CN114305451B - Construction method of children brain electrical function connection map of entropy stability criterion - Google Patents
Construction method of children brain electrical function connection map of entropy stability criterion Download PDFInfo
- Publication number
- CN114305451B CN114305451B CN202210094726.0A CN202210094726A CN114305451B CN 114305451 B CN114305451 B CN 114305451B CN 202210094726 A CN202210094726 A CN 202210094726A CN 114305451 B CN114305451 B CN 114305451B
- Authority
- CN
- China
- Prior art keywords
- connection
- network
- functional
- brain
- channel
- 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
- 210000004556 brain Anatomy 0.000 title claims abstract description 40
- 238000010276 construction Methods 0.000 title claims abstract description 14
- 238000000034 method Methods 0.000 claims abstract description 32
- 238000005259 measurement Methods 0.000 claims abstract description 13
- 230000010354 integration Effects 0.000 claims abstract description 10
- 238000000926 separation method Methods 0.000 claims abstract description 10
- 238000000605 extraction Methods 0.000 claims abstract description 6
- 238000007781 pre-processing Methods 0.000 claims abstract description 5
- 239000011159 matrix material Substances 0.000 claims description 23
- 230000003925 brain function Effects 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 12
- 238000012935 Averaging Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 10
- 238000001514 detection method Methods 0.000 claims description 9
- 238000001914 filtration Methods 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 8
- 239000012634 fragment Substances 0.000 claims description 7
- 238000005070 sampling Methods 0.000 claims description 7
- 230000000694 effects Effects 0.000 claims description 6
- 238000000354 decomposition reaction Methods 0.000 claims description 5
- 239000013598 vector Substances 0.000 claims description 5
- 230000003595 spectral effect Effects 0.000 claims description 4
- 238000000585 Mann–Whitney U test Methods 0.000 claims description 3
- 238000000528 statistical test Methods 0.000 claims description 3
- 238000003012 network analysis Methods 0.000 claims description 2
- 230000011218 segmentation Effects 0.000 claims description 2
- 238000011282 treatment Methods 0.000 claims description 2
- 239000003795 chemical substances by application Substances 0.000 claims 1
- 238000004458 analytical method Methods 0.000 abstract description 13
- 208000014644 Brain disease Diseases 0.000 description 12
- 238000010586 diagram Methods 0.000 description 4
- 238000002599 functional magnetic resonance imaging Methods 0.000 description 3
- 230000008667 sleep stage Effects 0.000 description 3
- 208000032274 Encephalopathy Diseases 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 208000024827 Alzheimer disease Diseases 0.000 description 1
- 206010003805 Autism Diseases 0.000 description 1
- 208000020706 Autistic disease Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 206010015037 epilepsy Diseases 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 208000013403 hyperactivity Diseases 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 201000000980 schizophrenia Diseases 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Landscapes
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention discloses a method for constructing a children brain electrical function connection map based on entropy stability criteria. Firstly, data preprocessing is carried out, then functional connection network construction is carried out, and finally network feature extraction and analysis are carried out. According to the invention, an optimal threshold value is searched by adopting a combination method of a random network and shannon entropy measurement, and a binary network constructed based on the method has stronger intra-group network stability compared with binary networks constructed in other modes, namely, personalized connection parts of the network are fewer, and age characteristics of the network are more highlighted. The graph theory features of individual level and age group level functional connection are extracted respectively, and the graph theory features comprise functional separation, functional integration, change rules of network centrality measurement analysis functional connection along with age and the characteristics of functional connection modes of specific age stages. Wherein, the change trend of the functional connection accords with the medical common sense, which verifies the rationality of the functional connection construction method.
Description
Technical Field
The invention belongs to the field of intelligent biomedical signal processing, and relates to a method for constructing a children brain electrical function connection map of entropy stability criterion.
Background
During childhood, the brain varies greatly in structure and function. Subtle changes in this stage may be greatly amplified as the different development processes develop, creating profound effects. Deviations from normal progression may lead to diseases such as hyperactivity, autism and schizophrenia. Macroscopic activities of the brain are mainly manifested in interactions between different brain regions. The traditional univariate feature cannot capture the information, and the functional connection is used as the relevance feature between the bivariate, so that the degree of information interaction between the brain areas can be described. Therefore, the construction of the reliable normal children brain function connection map not only can analyze the change track of the brain function connection of children, but also has important significance for brain disease detection.
