CN106485208B - The automatic removal method of eye electrical interference in single channel EEG signals - Google Patents
The automatic removal method of eye electrical interference in single channel EEG signals Download PDFInfo
- Publication number
- CN106485208B CN106485208B CN201610840591.2A CN201610840591A CN106485208B CN 106485208 B CN106485208 B CN 106485208B CN 201610840591 A CN201610840591 A CN 201610840591A CN 106485208 B CN106485208 B CN 106485208B
- Authority
- CN
- China
- Prior art keywords
- eeg signals
- interference
- electrical interference
- single channel
- eye
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention discloses a kind of automatic removal methods of eye electrical interference in single channel EEG signals, the detection of the eye electrical interference section based on MSDW is carried out to the single channel EEG signals after collected single channel EEG signals and reverse phase respectively first, obtain an electrical interference section, then electro-ocular signal estimation is carried out to the single channel EEG signals in each interference section using the eye electricity estimation method based on wavelet transformation, then the electro-ocular signal that estimation obtains is subtracted from single channel EEG signals, EEG signals after obtaining an electrical interference removal, realize the removal of eye electrical interference.The present invention is only needed to can be carried out the detection of electro-ocular signal based on single channel EEG signals and removes an electrical interference, by the improvement to the detection of interference section and electro-ocular signal estimation, is improved electro-ocular signal and is estimated accuracy rate, to improve an electrical interference removal effect.
Description
Technical field
The invention belongs to EEG Processing technical fields, more specifically, are related in a kind of single channel EEG signals
The automatic removal method of eye electrical interference.
Background technique
EEG signals (Electroencephalogram, EEG) are in being generated by brain neurological motion and being present in always
The autonomous potential activity of pivot nervous system, brain activity information rich in are brain research, physiological Study, clinical brain
The important means of medical diagnosis on disease.However, EEG signal has the characteristics that non-stationary height, randomness and nonlinear, and signal is micro-
It is weak, right using special purpose machinery (usual port number is more) or portable brain electric acquisition equipment (few channel is even single pass)
When EEG is acquired, be highly prone to eye electric (Electrooculogram, EOG), myoelectricity (Electromyography, EMG),
The interference of electrocardio (Electrocardiography, EKG), especially eye electrical interference, amplitude is larger, the extraction to EEG signals
Have a great impact with analysis.Therefore, the minimizing technology for being especially eye electrical interference to various interference (artefact) is studied, and is brain always
The major issue in Electric signal processing field.
Current main eye electricity artefact removal technology has two classes: artefact excludes and artefact correction.Artefact exclusion is to include
The brain of the eye electrical interference electric period simply excludes;And artefact correction refers to using various methods elimination eye electricity ingredient to acquired EEG
Influence, which specifically includes that 1) average artefact regression analysis, principle assume that eye electricity electrode and each scalp electrode
Between the coefficient of conductivity it is constant, estimate the coefficient of conductivity using the correlation in eye electric channel and other multiple channels, and from each
Electro-ocular signal is subtracted by the coefficient of conductivity in channel and obtains normal EEG signal;2) blind source separation algorithm (Blind Source
Separation, BSS), refer to and mixed signal is separated in the case where source signal and unknown Transmission system characteristic, will divide
Reconstruction signal can be obtained by the signal after removal artefact again after the artefact ingredient removal separated out, and common algorithm has principal component point
Analyse (Principal component analysis, PCA), independent component analysis (Independent component
Analysis, ICA) and second-order blind identification (Second order blind identification, SOBI) algorithm etc.;3) small
Wave conversion (wavelettransform, WT) method, by wavelet transformation to EEG signal carry out wavelet decomposition and remove artefact at
Point, it then carries out wavelet reconstruction and obtains pure EEG.Average artefact regression analysis needs independent eye electric channel;It is blind
Source separation algorithm requires different channels to have certain distribution in space, and port number is greater than the number of signal source, and key is asked
Topic is how to find out artefact ingredient from decomposing in obtained independent element;Small wave converting method has multi-resolution characteristics, is one
The denoising mode of the good non-stationary signal of kind.
For above-mentioned two classes eye electricity artefact removal technology, artefact elimination technique must have the process of Interference Detection, in this way
It could exclude the brain electric period by eye electrical interference;And detected in artefact alignment technique using interference section, then to detection
To interference section handled, the technology can be minimized in correction course to the distortion of EEG data, therefore, to eye electricity
Interference section, which carries out detection, to be very important.
Summary of the invention
It is an object of the invention to overcome the deficiencies of the prior art and provide eye electrical interference in a kind of single channel EEG signals
Automatic removal method, it is only necessary to can be carried out the detection of electro-ocular signal based on single channel EEG signals and remove an electrical interference, lead to
The improvement to the detection of interference section and electro-ocular signal estimation is crossed, electro-ocular signal is improved and estimates accuracy rate, to improve an electrical interference
Removal effect.
