CN117311516B - Motor imagery electroencephalogram channel selection method and system - Google Patents
Motor imagery electroencephalogram channel selection method and system Download PDFInfo
- Publication number
- CN117311516B CN117311516B CN202311599176.9A CN202311599176A CN117311516B CN 117311516 B CN117311516 B CN 117311516B CN 202311599176 A CN202311599176 A CN 202311599176A CN 117311516 B CN117311516 B CN 117311516B
- Authority
- CN
- China
- Prior art keywords
- neurons
- electroencephalogram
- neuron
- input
- cluster
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000010187 selection method Methods 0.000 title claims abstract description 16
- 210000002569 neuron Anatomy 0.000 claims abstract description 214
- 210000002364 input neuron Anatomy 0.000 claims abstract description 72
- 238000012549 training Methods 0.000 claims abstract description 58
- 230000003993 interaction Effects 0.000 claims abstract description 36
- 238000004891 communication Methods 0.000 claims abstract description 32
- 238000000034 method Methods 0.000 claims abstract description 32
- 230000008569 process Effects 0.000 claims abstract description 20
- 238000013507 mapping Methods 0.000 claims abstract description 17
- 210000004556 brain Anatomy 0.000 claims description 27
- 238000001208 nuclear magnetic resonance pulse sequence Methods 0.000 claims description 14
- 238000007781 pre-processing Methods 0.000 claims description 10
- 238000013528 artificial neural network Methods 0.000 claims description 8
- 230000036962 time dependent Effects 0.000 claims description 5
- 238000010586 diagram Methods 0.000 description 8
- 230000015654 memory Effects 0.000 description 6
- 230000004913 activation Effects 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 239000007788 liquid Substances 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002964 excitative effect Effects 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000001771 impaired effect Effects 0.000 description 1
- 230000002401 inhibitory effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008433 psychological processes and functions Effects 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/369—Electroencephalography [EEG]
- A61B5/372—Analysis of electroencephalograms
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
- G06F3/015—Input arrangements based on nervous system activity detection, e.g. brain waves [EEG] detection, electromyograms [EMG] detection, electrodermal response detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/09—Supervised learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2203/00—Indexing scheme relating to G06F3/00 - G06F3/048
- G06F2203/01—Indexing scheme relating to G06F3/01
- G06F2203/011—Emotion or mood input determined on the basis of sensed human body parameters such as pulse, heart rate or beat, temperature of skin, facial expressions, iris, voice pitch, brain activity patterns
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- Psychiatry (AREA)
- Animal Behavior & Ethology (AREA)
- Veterinary Medicine (AREA)
- Pathology (AREA)
- Heart & Thoracic Surgery (AREA)
- Public Health (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Surgery (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Medical Informatics (AREA)
- Signal Processing (AREA)
- Psychology (AREA)
- Human Computer Interaction (AREA)
- Neurosurgery (AREA)
- Neurology (AREA)
- Dermatology (AREA)
- Physiology (AREA)
- Fuzzy Systems (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention provides a motor imagery electroencephalogram channel selection method and a motor imagery electroencephalogram channel selection system, comprising the following steps: mapping motor imagery electroencephalogram data having a number of electroencephalogram channels to a NeuCube model that builds an SNN cube comprising input neurons and other neurons mapped by different electroencephalogram channels; analyzing the input neurons by adopting a clustering method, and grouping the neurons according to pulse communication of the neurons in the training process to obtain a neuron cluster taking the input neurons as the center; comparing the number of neurons in each neuron cluster with the total number of neurons in the model to obtain the proportion index of each input neuron; determining an average interaction between input neurons based on impulse communications of neurons of different clusters of neurons during training; and deleting the electroencephalogram channels corresponding to the input neurons with large proportion indexes and small average interaction, and selecting electroencephalogram data of other electroencephalogram channels to retrain and complete classification. The invention can effectively select the electroencephalogram channel and improve the classification accuracy.
Description
Technical Field
The invention relates to the technical field of brain-computer interfaces, in particular to a motor imagery electroencephalogram channel selection method and system.
Background
The Brain-computer interface (Brain-Computer Interface, BCI) is a man-machine interaction mode for realizing communication and control of human Brain and external electronic equipment based on Brain signals. Motor Imagery (MI) is a psychological process that imagines specific behaviors rather than actually performing those behaviors. BCI systems based on electroencephalogram (EEG) data generated by motor imagery can help establish a connection between the brain and external devices, can be used to maneuver, control unmanned aerial vehicles, help disabled or communication impaired patients resume control and communication functions, and like tasks.
While a large number of brain electrical channels can provide more background neural activity information, they also introduce redundancy and noise data, resulting in high-dimensional data. The channel selection method can be used for reducing irrelevant noise and redundant information, thereby reducing computational complexity, improving classification accuracy by avoiding overfitting, reducing setup time, and providing a BCI with better performance. However, the existing channel selection and classification technologies have the problems of high operation complexity and high energy consumption.
