signals have higher quality. After the dedicated integrated circuit obtains the target EEG signal based on the EEG signal acquired by the implanted EEG signal acquisition device, it compresses the target EEG signal based on the low-bit quantization technology, and after obtaining the target compressed data, the target compressed data is clustered, and the data of each virtual neuron is obtained according to the clustering result. The discharge rate data of the virtual neuron is used to calculate the intention information of the target subject, and then based on the discharge rate data of each virtual neuron, the intention information of the target subject is calculated, which can fundamentally reduce the processing complexity, computing resource overhead and memory bandwidth requirements of processing EEG signals with high spatiotemporal resolution and high sampling rate, and can reduce the data volume overhead and transmission bandwidth consumption of the above-mentioned EEG signal processing, thereby reducing the computational delay in processing the above-mentioned EEG signals, and can realize real-time, low-power decoding of the intention information of EEG signals, reduce the data storage and transmission requirements of the EEG interface system, have good scalability by channel expansion, and have high flexibility for different EEG signals.
[0169] Figure 8 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 8As shown, the electronic device may include: a processor 810, a communications interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communications interface 820 and the memory 830 communicate with each other through the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to execute the EEG signal processing method, which includes: using an implantable EEG signal acquisition device to acquire EEG signals in the cerebral cortex of the target subject at a predefined sampling rate, and for any channel of the implantable EEG signal acquisition device, when it is determined that the current sampling period has ended, based on the EEG signals acquired by the implantable EEG signal acquisition device through any channel in the current sampling period, obtaining the target EEG signal corresponding to any channel in the current sampling period, the EEG signal including the action potential signal; based on the low-bit quantization technology, compressing the target EEG signal corresponding to any channel in the current sampling period, and obtaining the target EEG signal corresponding to any channel in the current sampling period. The target compressed data corresponding to any channel of the current sampling period is clustered to determine the cluster to which the target compressed data corresponding to any channel of the current sampling period belongs; when it is determined that the current processing period is over, based on the target compressed data included in each cluster corresponding to any channel of the current processing period and the target compressed data included in each cluster corresponding to any channel of the previous processing period, the discharge rate data of each virtual neuron corresponding to any channel of the current processing period is obtained, and any processing period includes multiple sampling periods; based on the discharge rate data of each virtual neuron corresponding to each channel of the current processing period, the intention information of the target subject in the current processing period is calculated.
[0170] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.
[0171] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the EEG signal processing method provided by the above methods, which includes: using an implantable EEG signal acquisition device to acquire EEG signals in the cerebral cortex of a target subject at a predefined sampling rate. For any channel of the implantable EEG signal acquisition device, when it is determined that the current sampling period has ended, based on the EEG signals acquired by the implantable EEG signal acquisition device through any channel during the current sampling period, the target EEG signal corresponding to any channel of the current sampling period is acquired, and the EEG signal includes an action potential signal; based on a low-bit quantization technology, the current sampling period is quantized. The target EEG signal corresponding to any channel of the period is compressed to obtain the target compressed data corresponding to any channel of the current sampling period; the target compressed data corresponding to any channel of the current sampling period is clustered to determine the cluster to which the target compressed data corresponding to any channel of the current sampling period belongs; when it is determined that the current processing period is over, based on the target compressed data included in each cluster corresponding to any channel of the current processing period and the target compressed data included in each cluster corresponding to any channel of the previous processing period, the discharge rate data of each virtual neuron corresponding to any channel of the current processing period is obtained, and any processing period includes multiple sampling periods; based on the discharge rate data of each virtual neuron corresponding to each channel of the current processing period, the intention information of the target subject in the current processing period is calculated.
[0172] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to perform the EEG signal processing method provided by the above methods, the method comprising:
[0173] An implantable electroencephalogram signal acquisition device is used to acquire electroencephalogram signals in the cerebral cortex of a target subject at a predefined sampling rate. For any channel of the implantable electroencephalogram signal acquisition device, when it is determined that the current sampling period ends, based on the electroencephalogram signals acquired by the implantable electroencephalogram signal acquisition device through any channel in the current sampling period, a target electroencephalogram signal corresponding to any channel in the current sampling period is acquired, and the electroencephalogram signal includes an action potential signal; based on a low-bit quantization technology, the target electroencephalogram signal corresponding to any channel in the current sampling period is compressed to acquire target compressed data corresponding to any channel in the current sampling period; the target compressed data corresponding to any channel in the current sampling period is clustered to determine the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs; when it is determined that the current processing period ends, based on the target compressed data included in each cluster corresponding to any channel in the current processing period and the target compressed data included in each cluster corresponding to any channel in the previous processing period, the discharge rate data of each virtual neuron corresponding to any channel in the current processing period is acquired, and any processing period includes multiple sampling periods; based on the discharge rate data of each virtual neuron corresponding to each channel in the current processing period, the intention information of the target subject in the current processing period is calculated.