For achieving the above object, in single channel EEG signals of the present invention the automatic removal method of eye electrical interference include with
Lower step:
S1: collected single channel EEG signals are detected using the eye electrical interference section based on MSDW first, are interfered
Section set r1;Then single channel EEG signals are subjected to reverse phase, the eye based on MSDW is used to reverse phase single channel EEG signals
The detection of electrical interference section obtains interference section set r2;Section set r will be interfered1With interference section set r2In interference section
It merges, exists if there is two or more interference sections and be overlapped, be merged into an interference section, obtain final
Interfere section set R;
S2: each interference section in interference section set R that step S1 is obtained, using the eye electricity based on wavelet transformation
Estimation method carries out electro-ocular signal estimation to the single channel EEG signals in the interference section, then subtracts from single channel EEG signals
The electro-ocular signal for going estimation to obtain, the EEG signals after obtaining an electrical interference removal.
The automatic removal method of eye electrical interference in single channel EEG signals of the present invention, first respectively to collected single channel
Single channel EEG signals after EEG signals and reverse phase carry out the detection of the eye electrical interference section based on MSDW, obtain an electrical interference area
Between, eye electricity is then carried out to the single channel EEG signals in each interference section using the eye electricity estimation method based on wavelet transformation
Signal estimation, then subtracts the electro-ocular signal that estimation obtains from single channel EEG signals, the brain electricity after obtaining an electrical interference removal
Signal realizes the removal of eye electrical interference.
The present invention has following technical effect that
1) detection from single channel EEG signals is detected using the eye electrical interference section based on MSDW and obtains an electrical interference area
Between, without using multiple channels EEG and the additional channel EOG, go eye electric for the EEG data that portable brain electric acquires equipment acquisition
Interference brings convenience, and Detection accuracy is high, and electro-ocular signal estimation accuracy rate can be improved;
2) using the eye electrical interference minimizing technology based on wavelet transformation to the interference section that detects carry out the estimation of eye electricity and
Removal, can accurately isolate an electrical interference, according to experimental verification it is found that eye electrical interference removal effect is preferable, and original brain
Electric signal and the EEG signals correlation after removal eye electrical interference are moderate, meet data characteristics.
Detailed description of the invention
Fig. 1 is the specific embodiment process of the automatic removal method of eye electrical interference in single channel EEG signals of the present invention
Figure;
Fig. 2 is the schematic diagram of SDW;
Fig. 3 is that amplitude increases eye electrical interference exemplary diagram;
Fig. 4 is actual capabilities eye electrical interference exemplary diagram;
Fig. 5 is the detection example figure in eye electrical interference of the present invention section;
Fig. 6 is the eye electricity estimation method flow chart in the present embodiment based on wavelet transformation;
Fig. 7 is 6 layers of wavelet decomposition schematic diagram in the present embodiment;
Fig. 8 is the reconstruction signal comparison diagram of small wave converting method and two kinds of control methods in the present embodiment;
Fig. 9 is the eye electrical interference removal comparison diagram of TP9 channel data before driving a car simulator;
Figure 10 is the eye electrical interference removal comparison diagram of TP10 channel data before driving a car simulator;
Figure 11 is the eye electrical interference removal comparison diagram of FP1 channel data before driving a car simulator;
Figure 12 is the eye electrical interference removal comparison diagram of TP9 channel data in driving simulator;
Figure 13 is the eye electrical interference removal comparison diagram of TP10 channel data in driving simulator;
Figure 14 is the eye electrical interference removal comparison diagram of FP1 channel data in driving simulator;
Figure 15 is the eye electrical interference removal comparison diagram of TP9 channel data after driving a car simulator;
Figure 16 is the eye electrical interference removal comparison diagram of TP10 channel data after driving a car simulator;
Figure 17 is the eye electrical interference removal comparison diagram of FP1 channel data after driving a car simulator.
Specific embodiment
A specific embodiment of the invention is described with reference to the accompanying drawing, preferably so as to those skilled in the art
Understand the present invention.Requiring particular attention is that in the following description, when known function and the detailed description of design perhaps
When can desalinate main contents of the invention, these descriptions will be ignored herein.
Embodiment
Fig. 1 is the specific embodiment process of the automatic removal method of eye electrical interference in single channel EEG signals of the present invention
Figure.As shown in Figure 1, the specific steps of the automatic removal method of eye electrical interference include: in single channel EEG signals of the present invention
S101: the eye electrical interference section detection based on MSDW:
Technical solution in order to better illustrate the present invention, first to the eye electrical interference section testing principle based on MSDW into
Row explanation.Eye electrical interference section detection based on MSDW is based on real based on digital filter and rule-based decision system
Existing.Filter can be described with following formula (1):
F (t)=S (t)-S (t- | W |) (1)
Wherein, | W | indicate that the width of sliding window, S (t) indicate original signal in the sampled value of moment t.