Disclosure of Invention
The invention provides a motor imagery electroencephalogram channel selection method and a motor imagery electroencephalogram channel selection system, which are used for solving the problems existing in the prior art, and the technical scheme provided by the invention is as follows:
in one aspect, a motor imagery electroencephalogram channel selection method is provided, the method comprising:
s1, acquiring and preprocessing motor imagery electroencephalogram data with a plurality of electroencephalogram channels;
s2, mapping the electroencephalogram data to a NeuCube model based on a pulse neural network, and training the NeuCube model, wherein the NeuCube model constructs a three-dimensional SNN cube which simulates a brain and comprises neurons mapped by different electroencephalogram channels and other neurons, and the neurons are mapped by different electroencephalogram channels
Is called an input neuron;
s3, analyzing the space-time relationship of the input neurons in the Neucube model by adopting a clustering method, and grouping the neurons in the model according to pulse communication of the neurons in the training process to obtain a neuron cluster taking the input neurons as a cluster center;
s4, comparing the number of neurons in each neuron cluster with the total number of neurons in the NeuCube model to obtain the proportion index of each input neuron;
s5, determining average interaction among all input neurons according to pulse communication of neurons of different neuron clusters in the training process;
and S6, deleting the electroencephalogram channels corresponding to the input neurons with large proportion indexes and small average interaction as unimportant electroencephalogram channels, selecting motor imagery electroencephalogram data of other undeleted electroencephalogram channels, retraining a NeuCube model, and completing classification.
Optionally, the NeuCube model includes: the device comprises a data coding module, a three-dimensional SNN cube module and an output module;
the data coding module is used for converting the input electroencephalogram data into pulse sequences of different electroencephalogram channels;
the three-dimensional SNN cube module is used for mapping pulse sequences of different brain electrical channels to neurons in the three-dimensional cube, adopting a small world network to perform initial connection among different neurons, randomly setting initial connection weights, training in an unsupervised mode after initializing the three-dimensional cube, and modifying the initial connection weights by utilizing a pulse time dependent plasticity rule;
the output module is used for recognizing the cube state by using the predefined mode category classification by adopting the supervision training after the non-supervision training stage.
Optionally, the step S3 specifically includes:
taking the input neurons as cluster centers of neuron clusters, and marking with input variables representing brain electrical channels;
the remaining neurons of a cluster of neurons are determined based on the number of pulses received from each cluster center, each neuron belonging to a particular cluster of neurons, with the number of pulses received from the cluster center of the particular cluster of neurons being the greatest compared to the cluster centers of other clusters of neurons.
Optionally, the step S5 specifically includes:
calculating the pulse communication total quantity of pulse communication between the neurons of each neuron cluster and the neurons of other neuron clusters in the training process;
the resulting value of the total amount of impulse traffic for each of the neuron clusters divided by the number of neurons for each of the neuron clusters is used as the average interaction between the input neurons for each of the neuron clusters.
Optionally, the unimportant electroencephalogram channel comprises an electroencephalogram channel corresponding to at least one input neuron with a large scale index and small average interactions.
In another aspect, there is provided a motor imagery electroencephalogram channel selection system, the system comprising:
the acquisition preprocessing module is used for acquiring and preprocessing motor imagery electroencephalogram data with a plurality of electroencephalogram channels;
the mapping training module is used for mapping the electroencephalogram data to a NeuCube model based on a pulse neural network and training the NeuCube model, and the NeuCube model is used for constructing a three-dimensional SNN cube which simulates a brain and comprises neurons mapped by different electroencephalogram channels and other neurons, wherein the neurons mapped by different electroencephalogram channels are called as input neurons;
the grouping module is used for analyzing the space-time relationship of the input neurons in the Neucube model by adopting a clustering method, and grouping the neurons in the model according to the pulse communication level of the neurons in the training process to obtain a neuron cluster taking the input neurons as a cluster center;
the comparison module is used for comparing the number of neurons in each neuron cluster with the total number of neurons in the NeuCube model to obtain the proportion index of each input neuron;
a determining module for determining an average interaction between the input neurons according to impulse communication of neurons of different neuron clusters during training;
and the selection module is used for deleting the electroencephalogram channels corresponding to the input neurons with large proportion indexes and small average interaction as unimportant electroencephalogram channels, selecting motor imagery electroencephalogram data of other undeleted electroencephalogram channels, retraining the NeuCube model, and finishing classification.