Functional connections are typically constructed based on functional magnetic resonance imaging (fMRI) data or electroencephalogram (EEG) data. Compared with fMRI, EEG data has higher resolution in the time domain, can meet the requirements of cortex function connection analysis in the space domain, and is relatively low in EEG signal acquisition cost. Thus, more and more researchers are using EEG data for brain function connection related studies. The current study of EEG-based functional connections is mainly focused on two directions of brain disease detection and age analysis:
and (3) encephalopathy detection: brain diseases include epilepsy, alzheimer's disease, depression, and the like. The difference of brain function connection of a brain disease patient and a normal person is analyzed by using an intelligent decision model, so that the detection of the brain disease is realized.
Age analysis: the effect of aging on brain function connections is often studied. For example, individuals are simply divided into young and old groups for comparison analysis, or the trend of change in functional connection characteristics with age is studied based on individuals over a larger age span.
The studies in the above fields have the following problems: 1. intelligent brain disease detection in order to distinguish patients from normal control groups of corresponding ages, functional connection construction is generally performed by maximizing the difference between groups, so that the functional connection of the normal control groups is contrary to the actual situation. 2. The age analysis field lacks a method of constructing and analyzing brain function connections for each important age group of children. The method is also of great significance in assisting the diagnosis of childhood encephalopathy.
In addition, many functional connection-related studies do not binarize the functional connection, which would result in noise-related connections not being removed.
Disclosure of Invention
Aiming at the problems and the defects of the brain function connection analysis method, the invention provides a children brain function connection map construction method of an entropy stability criterion. The method divides children aged 0-17 into 7 age groups, weakens individual part functional connection, extracts individual level functional connection and age group level functional connection with higher stability, and analyzes brain functional connection differences and frequency band importance of different age groups by using graph theory and statistical methods.
Since age influences the pattern of functional connections, functional connection-based brain disease detection is often compared with normal functional connections of the corresponding age and analyzed for the presence of abnormal connections, but the functional connection construction method aims at maximizing the inter-group variability, under which the functional connection of the normal control group fails to be close to the real case. Under the stability criterion, the invention can construct the function connection with outstanding age correlation, and can overcome the defects of the function connection construction method of the normal control group in the current brain disease detection. In addition, analysis is performed based on age groups, so that the change rule of the functional connection of normal children along with the age can be obtained, and the functional connection characteristics of a specific age group are reflected. The method is suitable for processing the electroencephalogram data of children in all age groups, and can obtain ideal analysis results.
In the invention, the original signal data is preprocessed firstly, and the method comprises channel selection, digital filtering, signal slicing, sample filtering based on a threshold value and standard frequency band decomposition. Threshold optimization is performed on the preprocessed EEG data by combining an entropy-based stability metric with a random network proxy method, thereby constructing individual-level binarized functional connections. Individual level functional connections in turn construct age group level functional connections by majority voting. In order to evaluate the degree of functional separation and functional integration of brain functional connection in different age groups, global clustering coefficients and characteristic path length characteristics of individual horizontal functional connection are extracted respectively to analyze the change trend of the global clustering coefficients and the characteristic path length characteristics along with the age, and a statistical test method is combined to analyze the frequency band importance; in order to evaluate the centrality of the brain network, the degree characteristics of functional connection of the age group level are extracted, and the balance degree of the area where the network center is located and the degree distribution is analyzed.
The technical scheme of the invention mainly comprises the following steps:
step 1: and (5) preprocessing data.
Channel selection is performed on the raw multi-channel quiet sleep period EEG signal data. And (3) carrying out 1-30HZ band-pass filtering and 50HZ power frequency filtering to remove signal noise based on the frequency band of the main information of the EEG signal. Cutting the signal into 1s segments, setting a 150uV threshold according to the amplitude range of the physiological signal, removing the segments containing the non-physiological artifact signals to obtain a clean EEG signal, and finally carrying out signal decomposition on the EEG according to a standard frequency band.
Step 2: and (5) constructing a functional connection network.