Because formula (1) derives from formula (2):
Therefore filter can be referred to as " single sliding window first derivative summation (summation of first
derivatives in a sliding window,SDW)”。
Fig. 2 is the schematic diagram of SDW.As shown in Fig. 2, when the size of sliding window, which arrives greatly, can cover certain waveform, it is former
First derivative summation of the beginning signal under the sliding window can filter out the waveform, be closely sized to target eye when sliding window
When the half of electrical waveform size, filter can extract target waveform and inhibit other waveforms less than target waveform.For
Shaped like the eye electrical interference in Fig. 2, when the size of sliding window is the half of an electrical interference width, SDW is in this electrical interference
Maximum value and minimum value are obtained respectively at peak point and the last one point, therefore, can use local minimum and local maximum
Difference and a preset threshold value comparison, if difference value be greater than threshold value if determine eye electrical interference presence.Use MaxiAnd MiniPoint
Not Biao Shi i-th of local maximum and i-th of local minimum in a SDW timing, then, estimate the time range of interference
δ is as follows:
δ={ [T (Maxi)-|W|,T(Mini)]|Maxi-Mini>θ} (3)
Wherein, T (Maxi) and T (Mini) respectively indicate i-th of local maximum MaxiWith local minimum M iniWhen
It carves, | W | indicate the size of sliding window, θ indicates that preset threshold, the threshold value are rule of thumb carried out with the stock size of eye electrical interference
Value.
It is not difficult to find out that SDW filter emphasizes the peak value in special Wave is wide according to formula (1) and Fig. 2.Due to eye electrical interference
There is the diversity of its own, and in order to emphasize the peak value under different waves are wide, it is contemplated that the SDW filtering under a variety of window sizes
Device selects optimal value as the window size of moment t from different windows size, because this process is needed using multiple
Sliding window is called " multiwindow first derivative sums (Multiple-window SDW, MSDW) ".
By eye electrical interference section detecting method of the analysis based on MSDW it is found that it is suitable for the eye electricity that amplitude increases to do
It disturbs.Fig. 3 is that amplitude increases eye electrical interference exemplary diagram.As shown in figure 3, eye electrical interference is the waveform in figure in dotted line frame.And according to
Actual experiment discovery, the variation of electro-ocular signal are not only situation shown in Fig. 2.Fig. 4 is actual capabilities eye electrical interference exemplary diagram.Such as
Shown in Fig. 4, the EEG signal when blink occurs is it is possible that such as four kinds of variations in dotted line frame, wherein after a class signal first rises
Drop, b class signal first drops to be risen afterwards, and c class signal is fallen after rising to be risen again, and d class signal first drops to rise afterwards to drop again.Fig. 5 is that eye electricity of the present invention is dry
Disturb the detection example figure in section.As shown in figure 5, the present invention is using such as in order to detect various types of electrical interference sections
Lower detection method:
Collected single channel EEG signals are detected using the eye electrical interference section based on MSDW first, obtain interference section
Set r1, can detecte the eye electrical interference of a type and c type shown in Fig. 4 in this way.Then single channel EEG signals are carried out
Reverse phase is detected reverse phase single channel EEG signals using the eye electrical interference section based on MSDW, obtains interference section set r2.So
Section set r will be interfered afterwards1With interference section set r2In interference section merge, if there is more than two interference ranges
Between exist overlapping, be merged into an interference section, after merging interfere section endpoint be overlapped interfere section minimum
Starting point and maximum terminating point, obtain final interference section set R.
Based on MSDW eye electrical interference section detection method may refer to bibliography " Chang, W.-D., et al.,
Detection of eye blink artifacts from single prefrontal channel
electroencephalogram.Computer Methods and Programs in Biomedicine,2016.124:
P.19-30. ", specific method can be summarized are as follows: remember that the sampled value of current single channel EEG signals moment t uses [Wmin,Wmax]
Each different sliding window size W calculates SDW value F in range|W|。WminAnd WmaxAccording to the one of sample frequency and eye electrical interference
As duration setting, usually empirical value.From all SDW value F|W|Middle selection maximum value is denoted as RF (t), corresponding window size
It is denoted as | WRF(t)|, judge whether to meet following two condition simultaneously:
A. [t, t- | WRF(t)|] local maximum in range is identical with the number of minimum value;
B. [t, t- | WRF(t)|+1] all first derivatives in range should belong in S ' (t- | WRF(t)|+1) and
Between S ' (t), wherein S ' (t) and S ' (t- | WRF(t)|+1) single channel EEG signals are respectively represented in moment t and moment t- |
WRF(t)|+1 first derivative;
If sliding window size corresponding to maximum value RF (t) is unique and sliding window size meets two above simultaneously
Condition enables | WMSDW(t)|=| WRF(t)|, if correspondence sliding window size is not unique and has multiple sliding window sizes while expiring
Sufficient two above condition enables | WMSDW(t)|=min (| WRF(t)|);If the corresponding all sliding window sizes of maximum value RF (t)
Two above condition can not be met simultaneously, then enabled | WMSDW(t)|=Wmin。
Therefore the calculation formula in section is interfered when moment t are as follows:
Wherein, j is a nonnegative integer, so that Maxi-j-Mini> θ and T (Maxi-j)-T(Mini)≤Wmax, T (Maxi-j)
At the time of indicating that the i-th-j local maximums correspond to.If j is not present, there is no interference by moment t, that is, interference range is not present
Between;IfHave with other sections Chong Die, deletes the interference section of moment t;If moment t
Interference section be contained in another interference section, then delete the interference section of moment t.