Optionally, the NeuCube model includes: the device comprises a data coding module, a three-dimensional SNN cube module and an output module;
the data coding module is used for converting the input electroencephalogram data into pulse sequences of different electroencephalogram channels;
the three-dimensional SNN cube module is used for mapping pulse sequences of different brain electrical channels to neurons in the three-dimensional cube, adopting a small world network to perform initial connection among different neurons, randomly setting initial connection weights, training in an unsupervised mode after initializing the three-dimensional cube, and modifying the initial connection weights by utilizing a pulse time dependent plasticity rule;
the output module is used for recognizing the cube state by using the predefined mode category classification by adopting the supervision training after the non-supervision training stage.
Optionally, the grouping module is specifically configured to:
taking the input neurons as cluster centers of neuron clusters, and marking with input variables representing brain electrical channels;
the remaining neurons of a cluster of neurons are determined based on the number of pulses received from each cluster center, each neuron belonging to a particular cluster of neurons, with the number of pulses received from the cluster center of the particular cluster of neurons being the greatest compared to the cluster centers of other clusters of neurons.
Optionally, the determining module is specifically configured to:
calculating the pulse communication total quantity of pulse communication between the neurons of each neuron cluster and the neurons of other neuron clusters in the training process;
the resulting value of the total amount of impulse traffic for each of the neuron clusters divided by the number of neurons for each of the neuron clusters is used as the average interaction between the input neurons for each of the neuron clusters.
Optionally, the unimportant electroencephalogram channel comprises an electroencephalogram channel corresponding to at least one input neuron with a large scale index and small interactions.
In another aspect, an electronic device is provided, the electronic device including a processor and a memory, the memory having instructions stored therein, the instructions being loaded and executed by the processor to implement the motor imagery electroencephalogram channel selection method described above.
In another aspect, a computer-readable storage medium having instructions stored therein that are loaded and executed by a processor to implement the motor imagery electroencephalogram channel selection method described above is provided.
Compared with the prior art, the technical scheme has at least the following beneficial effects:
according to the invention, a low-energy architecture based on a pulse neural network is adopted to classify a motor imagery small data set, in the classifying process, the time and space information of the electroencephalogram signals are considered, the interaction information between the neuron cluster occupancy rate and the neurons corresponding to different electroencephalogram channels is calculated, the unimportant electroencephalogram channels are selected to be deleted, then the NeuCube model is retrained and classified, and the electroencephalogram channels can be selected while learning and understanding the electroencephalogram signals, so that the energy consumption is reduced, the overfitting is avoided, the classifying precision is improved, and the practical application performance of a BCI system is improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a motor imagery electroencephalogram channel selection method provided by an embodiment of the invention;
FIG. 2 is a structure diagram of a NeuCube model provided by an embodiment of the invention;
FIG. 3 is a diagram of a neuron cube according to an embodiment of the invention;
FIG. 4 is a diagram of initialized neuronal cube network connections according to an embodiment of the present invention;
FIG. 5 is a diagram of a cube connection after training in an unsupervised manner, provided by an embodiment of the present invention;
fig. 6 is a graph of a result after clustering neurons corresponding to twenty-two brain electrical channels according to an embodiment of the present invention;
FIG. 7 is a schematic diagram of the percentage of the number of neurons in each neuron cluster to the total number of neurons in the NeuCube model according to an embodiment of the present invention;
FIG. 8 is a schematic diagram illustrating the average interactions between input neurons of each of the neuron clusters provided by an embodiment of the present invention;
fig. 9 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. It will be apparent that the described embodiments are some, but not all, embodiments of the invention. All other embodiments, which can be made by a person skilled in the art without creative efforts, based on the described embodiments of the present invention fall within the protection scope of the present invention.
As shown in fig. 1, an embodiment of the present invention provides a motor imagery electroencephalogram channel selection method, which includes:
s1, acquiring and preprocessing motor imagery electroencephalogram data with a plurality of electroencephalogram channels;
s2, mapping the electroencephalogram data to a NeuCube model based on a pulse neural network, and training the NeuCube model, wherein the NeuCube model constructs a three-dimensional SNN cube which simulates a brain and comprises neurons mapped by different electroencephalogram channels and other neurons, and the neurons mapped by different electroencephalogram channels are called as input neurons;
s3, analyzing the space-time relationship of the input neurons in the Neucube model by adopting a clustering method, and grouping the neurons in the model according to pulse communication of the neurons in the training process to obtain a neuron cluster taking the input neurons as a cluster center;
s4, comparing the number of neurons in each neuron cluster with the total number of neurons in the NeuCube model to obtain the proportion index of each input neuron;
s5, determining average interaction among all input neurons according to pulse communication of neurons of different neuron clusters in the training process;
and S6, deleting the electroencephalogram channels corresponding to the input neurons with large proportion indexes and small average interaction as unimportant electroencephalogram channels, selecting motor imagery electroencephalogram data of other undeleted electroencephalogram channels, retraining a NeuCube model, and completing classification.