An associative feature matrix is computed ICoh for the EEG segments of each frequency band. Based on the stability criteria and by means of a random network proxy, an optimal threshold is found to construct individual-level binarization functional connections. And finally, constructing the age group level functional connection by using a voting method.
Step 3: and (5) extracting and analyzing network characteristics. And extracting global clustering coefficients of individual horizontal functional connection, evaluating key frequency bands based on a statistical test method, and analyzing the change rule of features under the key frequency bands along with the age. In addition, the degree of age group horizontal functional connection is extracted, and the balance degree of network centrality and degree distribution is analyzed.
The specific flow of the step 1 is as follows:
1-1 age group division. The brain electrical data set of children aged 0-17 is divided into 7 age groups of less than 1 month, 1-3 months, 3 months-1 year, 1-3 years, 3-7 years, 7-14 years, and 14-17 years. Each age group contained 6 individual data for children.
1-2, Selecting channels. The A1, A2, fz, cz and Pz channels in the original multi-channel EEG signal are removed, and the selected channels are Fp1, fp2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, T3, T4, T5 and T6 which are 16 in total.
1-3. Unifying sampling frequency. The channel-selected EEG signals are downsampled to the lowest sampling frequency of the data set for unification.
And 1-4, removing high and low frequency and alternating current noise. The EEG signal is base pre-processed by a 1-30HZ band pass filter, a 50HZ notch filter.
And 1-5, data segmentation and non-physiological signal filtering. Dividing the preprocessed data into a plurality of periods with the duration of 1 second, setting 150uV as a threshold value, deleting the signal fragments, and removing irrelevant non-physiological signals. Each individual retains 1001 second 16 channel EEG signal segments.
1-6. Signal decomposition. The signal is decomposed into delta (1-4 hz), theta (4-8 hz), alpha (8-14 hz), beta (14-30 hz) 4 standard frequency bands using a band pass filter.
The specific flow of the step 2 is as follows:
And 2-1, computing ICoh relevance feature matrixes of all EEG fragments under each standard frequency band.
The correlation characteristic extraction of the coherence imaginary part ICoh is carried out on the preprocessed 16-channel 1-second EEG sample, and the index can overcome false connection caused by volume conduction effect. ICoh the characteristic calculation formula is:
P i,j (f) is the cross-power spectral density between channel i and channel j (channel i, j is set to node v i,vj during the construction of the functional connection), P i,i and P j,j are the self-power spectral densities of channel i and channel j, respectively, and N is the number of signal sampling points. The ICoh index sets between all channel pairs form an associative feature matrix, and the diagonal of the matrix is set to 0.
And 2-2, averaging ICoh correlation feature matrixes of all fragments in the individual body to obtain an individual-level weighted connection matrix.
2-3, Threshold optimization:
1) Firstly, an initial proportion threshold t is selected in the range of (0, 0.5), and the weighted connection matrix is binarized to obtain a binary connection matrix. The individual binary connection matrices for the same age group are then averaged and the shannon entropy of the matrix is calculated to measure the binary network stability below this threshold. The calculation formula of shannon entropy is as follows:
Where x represents the channel pair connection state variable. P (x) is the connection probability between each channel pair, i.e. the average value of the elements of all binary network corresponding positions of the same age group.
2) A random network is constructed to assist in threshold value optimization, and the method comprises the following steps:
the degree of each node of the actual individual binary connection matrix is kept unchanged, and the connection edges of the nodes are disturbed. In this way 100 random networks are built for each individual. And (3) carrying out stability measurement on the random network by using shannon entropy, and averaging measurement results to represent the entropy value of the unstable state network under the threshold value, and comparing and making difference with the actual network.
3) And traversing all the proportion thresholds in the range of the (0, 0.5) interval by taking 0.02 as a step length, finding the proportion threshold corresponding to the position with the largest difference between the stability of the random network and the stability of the actual network, and determining the proportion threshold as the optimal threshold. The optimal threshold corresponding to the functional connection of each standard frequency band of each age group is obtained through the method.
2-4, Unifying the threshold values.
After the optimal threshold value is obtained, the influence of different numbers of connecting edges on graph theory indexes needs to be avoided. And averaging the proportion thresholds of different age groups to unify, and finally constructing a stable individual binary network under each standard frequency band by using the averaged thresholds.