S102: the estimation of eye electricity and removal are carried out based on wavelet transformation:
Each interference section in interference section set R that step S101 is obtained, using the eye electricity based on wavelet transformation
Estimation method carries out electro-ocular signal estimation to the single channel EEG signals in the interference section, then subtracts from single channel EEG signals
The electro-ocular signal for going estimation to obtain, the EEG signals after obtaining an electrical interference removal.
Wavelet analysis is a kind of Time-Frequency Analysis Method of variable resolution, has good time domain, frequency domain resolution and adaptation
Property.In recent years wavelet analysis be widely used in signal processing, pattern-recognition, data compression, denoising, feature extraction, classification and
The fields such as prediction.Since normal brain wave frequency is higher, amplitude is smaller, energy spectrum relative distribution, and electro-ocular signal have it is low
Frequently the features such as (< 5Hz), high-energy, limited time and energy Relatively centralized, reconstructing method is reasonably selected, can be achieved with single channel
The eye of EEG signals is electrically separated.
Single channel EEG signals by eye electrical interference can be represented simply as:
X (n)=S (n)+A (n) (5)
Wherein, X (n) is one section of single channel EEG signals by eye electrical interference, and S (n) is normal EEG signals, and A (n) is eye
Electric signal.Since electro-ocular signal is low frequency signal, there is amplitude more higher than normal EEG signals and more smooth waveform, because
This, electro-ocular signal, which extracts problem, may be considered one smooth function A ' (n) similar with X (n) low frequency characteristic of searching as eye
The estimation signal of electric signal A (n), and keep S ' (n)=X (n)-A ' (n) average variance as small as possible.Due to orthogonal multiresolution
Wavelet basis in analysis is the unconditional orthogonal basis in the space Banach, so the decaying of wavelet coefficient can make the mould of reconstruction of function
Original constant times are increased to, wherein the decaying of detail wavelet coefficients can make reconstruction of function more more smooth than original function.In view of eye
Electric signal Wavelet Energy Spectrum Relatively centralized is in low frequency region, therefore the detail wavelet coefficients for eye electricity part of reasonably decaying can obtain
Signal A ' (n) is estimated to eye electricity.
In order to be better achieved eye electricity estimation, the present embodiment to the particular technique means in wavelet transformation carried out screening and
It improves.Fig. 6 is the eye electricity estimation method flow chart in the present embodiment based on wavelet transformation.As shown in fig. 6, being based in the present embodiment
The specific steps of the eye electricity estimation method of wavelet transformation include:
S601: EEG signals between interception interference range:
Note interference section is [start, end], and it is corresponding in EEG signals that start and end respectively indicate interference section
Initial position and rest position need to intercept out the signal in corresponding section from single channel EEG signals.In view of wavelet transformation
The influence of end effect, in intercept signal need to interference section carry out certain extension, i.e., will interference section [start,
End] it is extended to [start-K, end+K], K indicates that growth data number, K > 0, size are arranged according to actual needs, this implementation
K=36 in example.Then from the signal intercepted out in single channel EEG signals in section [start-K, end+K].
S602:Mallat wavelet decomposition:
Mallat wavelet decomposition is carried out to the EEG signals that step S601 is intercepted.The key of Mallat wavelet decomposition is asked
Topic is to select suitable wavelet basis function and determines the number of plies of wavelet decomposition.It is since human eye ball can be considered as a cornea end
When anode, the bipolarity sphere that retina end is cathode, Rotation of eyeball and blink, the Potential distribution near eyeball can be changed and formed
Complicated electro-ocular signal.Studies have shown that typical blink eye electrical waveform shows as continuing the single-phase of about 100~500ms (< 5Hz)
It deviates (quarter-phase shift may also occur in recording mode difference), in order to obtain the smooth approximation of electro-ocular signal, is selected in the present embodiment
There is higher similitude, and the sym5 wavelet basis function with good symmetry and slickness with it, so that decomposition is sparse,
And reduce phase loss.Sym5 wavelet basis function illustrate and methods for using them may refer to document " Wu Mingquan, Li Haifeng,
And Ma Lin, the automatic separation method electronics and information journal of eye electrical interference, 2015 (02) " in single channel EEG signals.