The following describes in detail a motor imagery electroencephalogram channel selection method provided by an embodiment of the present invention with reference to fig. 2 to 8, including:
s1, acquiring and preprocessing motor imagery electroencephalogram data with a plurality of electroencephalogram channels;
the embodiment of the invention adopts a Butterworth band-pass filter to screen the motor imagery electroencephalogram data in a certain frequency range, and intercepts the data with fixed time length in the middle of the data for training and classification. The preferred scheme is that a 6-order Butterworth band-pass filter is adopted to screen motor imagery electroencephalogram data in the frequency range of 8-30Hz, and data in the middle of the data (motor imagery of a tested person) are intercepted for training and classification.
The electroencephalogram data has 22 electroencephalogram channels, namely 22 electrodes are arranged on electrode caps worn by testees, 22 different electroencephalogram data (space-time data) with time space information are collected, and all data in the finally obtained data set are in the format of:wherein the processed dataset comprises 9 folders with 576 csv files under each folder, i.e. 576 samples, according to 7:3 separate, 403 for training and 173 for testing, within each. Csv file isFor each value, the following table describes:
s2, mapping the electroencephalogram data to a NeuCube model based on a pulse neural network, and training the NeuCube model, wherein the NeuCube model constructs a three-dimensional SNN cube which simulates a brain and comprises neurons mapped by different electroencephalogram channels and other neurons, and the neurons mapped by different electroencephalogram channels are called as input neurons;
optionally, as shown in fig. 2, the NeuCube model includes: the device comprises a data coding module, a three-dimensional SNN cube module and an output module;
the data coding module is used for converting the input electroencephalogram data into pulse sequences of different electroencephalogram channels;
specifically, the embodiment of the invention adopts a SF (step forward) coding algorithm to convert the input electroencephalogram data into pulse sequences of different electroencephalogram channels, the floating point number in metadata is similarly converted into data only containing 0 and 1, and then the pulse sequences are imported into a three-dimensional SNN cube of NeuCube for processing and classification.
The three-dimensional SNN cube module is configured to map pulse sequences of the different brain electrical channels to neurons in the three-dimensional cube, and the different neurons are initially connected by using a small world network, where the small world network neurons in the embodiment of the invention are connected within a certain radius (for example, r=3), and on the basis of mapping of the international 10-5 positioning system, a structure composed of 346 neurons is designed, and all 22 brain electrical channels are mapped to the cube, as shown in fig. 3, the 22 input neurons include: FZ, POz, P2, FC3, etc., are indicated by dots in fig. 3, and other neurons are indicated by star points.
The initial connection weights are set randomly, the initial three-dimensional cube can be regarded as a liquid machine (Liquid state machines, LSM) which is composed of a large number of single neurons which are randomly interconnected to form an LSM network, the initialized network connection is shown in fig. 4, wherein the brightness of the neuron points represents the initial activation intensity of the initial connection weights, and the connection lines among the neurons represent the random initial connection weights.
After initializing the three-dimensional cube, training is performed in an unsupervised manner, using pulse-time dependent plasticity (STDP) rules to modify initial connection weights, and the cube connections after training in an unsupervised manner are shown in fig. 5, where the brightness of the neuron points represents the intensity of their activation after training, and the links between neurons represent the connection weights after training, including excitatory and inhibitory connections.
The output module is used for recognizing the cube state by using the predefined mode category classification by adopting the supervision training after the non-supervision training stage.
S3, analyzing the space-time relationship of the input neurons in the Neucube model by adopting a clustering method, and grouping the neurons in the model according to pulse communication of the neurons in the training process to obtain a neuron cluster taking the input neurons as a cluster center;
optionally, the step S3 specifically includes:
taking the input neurons as cluster centers of neuron clusters, and marking with input variables representing brain electrical channels;
the remaining neurons of a cluster of neurons are determined based on the number of pulses received from each cluster center, each neuron belonging to a particular cluster of neurons, with the number of pulses received from the cluster center of the particular cluster of neurons being the greatest compared to the cluster centers of other clusters of neurons.
The result of clustering the neurons corresponding to twenty-two brain electrical channels is shown in fig. 6.
S4, comparing the number of neurons in each neuron cluster with the total number of neurons in the NeuCube model to obtain the proportion index of each input neuron;
the embodiment of the invention designs a proportional index, and the calculation formula of the index is as follows: the percentage of the number of neurons in each cluster to the total number of neurons in the NeuCube model is shown in fig. 7 (test subject No. 3), where the input neurons FZ and POz have the largest cluster proportions of 26.1% and 25.8%, respectively. This suggests that they are two brain electrical channels that require attention.