[0174] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.
[0175] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.
[0176] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for processing electroencephalogram signals, It is characterized in that include: Using an implantable EEG signal acquisition device to acquire EEG signals in the cerebral cortex of a target subject at a predefined sampling rate, the EEG signals including action potential signals; For any channel of the implantable EEG signal acquisition device, when it is determined that the current sampling period ends, based on the EEG signal acquired by the implantable EEG signal acquisition device through the any channel in the current sampling period, a target EEG signal corresponding to the any channel in the current sampling period is acquired, and based on the low-bit quantization technology, the target EEG signal corresponding to the any channel in the current sampling period is compressed to acquire target compressed data corresponding to the any channel in the current sampling period; Clustering the target compressed data corresponding to any channel in the current sampling period, and determining the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs; When it is determined that the current processing cycle is over, based on the target compressed data included in each cluster corresponding to any channel in the current processing cycle and the target compressed data included in each cluster corresponding to any channel in the previous processing cycle, the discharge rate data of each virtual neuron corresponding to any channel in the current processing cycle is obtained, and any processing cycle includes multiple sampling cycles; Based on the discharge rate data of each virtual neuron corresponding to each channel in the current processing cycle, the intention information of the target subject in the current processing cycle is calculated.
2. The method for processing electroencephalogram signals according to claim 1, It is characterized in that The method of compressing the target EEG signal corresponding to any channel in the current sampling period based on the low-bit quantization technology to obtain the target compressed data corresponding to any channel in the current sampling period includes: Determine any two adjacent sampling points in the target EEG signal corresponding to any channel in the current sampling period as a sampling point pair, and for each sampling point pair in the target EEG signal corresponding to any channel in the current sampling period, encode the target EEG signal based on the voltage rise and fall relationship of the action potential signals of the two sampling points in each sampling point pair, and obtain the encoded data corresponding to each sampling point pair; The encoding data corresponding to each sampling point pair in the target EEG signal corresponding to any channel in the current sampling period are arranged in time sequence to obtain the target compressed data corresponding to any channel in the current sampling period.
3. The method for processing electroencephalogram signals according to claim 2, It is characterized in that The step of encoding the target EEG signal based on the voltage rise and fall relationship of the action potential signals of the two sampling points in each sampling point pair to obtain the encoding data corresponding to each sampling point pair includes: In the case where the voltage of the action potential signal of the sampling point preceding the timing in each sampling point pair is less than the voltage of the action potential signal of the sampling point following the timing, the coding data corresponding to each sampling point pair is determined to be 1; in the case where the voltage of the action potential signal of the sampling point preceding the timing in each sampling point pair is not less than the voltage of the action potential signal of the sampling point following the timing, the coding data corresponding to each sampling point pair is determined to be 0, or, When the voltage of the action potential signal of the sampling point preceding the timing in each sampling point pair is less than the voltage of the action potential signal of the sampling point following the timing, the encoding data corresponding to each sampling point pair is determined to be 0; when the voltage of the action potential signal of the sampling point preceding the timing in each sampling point pair is not less than the voltage of the action potential signal of the sampling point following the timing, the encoding data corresponding to each sampling point pair is determined to be 1.
4. The method for processing electroencephalogram signals according to claim 3, It is characterized in that The clustering of the target compressed data corresponding to any channel in the current sampling period to determine the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs includes: Calculating the Euclidean distance between the target compressed data corresponding to any channel in the current sampling period and the feature information of the cluster center of each cluster corresponding to any channel, and then determining the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs based on each Euclidean distance; Based on the target compressed data corresponding to any one of the channels in the current sampling period, the feature information of the cluster center of the cluster to which the target compressed data corresponding to any one of the channels in the current sampling period belongs is updated.