And in terms of the number of plies, the wavelet decomposition number of plies is excessive, more local eye information can be lost through threshold process, and divide
It is less to solve the number of plies, and can make to reconstruct in eye electricity and be mixed into excessive EEG signals.Through experiments, it was found that the number of plies can in the range of 5~8
It is better to achieve the effect that.6 layers of wavelet decomposition (i.e. level=6) are used in the present embodiment, are effectively balanced to small wavelength-division
Solve number of plies requirement different from signal detail.Fig. 7 is 6 layers of wavelet decomposition schematic diagram in the present embodiment.As shown in fig. 7, firstly, will
Original EEG signals X comprising eye electricity is as the bottom;Then bottom-up successively to decompose this layer of approximation coefficient cAk(k is the number of plies
0≤k≤level), obtain one layer of approximation coefficient cAk+1With detail coefficients cDk+1;Wavelet decomposition structure [C, L] is finally obtained,
Its expression signal X is 6 in Decomposition order, and wavelet basis function is the wavelet decomposition structure under " sym5 ", and wherein C indicates wavelet decomposition
Vector (wavelet decomposition vector), L indicate record vector (bookkeeping vector).
S603: wavelet coefficient threshold is determined:
Wavelet coefficient threshold is adaptively determined using Birg é-Massart strategy.
Wavelet coefficient extract it is accurate whether be influence eye electricity reconstruction quality key factor, be usually taken in whole reservations
On the basis of the approximation coefficient of top layer, every layer retains the detail coefficients that mould is greater than specified value, and mould is less than the thin of defined threshold
The method of coefficient zero setting is saved to select reconstruct wavelet coefficient.Wavelet coefficient threshold is determined, Birg é-is used in the present embodiment
Massart strategy is adaptively chosen, and specific method is divided into following three aspects:
1. retaining whole approximation coefficients of top J+1, wherein length is sought in J=length (L) -2, length () expression;
2. determining that the number of kth layer (1≤k < J) retention factor is nj:
Wherein, m value range is L (1)≤m≤2L (1), wherein L (1) indicates the number of top approximation coefficient, and α takes
Value range is 2≤α≤3, α=2 in the present embodiment.
3. according to the retention factor number n of kth layerk, the threshold value of kth layer is set as n-thkBig coefficient module | cDk|。
S604: wavelet reconstruction:
EEG signals after being decomposed according to the wavelet coefficient threshold determined in step S603 to step S602 carry out small echo weight
Structure is restored, the practice after reconstitution due to being extended in step s 601 to interference section are as follows: from
Signal in signal after reconstruct in interception interference section [start, end] estimates signal as eye electricity.
In order to illustrate the advantage of wavelet transformation employed in the present embodiment, as a comparison using two kinds of wavelet transformation modes
Method is compared with the present embodiment small wave converting method, and wherein method 1 is using reconstruct approximation coefficient as eye electrical interference
Approximate evaluation, method 2 are to obtain the approximate evaluation of an electrical interference using wavelet reconstruction function.Fig. 8 is that small echo becomes in the present embodiment
Change the reconstruction signal comparison diagram of method Yu two kinds of control methods.Table 1 be in the present embodiment small wave converting method and two kinds to analogy
The reconstruction signal of method and original signal correlation.
Table 1
As shown in Fig. 8 and table 1, using the present invention and two kinds of control methods to the list interfered comprising 4 seed type electro-ocular signals
Channel EEG signals are reconstructed, and the reconstruction signal of method 1 differs too big with original signal, and the reconstruction signal of method 2
The detailed information (including noise etc.) of original signal is remained too much, and is had the reconstruction signal that the present embodiment method obtains only and existed
It is approximate with original signal in shape, but also do not bring many noises into, the estimated value is smoother.Therefore the present embodiment method
Employed in wavelet transformation be a kind of more preferred mode.
After estimation obtains electro-ocular signal, needs to subtract the electro-ocular signal that estimation obtains from single channel EEG signals and obtain eye
EEG signals after electrical interference removal.Discontinuous point may be formed at endpoint by executing the operation.It, can be in order to avoid this problem
Endpoint compensation is carried out to the signal after removal eye electrical interference, method particularly includes: for the EEG signals after eye electrical interference removal, In
The endpoint location of each electro-ocular signal interferes the endpoint of section [start, end], using its own sampled value and two sides Q
Sampled value carries out median filtering, using median filtering value as the sampled value of the endpoint location.Q=2 in the present embodiment, i.e., to 5
Sampled value carries out median filtering.
In order to more preferably illustrate technical effect of the invention, simulating, verifying is carried out to the present invention using specific embodiment.This
The data when data that simulating, verifying uses are the driving simulators for using Muse to acquire, it is (left that Muse can acquire TP9
Ear), the EEG data of FP1 (left front volume), FP2 (right forehead) and TP10 (auris dextra) this four channels, sample frequency 220Hz.This
Secondary simulating, verifying uses the data in tri- channels TP9, FP1 and TP10, each channel data duration 60s (60x220=13200
Data) it is handled.It will show below using context of methods to EEG data under different automobile simulator driving condition difference channels
Eye electrical interference removal effect.