S5, determining average interaction among all input neurons according to pulse communication of neurons of different neuron clusters in the training process;
optionally, the step S5 specifically includes:
calculating the pulse communication total quantity of pulse communication between the neurons of each neuron cluster and the neurons of other neuron clusters in the training process;
specifically, a pulse is sent between two neurons, that is, a communication is generated, and how many times the neurons of each neuron cluster and the neurons of other neuron clusters share in the training process are counted.
The resulting value of the total amount of impulse traffic for each of the neuron clusters divided by the number of neurons for each of the neuron clusters is used as the average interaction between the input neurons for each of the neuron clusters.
The average interactions between the input neurons of each neuron cluster are shown in fig. 8, wherein the thickness of the line represents the intensity between the input neurons of each neuron cluster in the unsupervised learning stage, the thicker the line is, the larger the average interactions are, the thinner the line is, and the smaller the average interactions are, and it can be seen that the input neurons with the largest proportion index, namely FZ and POz, have the smallest average interactions of the electroencephalogram channels corresponding to the other input neurons.
And S6, deleting the electroencephalogram channels corresponding to the input neurons with large proportion indexes and small average interaction as unimportant electroencephalogram channels, selecting motor imagery electroencephalogram data of other undeleted electroencephalogram channels, retraining a NeuCube model, and completing classification.
Optionally, the unimportant electroencephalogram channel comprises an electroencephalogram channel corresponding to at least one input neuron with a large scale index and small average interactions.
In the embodiment of the invention, FZ and POz are input neurons with large proportion index and small average interaction, the size and the size are not absolute, and at least one of the numerical values in the front can be selected according to the requirement.
The embodiment of the invention does not select one of the proportional index or the average interaction as the selection basis, but combines the results of the proportional index and the average interaction to select the unimportant electroencephalogram channel more accurately.
The embodiment of the invention also provides a motor imagery electroencephalogram channel selection system, which comprises:
the acquisition preprocessing module is used for acquiring and preprocessing motor imagery electroencephalogram data with a plurality of electroencephalogram channels;
the mapping training module is used for mapping the electroencephalogram data to a NeuCube model based on a pulse neural network and training the NeuCube model, and the NeuCube model is used for constructing a three-dimensional SNN cube which simulates a brain and comprises neurons mapped by different electroencephalogram channels and other neurons, wherein the neurons mapped by different electroencephalogram channels are called as input neurons;
the grouping module is used for analyzing the space-time relationship of the input neurons in the Neucube model by adopting a clustering method, and grouping the neurons in the model according to the pulse communication level of the neurons in the training process to obtain a neuron cluster taking the input neurons as a cluster center;
the comparison module is used for comparing the number of neurons in each neuron cluster with the total number of neurons in the NeuCube model to obtain the proportion index of each input neuron;
a determining module for determining an average interaction between the input neurons according to impulse communication of neurons of different neuron clusters during training;
and the selection module is used for deleting the electroencephalogram channels corresponding to the input neurons with large proportion indexes and small average interaction as unimportant electroencephalogram channels, selecting motor imagery electroencephalogram data of other undeleted electroencephalogram channels, retraining the NeuCube model, and finishing classification.
Optionally, the NeuCube model includes: the device comprises a data coding module, a three-dimensional SNN cube module and an output module;
the data coding module is used for converting the input electroencephalogram data into pulse sequences of different electroencephalogram channels;
the three-dimensional SNN cube module is used for mapping pulse sequences of different brain electrical channels to neurons in the three-dimensional cube, adopting a small world network to perform initial connection among different neurons, randomly setting initial connection weights, training in an unsupervised mode after initializing the three-dimensional cube, and modifying the initial connection weights by utilizing a pulse time dependent plasticity rule;
the output module is used for recognizing the cube state by using the predefined mode category classification by adopting the supervision training after the non-supervision training stage.
Optionally, the grouping module is specifically configured to:
taking the input neurons as cluster centers of neuron clusters, and marking with input variables representing brain electrical channels;
the remaining neurons of a cluster of neurons are determined based on the number of pulses received from each cluster center, each neuron belonging to a particular cluster of neurons, with the number of pulses received from the cluster center of the particular cluster of neurons being the greatest compared to the cluster centers of other clusters of neurons.
Optionally, the determining module is specifically configured to:
calculating the pulse communication total quantity of pulse communication between the neurons of each neuron cluster and the neurons of other neuron clusters in the training process;
the resulting value of the total amount of impulse traffic for each of the neuron clusters divided by the number of neurons for each of the neuron clusters is used as the average interaction between the input neurons for each of the neuron clusters.
Optionally, the unimportant electroencephalogram channel comprises an electroencephalogram channel corresponding to at least one input neuron with a large scale index and small interactions.