5. The method for processing electroencephalogram signals according to claim 4, It is characterized in that The calculating of the Euclidean distance between the target compressed data corresponding to any channel in the current sampling period and the feature information of the cluster center of each cluster corresponding to any channel, and then determining the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs based on each Euclidean distance, includes: The specific calculation formula for calculating the Euclidean distance between the target compressed data corresponding to any channel in the current sampling period and the feature information of the cluster center of each cluster corresponding to any channel includes: ; in, Indicates the first m The coded data corresponding to the target EEG signal is the same as the coded data corresponding to the first j The Euclidean distance between the feature information of the cluster centers of clusters, j represents a positive integer not exceeding the number of clusters of any channel cluster; i Indicates less than n A non-negative integer; n Indicates the number of sampling points in the target EEG signal corresponding to any channel in the current sampling period, n Determined based on the sampling rate and the duration of the sampling period; Indicates the first m The encoding data corresponding to the i-th sampling point pair in the target EEG signal; m Indicates that the current sampling period is m Sampling period; Indicates the first j The characteristic information of the cluster center of the cluster i The value of the dimension; Indicates the first m Clustered to the first j The number of target compressed data in the clusters; Will Recorded as after, ; Any of the channels are compared in turn The Euclidean distance between the cluster center feature information of two adjacent clusters, that is, and , j and j +1 are all positive integers not exceeding the number of channel clusters, and are compared and When the equivalence is compared and ; reserve and The smaller value corresponds to the cluster and ,and j After increasing by 1 Compare and repeat the and The smaller value corresponds to the cluster and ,and j After increasing by 1 The comparison steps are as follows j The maximum value is reached to achieve the clustering cluster with the smallest Euclidean distance between the characteristic information of the cluster center and the target compressed data corresponding to any channel in the current sampling period, and is determined as the clustering cluster to which the target compressed data corresponding to any channel in the current sampling period belongs.
6. The method for processing electroencephalogram signals according to claim 5, It is characterized in that After clustering the target compressed data corresponding to any channel in the current sampling period, the method further includes: When the preset time length is reached, the similarity or distance between the feature information of the cluster centers of any two clusters corresponding to any one of the channels is obtained; If the similarity between the feature information of the cluster centers of any two clusters corresponding to any channel is higher than a similarity threshold or the distance is lower than a distance threshold, the any two clusters corresponding to any channel are merged.
7. An electroencephalogram signal processing device, It is characterized in that include: A signal acquisition module, used to acquire EEG signals in the cerebral cortex of a target subject at a predefined sampling rate using an implantable EEG signal acquisition device, wherein the EEG signals include action potential signals; A signal detection module, for any channel of the implantable EEG signal acquisition device, when determining that the current sampling period ends, based on the EEG signal acquired by the implantable EEG signal acquisition device through the any channel in the current sampling period, obtaining a target EEG signal corresponding to the any channel in the current sampling period, and compressing the target EEG signal corresponding to the any channel in the current sampling period based on a low-bit quantization technique, to obtain target compressed data corresponding to the any channel in the current sampling period; A signal clustering module, used for clustering the target compressed data corresponding to any channel in the current sampling period, and determining the cluster to which the target compressed data corresponding to any channel in the current sampling period belongs; A data conversion module, for obtaining the discharge rate data of each virtual neuron corresponding to any channel in the current processing cycle based on the target compressed data included in each cluster corresponding to any channel in the current processing cycle and the target compressed data included in each cluster corresponding to any channel in the previous processing cycle when determining that the current processing cycle is over, wherein any processing cycle includes multiple sampling cycles; The signal decoding module is used to calculate the intention information of the target subject in the current processing cycle based on the discharge rate data of each virtual neuron corresponding to each channel in the current processing cycle.
8. An application specific integrated circuit, It is characterized in that The ASIC is equipped with the electroencephalogram signal processing device as claimed in claim 7.
9. A brain-computer interface system, It is characterized in that include: The application specific integrated circuit and implantable EEG signal acquisition device as claimed in claim 8; The implantable EEG signal acquisition device is connected to the ASIC; After being implanted in the brain of a target subject, the implantable EEG signal acquisition device is used to acquire EEG signals in the cerebral cortex of the target subject at a predefined sampling rate, and transmit the acquired EEG signals to the dedicated integrated circuit.
10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, It is characterized in that When the processor executes the program, the electroencephalogram signal processing method according to any one of claims 1 to 6 is implemented.
Citation Information
Patent Citations
Electroencephalogram signal decoding method and device, electronic equipment and storage medium
CN116796223A
Implantable brain-computer interface system and information processing method thereof
CN116931726A