Fig. 9 is the eye electrical interference removal comparison diagram of TP9 channel data before driving a car simulator.Figure 10 is driving mould
The eye electrical interference of TP10 channel data removes comparison diagram before quasi- device.Figure 11 is the eye of FP1 channel data before driving a car simulator
Electrical interference removes comparison diagram.
Figure 12 is the eye electrical interference removal comparison diagram of TP9 channel data in driving simulator.Figure 13 is driving
The eye electrical interference of TP10 channel data removes comparison diagram in simulator.Figure 14 is FP1 channel data in driving simulator
Eye electrical interference removes comparison diagram.
Figure 15 is the eye electrical interference removal comparison diagram of TP9 channel data after driving a car simulator.Figure 16 is driving
The eye electrical interference of TP10 channel data removes comparison diagram after simulator.Figure 17 is FP1 channel data after driving simulator
Eye electrical interference removes comparison diagram.
It is can be seen that from Fig. 9 to Figure 17 using method proposed in this paper, can accurately be found most of in EEG
Eye electrical interference section, and the eye electrical interference found out is able to carry out and is relatively accurately estimated, and can be effectively removed.
Table 2 is a correlation contrast table for electrical interference removal front and back different data.
Table 2
It shown in table 2 as above, has calculated separately under different automobile simulator driving condition difference channels, X (n) and noise, S '
(n) with the correlation of noise and X (n) and S ' (n), wherein X (n) indicates that the raw EEG data of Muse acquisition, noise indicate
The eye electrical interference (merging that electrical interference signal is returned in each interference section) obtained using present invention estimation, S ' (n) is indicated using this
EGG data after invention eye electrical interference removal.Because X (n) contains eye electrical interference information, therefore X (n) is compared with the correlation of noise
Height, and S ' (n) is data of the X (n) after the removal eye electrical interference that the method for the present invention obtains, therefore S ' (n) is related to noise's
Property should be relatively low, from table 2 it is not difficult to find that the average correlation of X (n) and noise reach 0.8042, and S ' (n) and noise
Average correlation is down to 0.1029, this illustrates that the method for the present invention can be effectively removed an electrical interference.Since S ' (n) is X (n) warp
Past eye electrical interference obtains, and the correlation between them is certainly less than 1, but X (n) is that an electrical interference is doped in EEG signal,
Main information or EEG in X (n), therefore the correlation of S ' (n) and the X (n) obtained after removing eye electrical interference should will not be very
Low, as known from Table 2, the average correlation of X (n) and S ' (n) are 0.5460, meet theory analysis.
Complex chart 9 to Figure 17 and table 2 it is found that using in single channel EEG signals proposed in this paper eye electrical interference it is automatic
Minimizing technology preferably can be separated and be removed to eye electrical interference.
Although the illustrative specific embodiment of the present invention is described above, in order to the technology of the art
Personnel understand the present invention, it should be apparent that the present invention is not limited to the range of specific embodiment, to the common skill of the art
For art personnel, if various change the attached claims limit and determine the spirit and scope of the present invention in, these
Variation is it will be apparent that all utilize the innovation and creation of present inventive concept in the column of protection.
Claims (6)
1. the automatic removal method of eye electrical interference in a kind of single channel EEG signals, it is characterised in that the following steps are included:
S1: collected single channel EEG signals are detected using the eye electrical interference section based on MSDW first, signal is obtained and first rises
Drop and signal are fallen after rising afterwards rises the interference section set r of two types eye electrical interference again1;Then by single channel EEG signals into
Row reverse phase, to reverse phase single channel EEG signals using based on MSDW eye electrical interference section detect, obtain signal first drop rise afterwards with
Signal first drops rises the interference section set r for dropping two types eye electrical interference again afterwards2;Section set r will be interfered1With interference Interval Set
Close r2In interference section merge, exist if there is two or more interference sections and be overlapped, be merged into one it is dry
Section is disturbed, final interference section set R is obtained;
S2: each interference section in interference section set R obtained to step S1 is estimated using the eye electricity based on wavelet transformation
Method carries out electro-ocular signal estimation to the single channel EEG signals in the interference section, then subtracts and estimates from single channel EEG signals
Obtained electro-ocular signal is counted, the EEG signals after obtaining an electrical interference removal.