The specific functional structure of the motor imagery electroencephalogram channel selection system provided by the embodiment of the invention corresponds to the motor imagery electroencephalogram channel selection method provided by the embodiment of the invention, and the detailed description is omitted.
Fig. 9 is a schematic structural diagram of an electronic device 900 according to an embodiment of the present invention, where the electronic device 900 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 901 and one or more memories 902, where the memories 902 store instructions, and the instructions are loaded and executed by the processors 901 to implement the steps of the motor imagery electroencephalogram channel selection method described above.
In an exemplary embodiment, a computer readable storage medium, such as a memory including instructions executable by a processor in a terminal to perform the motor imagery electroencephalogram channel selection method described above, is also provided. For example, the computer readable storage medium may be ROM, random Access Memory (RAM), CD-ROM, magnetic tape, floppy disk, optical data storage device, etc.
It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program for instructing relevant hardware, where the program may be stored in a computer readable storage medium, and the storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.
The foregoing description of the preferred embodiments of the invention is not intended to limit the invention to the precise form disclosed, and any such modifications, equivalents, and alternatives falling within the spirit and scope of the invention are intended to be included within the scope of the invention.
Claims (6)
1. A motor imagery electroencephalogram channel selection method, characterized in that the method comprises:
s1, acquiring and preprocessing motor imagery electroencephalogram data with a plurality of electroencephalogram channels;
s2, mapping the electroencephalogram data to a NeuCube model based on a pulse neural network, and training the NeuCube model, wherein the NeuCube model constructs a three-dimensional SNN cube which simulates a brain and comprises neurons mapped by different electroencephalogram channels and other neurons, and the neurons mapped by different electroencephalogram channels are called as input neurons;
s3, analyzing the space-time relationship of the input neurons in the Neucube model by adopting a clustering method, and grouping the neurons in the model according to pulse communication of the neurons in the training process to obtain a neuron cluster taking the input neurons as a cluster center;
s4, comparing the number of neurons in each neuron cluster with the total number of neurons in the NeuCube model to obtain the proportion index of each input neuron;
s5, determining average interaction among all input neurons according to pulse communication of neurons of different neuron clusters in the training process;
s6, deleting an electroencephalogram channel corresponding to an input neuron with a large proportion index and small average interaction as an unimportant electroencephalogram channel, and selecting motor imagery electroencephalogram data of other undeleted electroencephalogram channels to retrain a NeuCube model and finish classification;
the NeuCube model comprises: the device comprises a data coding module, a three-dimensional SNN cube module and an output module;
the data coding module is used for converting the input electroencephalogram data into pulse sequences of different electroencephalogram channels;
the three-dimensional SNN cube module is used for mapping pulse sequences of different brain electrical channels to neurons in the three-dimensional cube, adopting a small world network to perform initial connection among different neurons, randomly setting initial connection weights, training in an unsupervised mode after initializing the three-dimensional cube, and modifying the initial connection weights by utilizing a pulse time dependent plasticity rule;
the output module is used for recognizing the cube state by using the predefined mode category classification by adopting the supervision training after the non-supervision training stage.
2. The method according to claim 1, wherein S3 specifically comprises:
taking the input neurons as cluster centers of neuron clusters, and marking with input variables representing brain electrical channels;
the remaining neurons of a cluster of neurons are determined based on the number of pulses received from each cluster center, each neuron belonging to a particular cluster of neurons, with the number of pulses received from the cluster center of the particular cluster of neurons being the greatest compared to the cluster centers of other clusters of neurons.
3. The method according to claim 1, wherein S5 specifically comprises:
calculating the pulse communication total quantity of pulse communication between the neurons of each neuron cluster and the neurons of other neuron clusters in the training process;
the resulting value of the total amount of impulse traffic for each of the neuron clusters divided by the number of neurons for each of the neuron clusters is used as the average interaction between the input neurons for each of the neuron clusters.