2-5. Functional links at age group level are constructed.
The majority voting method is adopted, namely when the connecting edge between certain two vertexes is more than half of 1 in the group, the connecting edge is reserved, otherwise, the connecting edge is set to 0.
The specific flow of the step 3 is as follows:
3-1. Extracting global cluster coefficients on individual level functional connections to measure functional separation of the brain.
Functional separation of the brain refers to the ability to perform specific treatments in closely related brain region groups. Global cluster coefficients are used here to measure functional separation of the brain. Global cluster coefficientsThe average of the local cluster coefficients for all vertices:
where n is the number of top points, and the calculation formula of the local cluster coefficient C i is as follows:
A ij is an element of the ith row and jth column of the individual binary connection matrix A, namely a connection weight (0 or 1) between the node v i and the node v j. And k is calculated as follows:
3-2. Functional integration of the feature path length metric brain is extracted over individual-level functional connections.
Functional integration of the brain refers to the ability to quickly integrate professional information from different brain regions. Here, the functional integration of the brain is measured using the characteristic path length, and the calculation formula of the characteristic path length l G is as follows:
Where d (v i,vj) is the shortest distance between nodes v i,vj.
And 3-3, analyzing the global clustering coefficient and the characteristic path length distribution difference among different age groups by utilizing a Wilcoxon rank sum test method. And averaging p values obtained by rank sum test to obtain the frequency band importance index.
3-4. Extraction of centrality features on age group level functional connections.
Degree centrality is the most direct measure of node centrality characterized in network analysis. The greater the degree of a node means the greater the centrality of the node, the more important the node is in the network. The calculation formula of the node degree centrality is as follows:
Where k i represents the number of edges connected to node v i. N-1 represents the number of edges that node v i connects to other nodes.
And 3-5, superposing the node degree centrality under each standard frequency band, calculating the variance of the node degree centrality, and finally normalizing. The calculation formula is as follows:
Where v a,b denotes a degree distribution vector in the b-th band of the a-th age group. The degree of imbalance of the centrality distribution of the brain function network at the cross-frequency level was measured by v a,b. The index is at a lower level at ages less than one month, because neonatal neurons less than one month have not yet developed, resulting in a more random discharge without the formation of a distinct network center. At the age of 1-3 months, the index rises significantly, which may be a result of the local brain regions developing rapidly at this stage and showing more activity than other brain regions during sleep. The index gradually decreases with age from 1 month to 14 years of age, i.e. the degree distribution of functional connectivity gradually becomes balanced. The change of the index is closely related to the increase of the age, and can reflect the degree of brain maturity to a certain extent.
The invention has the following beneficial effects:
1. the binary network constructed based on the method has stronger intra-group network stability, namely fewer personalized connection parts of the network and more outstanding age characteristics compared with binary networks constructed in other modes.
2. The graph theory features of individual level and age group level functional connection are extracted respectively, and the graph theory features comprise functional separation, functional integration, change rules of network centrality measurement analysis functional connection along with age and the characteristics of functional connection modes of specific age stages. Wherein, the change trend of the functional connection accords with the medical common sense, which verifies the rationality of the functional connection construction method.
Drawings
FIG. 1 is a functional connection construction flow chart of an embodiment of the present invention;
FIG. 2 is a diagram showing a map of a correlation feature matrix and ICoh in accordance with an embodiment ICoh of the present invention;
FIG. 3 is a view of an iteration of threshold optimization in accordance with an embodiment of the present invention;
FIG. 4 is a schematic diagram of an age group horizontal functional connection according to an embodiment of the present invention;
FIG. 5 is a diagram of global cluster coefficients and characteristic path lengths in the β -band according to an embodiment of the present invention;
Fig. 6 is a variance diagram of a cross-frequency node centrality distribution in an embodiment of the present invention.
Detailed Description
The invention is further described below with reference to the drawings and examples.
The first main step of the invention is signal acquisition and preprocessing, and the specific implementation steps are as follows:
1-1. Collecting brain electrical data of a normal child aged 0-17 years including sleep stages from a hospital, manually screening the original data, discarding data with poor signal quality (such as including electrocardiographic artifact, ref artifact, bad track and the like), and intercepting an EEG signal in a quiet sleep stage. The screened data is then divided by age group. The brain electrical data set of children aged 0-17 is divided into 7 age groups of less than 1 month, 1-3 months, 3 months-1 year, 1-3 years, 3-7 years, 7-14 years, and 14-17 years. Each age group contained 6 children individuals.