2. the automatic removal method of eye electrical interference in single channel EEG signals according to claim 1, which is characterized in that institute
The eye electrical interference section detection based on MSDW stated method particularly includes:
Remember that the sampled value of current single channel EEG signals moment t uses [Wmin,Wmax] each different sliding window ruler in range
Very little W calculates SDW value F|W|;WminAnd WmaxThe empirical value being arranged according to the duration of sample frequency and eye electrical interference;From all
SDW value F|W|Middle selection maximum value is denoted as RF (t), and corresponding window size is denoted as | WRF(t)|, judge whether to meet following two simultaneously
Condition:
A. [t, t- | WRF(t)|] local maximum in range is identical with the number of minimum value;
B. [t, t- | WRF(t)|+1] all first derivatives in range should belong in S ' (t- | WRF(t)|+1) and with S '
(t) between, wherein S ' (t) and S ' (t- | WRF(t)|+1) single channel EEG signals are respectively represented in moment t and moment t- | WRF(t)|
+ 1 first derivative;
If sliding window size corresponding to maximum value RF (t) is unique and sliding window size meets two above item simultaneously
Part enables | WMSDW(t)|=| WRF(t)|, if corresponding sliding window size is not unique and has multiple sliding window sizes while meeting
Two above condition enables | WMSDW(t)|=min (| WRF(t)|);If the corresponding all sliding window sizes of maximum value RF (t) are equal
Two above condition can not be met simultaneously, then enabled | WMSDW(t)|=Wmin;
The calculation formula in section is interfered when moment t are as follows:
Wherein, j is a nonnegative integer, so that Maxi-j-Mini> θ and T (Maxi-j)-T(Mini)≤M, T (Maxi-j) indicate the
At the time of i-j local maximum corresponds to;If j is not present, there is no interference by moment t;IfHave with other sections Chong Die, deletes the interference section of moment t;If the interference of moment t
Section is contained in another interference section, then deletes the interference section of moment t.
3. the automatic removal method of eye electrical interference in single channel EEG signals according to claim 1, which is characterized in that institute
The specific steps for stating the eye electricity estimation method in step S2 based on wavelet transformation include:
S2.1: note interference section is [start, end], and it is corresponding in EEG signals that start and end respectively indicate interference section
Interference section [start, end] is extended to [start-K, end+K] by initial position and rest position, and K indicates growth data
Number, K > 0, size is arranged according to actual needs, is intercepted out in section [start-K, end+K] from single channel EEG signals
Signal;
S2.2: Mallat wavelet decomposition is carried out to the EEG signals that step S2.1 is intercepted;
S2.3: wavelet coefficient threshold is adaptively determined using Birg é-Massart strategy;
S2.4: the EEG signals after being decomposed according to the wavelet coefficient threshold determined in step S2.3 to step S2.2 carry out small echo weight
Structure estimates signal from the signal in the signal after reconstruct in interception interference section [start, end] as eye electricity.
4. the automatic removal method of eye electrical interference in single channel EEG signals according to claim 3, which is characterized in that institute
It states Mallat wavelet decomposition in step S2.2 and uses sym5 wavelet basis function.
5. the automatic removal method of eye electrical interference in single channel EEG signals according to claim 3, which is characterized in that institute
The number of plies value range for stating Mallat wavelet decomposition in step S2.2 is 5~8.
6. the automatic removal method of eye electrical interference in single channel EEG signals according to claim 1, which is characterized in that right
EEG signals in step S2 after eye electrical interference removal also need to carry out endpoint compensation, method particularly includes: in each telecommunications
Number endpoint location, median filtering is carried out using Q sampled value of its own sampled value and two sides, using median filtering value as the end
The sampled value of point position.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610840591.2A CN106485208B (en) | 2016-09-22 | 2016-09-22 | The automatic removal method of eye electrical interference in single channel EEG signals |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610840591.2A CN106485208B (en) | 2016-09-22 | 2016-09-22 | The automatic removal method of eye electrical interference in single channel EEG signals |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106485208A CN106485208A (en) | 2017-03-08 |
CN106485208B true CN106485208B (en) | 2019-10-29 |
Family
ID=58267591
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610840591.2A Active CN106485208B (en) | 2016-09-22 | 2016-09-22 | The automatic removal method of eye electrical interference in single channel EEG signals |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106485208B (en) |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108836321A (en) * | 2018-05-03 | 2018-11-20 | 江苏师范大学 | A kind of EEG signals preprocess method based on adaptive noise cancel- ation system |
CN109157214A (en) * | 2018-09-11 | 2019-01-08 | 河南工业大学 | A method of the online removal eye electricity artefact suitable for single channel EEG signals |
CN109820503A (en) * | 2019-04-10 | 2019-05-31 | 合肥工业大学 | The synchronous minimizing technology of a variety of artefacts in single channel EEG signals |
CN110236532A (en) * | 2019-04-30 | 2019-09-17 | 深圳和而泰家居在线网络科技有限公司 | Processing of bioelectric signals method, apparatus, computer equipment and storage medium |
CN111685759B (en) * | 2020-05-12 | 2022-09-09 | 河北大学 | P, T characteristic wave detection method of electrocardiosignals |
CN111956217B (en) * | 2020-07-15 | 2022-06-24 | 山东师范大学 | Blink artifact identification method and system for real-time electroencephalogram signals |
CN114041803B (en) * | 2021-11-16 | 2022-10-14 | 电子科技大学 | TMS pulse artifact removing device based on interval median filtering |
CN117064409B (en) * | 2023-10-12 | 2024-01-30 | 深圳般意科技有限公司 | Method, device and terminal for evaluating transcranial direct current intervention stimulation effect in real time |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101869477A (en) * | 2010-05-14 | 2010-10-27 | 北京工业大学 | Self-adaptive EEG signal ocular artifact automatic removal method |
CN104688220A (en) * | 2015-01-28 | 2015-06-10 | 西安交通大学 | Method for removing ocular artifacts in EEG signals |
CN104809434A (en) * | 2015-04-22 | 2015-07-29 | 哈尔滨工业大学 | Sleep staging method based on single-channel electroencephalogram signal ocular artifact removal |
-
2016
- 2016-09-22 CN CN201610840591.2A patent/CN106485208B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101869477A (en) * | 2010-05-14 | 2010-10-27 | 北京工业大学 | Self-adaptive EEG signal ocular artifact automatic removal method |
CN104688220A (en) * | 2015-01-28 | 2015-06-10 | 西安交通大学 | Method for removing ocular artifacts in EEG signals |
CN104809434A (en) * | 2015-04-22 | 2015-07-29 | 哈尔滨工业大学 | Sleep staging method based on single-channel electroencephalogram signal ocular artifact removal |
Non-Patent Citations (3)
Title |
---|
An unsupervised eye blink artifact detection method for real-time electroencephalogram processing;Won-Du Chang et al.;《Institute of Physics and Engineering in Medicine》;20160219;第401-417页 * |
Detection of eye blink artifacts from single prefrontal channel electroencephalogram;Won-Du Chang et al.;《Computer Methods and Programs in Biomedicine》;20160229;第21-22页 * |
单通道脑电信号中眼电干扰的自动分离方法;吴明权 等;《电子与信息学报》;20150228;第37卷(第2期);第367、369-370页 * |
Also Published As
Publication number | Publication date |
---|---|
CN106485208A (en) | 2017-03-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106485208B (en) | The automatic removal method of eye electrical interference in single channel EEG signals | |
Merah et al. | R-peaks detection based on stationary wavelet transform | |
Urigüen et al. | EEG artifact removal—state-of-the-art and guidelines | |
Chang et al. | Detection of eye blink artifacts from single prefrontal channel electroencephalogram | |
Iriarte et al. | Independent component analysis as a tool to eliminate artifacts in EEG: a quantitative study | |
CN112690793B (en) | Emotion electroencephalogram migration model training method and system and emotion recognition method and equipment | |
Milanesi et al. | Independent component analysis applied to the removal of motion artifacts from electrocardiographic signals | |
Escudero et al. | Quantitative evaluation of artifact removal in real magnetoencephalogram signals with blind source separation | |
Bono et al. | Hybrid wavelet and EMD/ICA approach for artifact suppression in pervasive EEG | |
Guarnieri et al. | Online EEG artifact removal for BCI applications by adaptive spatial filtering | |
Sameni et al. | An iterative subspace denoising algorithm for removing electroencephalogram ocular artifacts | |
CN107260166A (en) | A kind of electric artefact elimination method of practical online brain | |
Ashokkumar et al. | RETRACTED: Implementation of deep neural networks for classifying electroencephalogram signal using fractional S‐transform for epileptic seizure detection | |
Kumar et al. | Detection and classification of ECG noises using decomposition on mixed codebook for quality analysis | |
CN104305992B (en) | A kind of interactive fast automatic extracting method of Fetal ECG | |
Dasgupta et al. | A two-stage framework for denoising electrooculography signals | |
Mohamed et al. | Towards automated quality assessment measure for EEG signals | |
Sheoran et al. | Methods of denoising of electroencephalogram signal: A review | |
Ranjan et al. | Cardiac artifact noise removal from sleep EEG signals using hybrid denoising model | |
Shahbakhti et al. | SWT-kurtosis based algorithm for elimination of electrical shift and linear trend from EEG signals | |
Dora et al. | Efficient detection and correction of variable strength ECG artifact from single channel EEG | |
Roy et al. | A NLMS based approach for artifacts removal in multichannel EEG signals with ICA and double density wavelet transform | |
CN108245154B (en) | Method for accurately determining blink interval in electroencephalogram or electrooculogram by using abnormal value detection | |
Luengo et al. | Blind analysis of atrial fibrillation electrograms: a sparsity-aware formulation | |
Kaushal et al. | Better approach for denoising EEG signals |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | 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 | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20210219 Address after: 311100 401v, building 7, 22 Xinyan Road, Donghu street, Yuhang District, Hangzhou City, Zhejiang Province Patentee after: Hangzhou wowei Medical Technology Co.,Ltd. Address before: No. 505, building 6, Zone D, Tianfu Software Park, No. 599, shijicheng South Road, high tech Zone, Chengdu, Sichuan 610015 Patentee before: XIAOCAIER CHENGDU INFORMATION TECHNOLOGY Co.,Ltd. Patentee before: Chongqing University |