4. A motor imagery electroencephalogram channel selection system, the system comprising:
the acquisition preprocessing module is used for acquiring and preprocessing motor imagery electroencephalogram data with a plurality of electroencephalogram channels;
the mapping training module is used for mapping the electroencephalogram data to a NeuCube model based on a pulse neural network and training the NeuCube model, and the NeuCube model is used for constructing a three-dimensional SNN cube which simulates a brain and comprises neurons mapped by different electroencephalogram channels and other neurons, wherein the neurons mapped by different electroencephalogram channels are called as input neurons;
the grouping module is used for analyzing the space-time relationship of the input neurons in the Neucube model by adopting a clustering method, and grouping the neurons in the model according to the pulse communication level of the neurons in the training process to obtain a neuron cluster taking the input neurons as a cluster center;
the comparison module is used for comparing the number of neurons in each neuron cluster with the total number of neurons in the NeuCube model to obtain the proportion index of each input neuron;
a determining module for determining an average interaction between the input neurons according to impulse communication of neurons of different neuron clusters during training;
the selection module is used for deleting the electroencephalogram channels corresponding to the input neurons with large proportion indexes and small average interaction as unimportant electroencephalogram channels, selecting motor imagery electroencephalogram data of other undeleted electroencephalogram channels, retraining a Neucube model, and finishing classification;
the NeuCube model comprises: the device comprises a data coding module, a three-dimensional SNN cube module and an output module;
the data coding module is used for converting the input electroencephalogram data into pulse sequences of different electroencephalogram channels;
the three-dimensional SNN cube module is used for mapping pulse sequences of different brain electrical channels to neurons in the three-dimensional cube, adopting a small world network to perform initial connection among different neurons, randomly setting initial connection weights, training in an unsupervised mode after initializing the three-dimensional cube, and modifying the initial connection weights by utilizing a pulse time dependent plasticity rule;
the output module is used for recognizing the cube state by using the predefined mode category classification by adopting the supervision training after the non-supervision training stage.
5. The system according to claim 4, wherein the grouping module is specifically configured to:
taking the input neurons as cluster centers of neuron clusters, and marking with input variables representing brain electrical channels;
the remaining neurons of a cluster of neurons are determined based on the number of pulses received from each cluster center, each neuron belonging to a particular cluster of neurons, with the number of pulses received from the cluster center of the particular cluster of neurons being the greatest compared to the cluster centers of other clusters of neurons.
6. The system according to claim 4, wherein the determining module is specifically configured to:
calculating the pulse communication total quantity of pulse communication between the neurons of each neuron cluster and the neurons of other neuron clusters in the training process;
the resulting value of the total amount of impulse traffic for each of the neuron clusters divided by the number of neurons for each of the neuron clusters is used as the average interaction between the input neurons for each of the neuron clusters.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311599176.9A CN117311516B (en) | 2023-11-28 | 2023-11-28 | Motor imagery electroencephalogram channel selection method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311599176.9A CN117311516B (en) | 2023-11-28 | 2023-11-28 | Motor imagery electroencephalogram channel selection method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117311516A CN117311516A (en) | 2023-12-29 |
CN117311516B true CN117311516B (en) | 2024-02-20 |
Family
ID=89260707
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311599176.9A Active CN117311516B (en) | 2023-11-28 | 2023-11-28 | Motor imagery electroencephalogram channel selection method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117311516B (en) |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
MX2018004398A (en) * | 2018-04-10 | 2019-10-11 | Rafael Roman Soltero Alberto | Process for the analysis of brain waves standardizing them with a tetra-dimensional space. |
KR20200068051A (en) * | 2018-11-26 | 2020-06-15 | 전자부품연구원 | Spiking neural network device and intelligent apparatus comprising the same |
CN111543983A (en) * | 2020-04-02 | 2020-08-18 | 天津大学 | Electroencephalogram signal channel selection method based on neural network |
CN112155549A (en) * | 2020-09-04 | 2021-01-01 | 西北师范大学 | ADHD disease diagnosis aid decision-making system based on deep convolution pulse neural network |
WO2021158085A1 (en) * | 2020-02-05 | 2021-08-12 | Samsung Electronics Co., Ltd. | Neural network update method, classification method and electronic device |
JP2021189727A (en) * | 2020-05-29 | 2021-12-13 | 株式会社日立製作所 | Computer system, learning method, and brain-type computing device |
CN114983439A (en) * | 2022-06-15 | 2022-09-02 | 四川省医学科学院·四川省人民医院 | Brain state identification method fusing impulse neural network and binary dynamic network |
WO2022183921A1 (en) * | 2021-03-01 | 2022-09-09 | 浙江大学 | Neural model mapping method of brain-like computer operating system |