1-2. Remove the A1, A2, fz, cz, pz channels in the original multichannel EEG signal.
1-3 Downsampling the channel selected EEG signal to the lowest sampling frequency of the data set for unification.
And 1-4, denoising the signals through a 1-30HZ band-pass filter and a 50HZ notch filter.
1-5, Dividing the denoised data into a plurality of periods with the duration of 1 second, and setting 150uV as a threshold value to screen signal fragments.
1-6. The signals are decomposed using bandpass filters into delta (1-4 hz), theta (4-8 hz), alpha (8-14 hz), beta (14-30 hz) 4 standard frequency bands.
The second main step of the present invention is to construct a functional connection, and fig. 1 is a flowchart of this step, and the specific implementation steps are as follows:
And 2-1, carrying out ICoh-channel correlation feature extraction on the EEG sample after pretreatment. ICoh relevance feature matrix and ICoh distribution are shown in fig. 2.
And 2-2, averaging ICoh relevance feature matrixes of all the fragments to obtain an individual-level weighted connection matrix.
2-3, Threshold optimization: 1) Firstly, an initial proportion threshold t is selected in the range of (0, 0.5), and the weighted connection matrix is binarized. The individual binary connected matrices for the same age group are then averaged and the shannon entropy of the matrix is calculated. 2) The degree of each node of the actual individual binary network is kept unchanged, the connecting edges of the nodes are disturbed, and 100 random networks are built for each individual. And (3) carrying out stability measurement on the random network by using the shannon entropy, averaging the measurement results, comparing the measurement results with an actual network, and calculating a difference value. 3) And traversing all the proportion thresholds in the range of the (0, 0.5) interval by taking 0.02 as a step length, finding the proportion threshold corresponding to the position with the largest difference between the stability of the random network and the stability of the actual network, and determining the proportion threshold as the optimal threshold. Fig. 3 illustrates an iterative process for threshold optimization based on shannon entropy. The triangle mark corresponds to the proportional threshold that maximizes the functional link stability.
2-4. Averaging the ratio thresholds of different age groups to unify the thresholds, and constructing a binary network of individuals stable under the standard frequency band using the unifying thresholds.
2-5, Constructing functional connection of age group level by adopting a majority voting method. Fig. 4 is an age group horizontal functional connection.
The third main step of the invention is to extract the graph theory characteristics of the functional connection, and the specific implementation steps are as follows:
3-1, extracting global cluster coefficient to measure the functional separation of the brain on the functional connection of the individual level, and extracting characteristic path length to measure the functional integration of the brain. Fig. 5 reflects the trend of global cluster coefficients and feature path lengths over age.
And 3-2, respectively carrying out distribution difference analysis on the global clustering coefficients and the characteristic path lengths of different age groups by utilizing a Wilcoxon rank sum test method. And calculates an average p-value as a band importance measure.
3-3. Extraction of centrality features on age group level functional connections.
3-4, Superposing the degree vectors of the functional connection under each standard frequency band, calculating the variance of the degree vectors, and finally normalizing the degree vectors to measure the centrality distribution balance of the brain functional network on the cross-frequency level, wherein fig. 6 shows the measurement results of different age groups under the index.
3-5, Analyzing the characteristics of the functional connection by combining the current medical common knowledge so as to test the rationality of the individual and age group level functional connection construction method.
The method provided by the invention is applied to the analysis of the brain function connection difference of children at different ages and the diagnosis of auxiliary brain diseases. After ICoh connection matrixes of sleep stage EEG under a standard frequency band are extracted, individual level functional connection constructed based on the stability measurement of shannon entropy and age group level functional connection constructed based on majority voting show a change rule conforming to medical common sense under graph theory measurement. Therefore, the functional connection constructed based on the method can be used as a brain network map of children aged 0-17 years, and provides a reliable control reference for EEG-based brain disease detection.
Claims (3)
1. The construction method of the children brain electrical function connection map of the entropy stability criterion is characterized by comprising the following steps:
Step 1: preprocessing data;
Channel selection of original multi-channel quiet sleep period EEG signal data; based on the frequency band of the main information of the EEG signal, carrying out 1-30HZ band-pass filtering and 50HZ power frequency filtering to remove signal noise; cutting the signal into 1s segments, setting a 150uV threshold according to the amplitude range of the physiological signal to remove the segments containing the non-physiological artifact signals to obtain pure EEG signals, and finally carrying out signal decomposition on the EEG according to the standard frequency band;
Step 2: constructing a functional connection network;
computing ICoh a correlation feature matrix for the EEG segments of each frequency band; searching an optimal threshold value through a random network agent based on a stability criterion to construct individual-level binarization function connection; finally, constructing age group level functional connection by using a voting method;
Step 3: extracting and analyzing network characteristics; extracting global clustering coefficients of individual horizontal functional connection, evaluating key frequency bands based on a statistical test method, and analyzing the change rule of features under the key frequency bands along with age; extracting the degree of the age group horizontal functional connection, and analyzing the network centrality and the balance degree of the degree distribution;
the specific flow of the step 2 is as follows:
2-1, computing ICoh relevance feature matrixes of all EEG fragments under each standard frequency band;
The correlation characteristic extraction of the coherence imaginary part ICoh is carried out on the preprocessed 16-channel 1-second EEG sample, and the index can overcome false connection caused by volume conduction effect; ICoh the characteristic calculation formula is:
P i,j (f) is the cross power spectral density between channel i and channel j, wherein in the process of constructing the functional connection, the channel i and j are set as nodes v i,vj,Pi,i and P j,j respectively are the self power spectral densities of the channel i and the channel j, and N is the number of signal sampling points; the ICoh index sets among all channel pairs form an association feature matrix, and the diagonal lines of the matrix are all set to be 0;
2-2, averaging ICoh relevance feature matrixes of all fragments in the individual body to obtain an individual-level weighted connection matrix;
2-3, threshold optimization:
1) Firstly, selecting an initial proportion threshold t in the range of (0, 0.5), and binarizing the weighted connection matrix to obtain a binary connection matrix; then, averaging individual binary connection matrixes of the same age group, and calculating shannon entropy of the matrix to measure the binary network stability under the threshold; the calculation formula of shannon entropy is as follows:
Wherein x represents a channel pair connection state variable; p (x) is the connection probability between each channel pair, namely the average value of elements at the corresponding positions of all binary networks of the same age group;
2) A random network is constructed to assist in threshold value optimization, and the method comprises the following steps:
The degree of each node of the actual individual binary connection matrix is kept unchanged, and the connection edges of the nodes are disturbed; constructing 100 random networks for each individual in this manner; the stability of the random network is measured by using shannon entropy, and the measurement result is averaged to represent the entropy value of the unstable state network under the threshold value, and the entropy value is compared with the actual network and is poor;
3) Traversing all proportion thresholds in a range of (0, 0.5) with 0.02 as a step length, finding a proportion threshold corresponding to the position with the largest difference between the stability of the random network and the stability of the actual network, and determining the proportion threshold as an optimal threshold; obtaining an optimal threshold corresponding to the functional connection of each standard frequency band of each age group by the method;
2-4, unifying threshold values;
After the optimal threshold value is obtained, the influence of different numbers of connecting edges on graph theory indexes needs to be avoided; averaging the proportion thresholds of different age groups to unify, and finally constructing a stable individual binary network under each standard frequency band by using the average threshold;
2-5, constructing age group level functional connection;
The majority voting method is adopted, namely when the connecting edge between certain two vertexes is more than half of 1 in the group, the connecting edge is reserved, otherwise, the connecting edge is set to 0.
2. The method for constructing the children brain electrical function connection map according to the entropy stability criterion of claim 1, wherein the specific flow of the step 1 is as follows:
1-1, age group division; dividing the 0-17 year old child brain electrical data set into 7 age groups of less than 1 month, 1-3 months, 3 months-1 year old, 1-3 years old, 3-7 years old, 7-14 years old, 14-17 years old; each age group contains 6 individual data for children;
1-2, selecting a channel; removing the A1, A2, fz, cz and Pz channels in the original multi-channel EEG signal, wherein the selected channels are Fp1, fp2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, T3, T4, T5 and T6 which are 16 in total;
1-3, unifying sampling frequencies; downsampling the channel-selected EEG signals to a lowest sampling frequency of the data set for unification;
1-4, removing high and low frequency and alternating current noise; performing basic preprocessing on the EEG signals through a 1-30HZ band-pass filter and a 50HZ notch filter;
1-5, data segmentation and non-physiological signal filtering; dividing the preprocessed data into a plurality of periods with the duration of 1 second, setting 150uV as a threshold value, deleting signal fragments, and removing irrelevant non-physiological signals; each individual retains 100 1 second 16 channel EEG signal segments;
1-6, signal decomposition; the signal is decomposed into delta (1-4 hz), theta (4-8 hz), alpha (8-14 hz), beta (14-30 hz) 4 standard frequency bands using a band pass filter.
3. The method for constructing the children brain electrical function connection map according to the entropy stability criterion of claim 2, wherein the specific flow of the step 3 is as follows:
3-1, extracting global cluster coefficient to measure the functional separation of the brain on the functional connection of the individual level;
Functional separation of the brain refers to the ability to perform specific treatments in closely connected brain region groups; here global cluster coefficients are used to measure the functional separation of the brain; global cluster coefficients The average of the local cluster coefficients for all vertices:
where n is the number of top points, and the calculation formula of the local cluster coefficient C i is as follows:
A ij is an element of the ith row and the jth column of the individual binary connection matrix A, namely a connection weight between a node v i and a node v j; and k is calculated as follows:
3-2, extracting feature path length metric brain function integration on individual level functional connections;
Functional integration of the brain refers to the ability to quickly integrate professional information from different brain regions; here, the functional integration of the brain is measured using the characteristic path length, and the calculation formula of the characteristic path length l G is as follows:
Wherein d (v i,vj) is the shortest distance between nodes v i,vj;
3-3, respectively analyzing global clustering coefficients and characteristic path length distribution differences among different age groups by utilizing a Wilcoxon rank sum test method; averaging p values obtained by rank sum detection to serve as a frequency band importance index;
3-4, extracting the centrality characteristic on the age group level functional connection;
The centrality is the most direct measurement index for describing node centrality in network analysis; the greater the degree of a node means the higher the centrality of the node, the more important the node is in the network; the calculation formula of the node degree centrality is as follows:
where k i represents the number of edges connected to node v i; n-1 represents the number of edges that node v i connects with other nodes;
3-5, superposing the node degree centrality under each standard frequency band, calculating the variance of the node degree centrality, and finally normalizing; the calculation formula is as follows:
Wherein v a,b denotes a degree distribution vector in the b-th band of the a-th age group; the degree of imbalance of the centrality distribution of the brain function network at the cross-frequency level was measured by v a,b.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210094726.0A CN114305451B (en) | 2022-01-26 | 2022-01-26 | Construction method of children brain electrical function connection map of entropy stability criterion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210094726.0A CN114305451B (en) | 2022-01-26 | 2022-01-26 | Construction method of children brain electrical function connection map of entropy stability criterion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114305451A CN114305451A (en) | 2022-04-12 |
| CN114305451B true CN114305451B (en) | 2024-04-23 |
Family
ID=81028536
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210094726.0A Active CN114305451B (en) | 2022-01-26 | 2022-01-26 | Construction method of children brain electrical function connection map of entropy stability criterion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114305451B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7433732B1 (en) * | 2004-02-25 | 2008-10-07 | University Of Florida Research Foundation, Inc. | Real-time brain monitoring system |
| CN109524112A (en) * | 2018-12-26 | 2019-03-26 | 杭州电子科技大学 | A kind of brain function network establishing method orienting coherent method based on part |
| CN110859614A (en) * | 2019-11-22 | 2020-03-06 | 东南大学 | Mathematical astronomical analysis method for teenager brain network based on weighted phase lag index |
| CN112914587A (en) * | 2021-02-18 | 2021-06-08 | 郑州大学 | Apoplexy rehabilitation assessment model construction method and assessment method based on resting state electroencephalogram signal coherence brain function network |
| CN113576491A (en) * | 2021-07-26 | 2021-11-02 | 深圳市人民医院 | Method and system for automatically analyzing frequency domain characteristics and brain network based on resting EEG |
| KR102322647B1 (en) * | 2021-05-25 | 2021-11-09 | 주식회사 아이메디신 | Method, server and computer program for diagnosing alzheimer's disease dementia using correlation between eeg signals |
-
2022
- 2022-01-26 CN CN202210094726.0A patent/CN114305451B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7433732B1 (en) * | 2004-02-25 | 2008-10-07 | University Of Florida Research Foundation, Inc. | Real-time brain monitoring system |
| CN109524112A (en) * | 2018-12-26 | 2019-03-26 | 杭州电子科技大学 | A kind of brain function network establishing method orienting coherent method based on part |
| CN110859614A (en) * | 2019-11-22 | 2020-03-06 | 东南大学 | Mathematical astronomical analysis method for teenager brain network based on weighted phase lag index |
| CN112914587A (en) * | 2021-02-18 | 2021-06-08 | 郑州大学 | Apoplexy rehabilitation assessment model construction method and assessment method based on resting state electroencephalogram signal coherence brain function network |
| KR102322647B1 (en) * | 2021-05-25 | 2021-11-09 | 주식회사 아이메디신 | Method, server and computer program for diagnosing alzheimer's disease dementia using correlation between eeg signals |
| CN113576491A (en) * | 2021-07-26 | 2021-11-02 | 深圳市人民医院 | Method and system for automatically analyzing frequency domain characteristics and brain network based on resting EEG |
Non-Patent Citations (1)
| Title |
|---|
| 基于电生理的颞叶癫痫脑网络机制研究;马涛;《中国优秀硕士论文全文数据库》;20180215;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114305451A (en) | 2022-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108427929B (en) | Depression identification and analysis system based on resting brain network | |
| CN110859614A (en) | Mathematical astronomical analysis method for teenager brain network based on weighted phase lag index | |
| Laskaris et al. | Exploratory data analysis of evoked response single trials based on minimal spanning tree | |
| CN111956221B (en) | Temporal lobe epilepsy classification method based on wavelet scattering factor and LSTM neural network model | |
| CN112515685B (en) | Multi-channel electroencephalogram signal channel selection method based on time-frequency co-fusion | |
| JP2006095266A (en) | Sensitive state judging method | |
| KR102105002B1 (en) | Method and apparatus for an automatic artifact removal of EEG based on a deep leaning algorithm | |
| CN113017627B (en) | Depression and bipolar disorder brain network analysis method based on two-channel phase synchronization feature fusion | |
| CN110942103A (en) | Training method of classifier and computer-readable storage medium | |
| KR20190033972A (en) | Method and apparatus for an automatic artifact removal of EEG based on a deep leaning algorithm | |
| CN111481192A (en) | Electrocardiosignal R wave detection method based on improved U-Net | |
| CN112634214A (en) | Brain network classification method combining node attributes and multilevel topology | |
| CN113712571A (en) | Abnormal electroencephalogram signal detection method based on Rinyi phase transfer entropy and lightweight convolutional neural network | |
| KR20150029969A (en) | Sensibility classification method using brain wave | |
| CN114305451B (en) | Construction method of children brain electrical function connection map of entropy stability criterion | |
| CN110689029A (en) | Method for determining sparsity of fMRI brain function connection network | |
| CN111931578B (en) | Electroencephalogram identification method based on minimum spanning tree and regional double-layer network | |
| CN112274145A (en) | Method and device for processing near-infrared brain function imaging data and storage medium | |
| CN116027888A (en) | P300 intention recognition method based on PLV dynamic brain function network | |
| Liu et al. | Remove motion artifacts from scalp single channel EEG based on noise assisted least square multivariate empirical mode decomposition | |
| CN113919387A (en) | Electroencephalogram signal emotion recognition based on GBDT-LR model | |
| Wei et al. | Automatic recognition of chewing noises in epileptic EEG based on period segmentation | |
| Anwar et al. | A generic approach for classification of psychological disorders diagnosis using EEG | |
| CN113558636B (en) | Method for classifying dementia degree of Alzheimer disease patient based on musical electroencephalogram permutation entropy | |
| CN114287908A (en) | Brain connection classification method with multiple band convolution fusion |
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 |