CN116584954A (en) * | 2023-03-31 | 2023-08-15 | 北京讯脑医疗科技有限公司 | Electroencephalogram visualization method, control device, storage medium and visualization system |
CN116756643A (en) * | 2023-06-20 | 2023-09-15 | 复旦大学 | Fisher score-based electroencephalogram channel selection method |
CN116796207A (en) * | 2023-06-29 | 2023-09-22 | 东南大学 | Self-organizing mapping clustering method based on impulse neural network |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10579925B2 (en) * | 2013-08-26 | 2020-03-03 | Aut Ventures Limited | Method and system for predicting outcomes based on spatio/spectro-temporal data |
-
2023
- 2023-11-28 CN CN202311599176.9A patent/CN117311516B/en active Active
Patent Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
MX2018004398A (en) * | 2018-04-10 | 2019-10-11 | Rafael Roman Soltero Alberto | Process for the analysis of brain waves standardizing them with a tetra-dimensional space. |
KR20200068051A (en) * | 2018-11-26 | 2020-06-15 | 전자부품연구원 | Spiking neural network device and intelligent apparatus comprising the same |
WO2021158085A1 (en) * | 2020-02-05 | 2021-08-12 | Samsung Electronics Co., Ltd. | Neural network update method, classification method and electronic device |
CN111543983A (en) * | 2020-04-02 | 2020-08-18 | 天津大学 | Electroencephalogram signal channel selection method based on neural network |
JP2021189727A (en) * | 2020-05-29 | 2021-12-13 | 株式会社日立製作所 | Computer system, learning method, and brain-type computing device |
CN112155549A (en) * | 2020-09-04 | 2021-01-01 | 西北师范大学 | ADHD disease diagnosis aid decision-making system based on deep convolution pulse neural network |
WO2022183921A1 (en) * | 2021-03-01 | 2022-09-09 | 浙江大学 | Neural model mapping method of brain-like computer operating system |
CN114983439A (en) * | 2022-06-15 | 2022-09-02 | 四川省医学科学院·四川省人民医院 | Brain state identification method fusing impulse neural network and binary dynamic network |
CN116584954A (en) * | 2023-03-31 | 2023-08-15 | 北京讯脑医疗科技有限公司 | Electroencephalogram visualization method, control device, storage medium and visualization system |
CN116756643A (en) * | 2023-06-20 | 2023-09-15 | 复旦大学 | Fisher score-based electroencephalogram channel selection method |
CN116796207A (en) * | 2023-06-29 | 2023-09-22 | 东南大学 | Self-organizing mapping clustering method based on impulse neural network |
Non-Patent Citations (4)
Title |
---|
EEG Feature Selection via Global Redundancy Minimization for Emotion Recognition;Xu, Xueyuan;《IEEE TRANSACTIONS ON AFFECTIVE COMPUTING》;全文 * |
Emotion Classification Using EEG Data in a Brain-Inspired Spiking Neural Network;Yulin He;《2021 11th International Conference on Intelligent Control and Information Processing (ICICIP)》;全文 * |
一种重复二分CSP 4类运动想象脑电信号特征提取算法;郑戍华;闫琛;王向周;;北京理工大学学报(第08期);全文 * |
郑戍华 ; 闫琛 ; 王向周 ; .一种重复二分CSP 4类运动想象脑电信号特征提取算法.北京理工大学学报.2016,(第08期),全文. * |
Also Published As
Publication number | Publication date |
---|---|
CN117311516A (en) | 2023-12-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107273845B (en) | Facial expression recognition method based on confidence region and multi-feature weighted fusion | |
CN111858989B (en) | Pulse convolution neural network image classification method based on attention mechanism | |
RU2586864C2 (en) | Method and arrangement for local rule of training competition, which leads to a sparse connectivity | |
Harnad et al. | Categorical perception and the evolution of supervised learning in neural nets | |
CN113180692B (en) | Electroencephalogram signal classification and identification method based on feature fusion and attention mechanism | |
CN113712573A (en) | Electroencephalogram signal classification method, device, equipment and storage medium | |
CN102279358A (en) | MCSKPCA based neural network fault diagnosis method for analog circuits | |
CN113297994B (en) | Pilot behavior analysis method and system | |
CN113722474A (en) | Text classification method, device, equipment and storage medium | |
EP4303759A1 (en) | Physiological electrical signal classification and processing method and apparatus, computer device, and storage medium | |
CN113143295A (en) | Equipment control method and terminal based on motor imagery electroencephalogram signals | |
CN113133769A (en) | Equipment control method, device and terminal based on motor imagery electroencephalogram signals | |
CN108985442B (en) | Handwriting model training method, handwritten character recognition method, device, equipment and medium | |
CN113705715B (en) | Time sequence classification method based on LSTM and multi-scale FCN | |
CN114781441A (en) | EEG motor imagery classification method and multi-space convolution neural network model | |
CN112259228B (en) | Depression screening method by dynamic attention network non-negative matrix factorization | |
CN117311516B (en) | Motor imagery electroencephalogram channel selection method and system | |
CN117292750A (en) | Cell type duty ratio prediction method, device, equipment and storage medium | |
US20230101539A1 (en) | Physiological electric signal classification processing method and apparatus, computer device and storage medium | |
Ridoy et al. | A lightweight convolutional neural network for white blood cells classification | |
CN115169384A (en) | Electroencephalogram classification model training method, intention identification method, equipment and medium | |
CN114626408A (en) | Electroencephalogram signal classification method and device, electronic equipment, medium and product | |
CN114721514A (en) | Geometric model selection intention distinguishing method and system based on electroencephalogram signals | |
CN115381467B (en) | Attention mechanism-based time-frequency information dynamic fusion decoding method and device | |
CN114092996A (en) | Expression recognition method and device, electronic equipment and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |