腦機接口與認知計算實驗報告2

1、哈爾濱工業(yè)大學計算機科學與技術學院實驗課程名稱：腦機接口與認知計算指導教師：課程類型：限選課實驗項目名稱：視覺實驗實驗題目：視覺實驗班級：1203106學號：1120310612：設計成績成績指導老師一、 實驗目的使用命令行對腦電數(shù)據(jù)進行預處理，得到去除眼電、市電和低頻成分干擾的腦電信號數(shù)據(jù)二、實驗要求與環(huán)境EEGLAB 13.4.42014b認知實驗中的 64 通道腦電數(shù)據(jù)Windows 7 PC三、實驗過程載入數(shù)據(jù)編輯電極信息數(shù)據(jù)濾波參考轉換ICA 眼電去除數(shù)據(jù)分段和移除基線偽跡去除繪制電極頻譜圖數(shù)據(jù)的疊加平均及成像 在同一坐標軸上繪制所有，并繪制腦地形圖繪制陣列圖成像腦地形圖序列Time

2、/Frequency 分解四、實驗結果 eeg = pop_chanedit(readlocs(location.ced);readlocs(): chanedit format amed from fileextenReading file (lines): 10 20 30 40 50 60 67readlocs() warning: Fewer columnshe inputn expected.See help readlocs EEG = pop_eegfiltneG, 0, 45);pop_eegfiltnew() pop_eegfiltnew() pop_eegfiltnew()

3、 pop_eegfiltnew() pop_eegfiltnew()-performing 295 polossfiltering. Hztransition band width: 11.25 passband edge(s): 45 Hzcutorequency(ies) (-6 dB): 50.625 Hzfiltering the data (zero-phase)firfilt(): |=| 100%, ETE 00:00 eeglab警告: 局部變量的值可能已發(fā)生更改，以便與全局變量匹配。以后版本的將要求您在使用變量之前將該變量 In eeg_global at 35 In eeg

4、lab at 328為全局變量。eeglab: EEGLAB: EEGLAB:options file is C:Usersmyasuseeg_options.m adding dipfit v2.3 (see help eegplugin_dipfit)adding firfilt v1.6.1 (see help eegplugin_firfilt)Could not check for the latest EEGLAB ver disconnected).(ernet maybeTo prevent long startup time, disable checking for ne

5、(FIle Memory and other options).GLAB vereeg_checksetnote:uppertimelimit(xmax)adjustedso(xmax-xmin)*srate+1 = number of frameseeg_checkset note: creating the original event table (EEG.urevent) Creating a new ALLEEG dataset 1Done. EEG = pop_reref( EEG, 28);EEG = pop_runica(EEG,extended,1, descending o

6、rdererupt,on);Sorting Scaling Scalingcomponents components componentsin to toofmeanprojectedvariance.RMS RMSmicrovolt microvoltRe-referencing dataScaling components Re-referencing ICA Scaling componentstoRMSmicrovoltmatrix to RMSmicrovoltRe-referencing dataScaling components Re-referencing ICA Scali

7、ng componentsto RMS matrixmicrovolttoRMSmicrovoltEEG =setname: filename: filepath: subject: group: condition:CNT file Original file: C:UsersmyasusDesktop腦機接口ses:comments: tnbchan:63trials: pnts: srate: xmin: xmax: times: data: icaact: icawinv: icasphere:icaweights: icachansind: chanlocs: urchanlocs:

8、 chaninfo:ref: event: urevent:eventdescription:epoch: epochdescription:reject:11020120100001.0201e+031x1020120 double 63x1020120 single 63x65 63x63 65x63double double double1x63 double 1x63 struct 1x1 struct common 1x121 struct 1x121 struct 1x1 1x1 struct structss:specdata: specicaact: splinefile: i

9、casplinefile:dipfit: history:EEG =t(C:UsersmyasusDesktop腦機接口2015-qile.c.saved:etc: datfile:no1x1 struct pop_topoplot(EEG,0, 1:32 ,題圖,4 8 ,0,electrodes,on); EEG =p( EEG, 4, 0);Computing projection . EEG = pop_epoch( EEG, 11, 01000);pop_epoch():10 Epoching.epochectedWarning: Warning: Warning: Warning:

10、 Warning:event event event event event12345out out out out outof of of of ofdata data data data databoundary boundary boundary boundary boundaryWarning: Warning: Warning: Warning: Warning:event event event event event6789out out out outof of of ofdata data data databoundary boundary boundary boundar

11、y10 out of data boundarypop_epoch():0 epochs generatedpop_epoch(): checking epochs for data discontinuity EEG = pop_eegthresh(EEG,1,1 3 4 5 7 8:32 ,-100,100,-1,1.9922,0,0);Warning : starttime inferior to minimum time, adjusted30 channel selected0/0 trials marked for rejection figure; pop_spectopo(EE

12、G, 1, 0 1000, EEG , percent, 15, freq,6 10 22, freqrange,2 pop_spectopo(): selecting 1 to 1001)Computing spectra (window25,electrodes,off);time range0.00 ms to 1000.00 ms (poslength 256;fft length: 512; overlap 0): NaN); figure; Scalp mapspop_timtopo(EEG, 0 1000,will show latencies:at frames:01p(ALL

13、EEG,1,vg,on,1,2,addavg,on,addst2d,off,subavg,on,diffstd, off, tplotopt, ydir -1);di ,figure;pop_image(EEG,1,1,FPz,10,1,ylabel,muV,on,cbar,on,topo,1EEG.chanlocs EEG.chaninfo );d executed by pop_image:image( mean(EEG.data(1, :),1), ones(1, EEG.trials)*EEG.xmax*1000, linspace(EEG.xmin*1000, EEG.xmax*10

14、00, EEG.pnts), FPz, 10, 1 ,ylabel, muV, , on, cbar, on, topo, 1optchanlocs optchaninfo );image() - Plot a colored image of a collection of single-trial data epochs, optionallysorted on and/or aligned to an input sorting variable and smoothed acrosstrials with a Gaussian weighted moving-average. (To

15、return event-aligned datawithout plotting, use eegalign(). trials on value,litudeOptionally sortor phase within a specified latency window. Optionally meanplot theand std. dev.and moving-window spectrallitude ander-trial coherenceat aselected or peak frequency. Optionally time warp the single trialt

16、ime-dao a series(potential) orer dao align the plottedof events with varying latenciest occur in each trial.Click onindividual figures parts to examine them separa zoom (using axcopy().Usage:y and figure; colored lines in input orderimage(data,times); % image trials as figure; outdata,outvar,outtria

17、ls,limits,axhndls, .,s,cohers,cohsig,sig,outs,.phsangls,phs=,sortidx,sig .image(data,sortvar,times,title,avewidth,decimate,.key, val, .); % useoptionsRequired datainput:= vector or matrix Single-channel input da Formats (1,frames*trials) or (frames,trials)o image.Optionalordered inputs with defaults

18、:sortvar= vector | Variable to sort epochs on (length(sortvar)= nepochs)Exle: sortvar may by subject response time in eachepoch (s)default|: plot in input order= vector | vector of latencies (s) for each epochtimes time po.Else startms ntimes srate = start latency (ms), timepos(=frames)per epoch, sl

19、ing rate (Hz). Else -0:nframes-1default: title= stringPlot title default: noneavewidth = set to boxcaritive scalar (may be non-eger). If avg_type is(the default), this is the number of trials used to smooth(vertically) wimoving-average. If avg_type is settoGaussian, this is the standard deviation (i

20、n unitsoftrials)oftheGaussianwindowusedtosmooth(vertically)wimoving-average.Gaussian window extends threestandard deviations below and three standard deviationsabove windowcenter (trials average). default: nosmoothingdecimate = Factor to decimate| non-eger)window are not incorporatedoolate ntrials b

21、y (may beElse, if this is large ( sqrrials), output thismany epochs.default|0-1Optionaordered keyword,argument pairs:Re-align data epochs:align= latency Time-lock dao sortvar. Plot sortvar at given latency(s). Else Inf - plot sortvar at median sortvarlatencydefault: do not aligntimewarp = events, wa

22、rpms, colors Time warp and phasetime-courses before smoothing. eventswhose columnsspecify the latencies (in ms) at which,litudeis a matrixa series ofsucsive events occurin each trial. warpms is an optional vector oflatencies(s) to whichthe series of events should be time locked. (Note: Epochendshoul

23、d not be declared as events or warpms. If warpmsor ,the median of each events column will be used. colorsstart andis absentcontains alist of marking the occurrenceof the timefor this eventcolumn. If through the given colorlinestyles to use for vertical lineswarped events. If , no line will be drawnf

24、ewer colorsn event columns, cycleslabels.Note: Not compatible with vert (below).renorm = yes|no| formula Normalize sorting variable to epoch latency range and plot. yes= autoscale. formula mustbe a lineartransformationhe format a*x+bExle of If sortingformula: 3*x+2. default: noby string values like

25、event type, suggestedformulas for:letterstring:1000*x,numberstring:30000*x-1500noplot = on|off Do not plot sortvar default: Plot sortvarifimes rangeNoShow = on|off Do not plotimage, simply return outputsdefault: offSort data nosortdata epochsepochs:= on|off Do not sort data epochs. default: Sortbyso

26、rtvar (see yes|no sortvar withsortvar input above)Replace trials with the same value ofreplace_ties =the mean of those trials.Only worksif sorting trialsby sortvar. default: nostartms endms direction Sort data on (mean) valuevalsort=betn startms and (optional) endms. Direction is1 or -1.If -1, plot

27、max-value epoch at bottom default: sorton sortvar phasesortby phase in=ms_centrct freq maxfreq topphase Sort epochsa 3-cycle window centeredatency ms_center (ms).Percentile (prct) in range 0,100 gives percent oftrialsto reject for (too) lowlitude. Else, if in range-100,0,percent of trials to reject

28、for (too) highlitude;freq (Hz) is the phase-sorting frequency. Withoptionalmaxfreq, sort by phase at freq of maxerhedata inthe range freq,maxfreq (Note: phasesort arg freqoverridesthe frequency specified in coher). With optionaltopphase,sort by phase, putting topphase (degrees, in range-180,180)at t

29、he top of the image. Note: phasesort now usescircularsmoothing. Use cycles (below) for wavelet length.default: 0 25 8 13 180sort= center_mst freq maxfreqSort epochs bylitude.(See phasesort above). If ms_center is Inf, thensortingis by meaner across the time window specifiedby sortwinbelow. If third

30、arg, freq, is (its default) 3 for better(below)wifrequencyspecificity,default: plot potential, notlitudes, with nominimum.The averageer(in log space) before time 0 is automatically removed.Notetthebaseline parameter has no effect on plots.Instead usechange baselinedb or parameter. By defaultthe base

31、line is removedbasedB in the limitsbefore time 0.Specify plot parameters:limits hicoher basedB= lotime hitime minmaxlodB hidB locoherPlot axes limits. Can use NaN (or nan, but notNan) for missing itemsand omit late items. Uselast input, basedB, toset thebaseline dBlitudeinplotsplotsdefault: from dat

32、a sortvar_limits = max minimum andum sorting variablevalues to image. This only affects visualizationofimage ands (not smoothing).Cannot beusedif sorting by any factor besides sortvar (e.g., phase).= lo_dB, hi_dB, coher_signif_level Usesignifputed significancethresholds (as from outputs save time. d

33、efault: nonesig,cohsig) tocaxisSet caxis limits at= lo hi Set color axis limits.Elsefraction(+/-)fraction*max(abs(data) in dB, based on data limitsdefault:symmetricalAddbelowepotheeanto plot:= on|off|1|2|3|4 Plottime average of the trialsimage. If on or 1, a single shown.If 2,(the mean ofalltrials)

34、istwos 3, the trials are splittertiles(super and sub median trials) are shown. oIfand their threes are shown.If 4,thetrials are splito quartilesand fours are shown. voltage plotted up, change YDIRto -1 in icadefs.m.If Note, if you want negativealpha option is used, anyvaluesof greatern1 will be rese

35、t to 1. default noplotted(i.e.,alpha = alpha Visualizes two-sided significance threshold a two-tailed test) for thenull hypothesis of a zero mean, symmetric distribution(range:.001 0.1).are value otherThresholds determined vian off.rmuion test. Requires to be aIf is setto a increase plot readability

36、.value greatern 1, it is resetto 1 todefault: no alpha significance thresholds plottedstd= on|off Plot+/- stdev. Requires default:no std. dev. plotted_grid = If _grid is added as an option voltage axis dashedgrid lines will beadded to theplot to facilie judginglitude rmbefore pro= on|off Subtract th

37、e average sing default: nofrom each trialAdd time/frequency information:coher= freq Plot at freq (Hz)averageplus meanlitude & coherenceElse minfrq maxfrq =same, but select frequency withmaxer ingiven range. (Note:the phasesort freq (above)overwrites theseparameters). Else minfrq maxfrq alpha = plot

38、coher. signif. level lineat probability alpha (range: 0,0.1) default: no coher,no alpha levelsrate= freq Specify the data s (if notling rate in Hz for/coherimplicithird arg, times) default: as defined inicadefs.mcycles = float Number of cycles ition default: 3he wavelet time/frequencyAdd plot featur

39、es: cbardefault nocbar_titlefor= on|off Plot color bar to right of-image= string The title for the color bar (e.g., muVmicrovolts).= map,chan_locs,eloc_info Plot a 2-D scalp map image.topo at upper leftofmap may be a singleeger, representing theplotted dhannel,or a vector of scalp map channel values

40、. be a channel locationsfile or a chanlocs structure (EEG.chanlocs).chan_locs maySee topoplot exleeloc_info (EEG.chaninfo), if empty () orabsent,imps the X directionpos towards the nose and all channels areplotteddefault: no scalp mapspec right of image.specaxis frequency scaling,= loHz,hiHz Plot th

41、e mean data spectrum at upper= log|lin Use lin for linear spectrumelse log for log scaling default: log= epochs_vector Plot horizontal lines athorz specified epoch vertnumbers.= times_vector Plot vertical dashed lines atspecified latencies auxvarvariable(s) for each= size(nvars,ntrials) matrix Plot

42、auxiliary trialasseparatetra.Else,auxvar,matrix,colorstringsto specify N trace colors.= r,bo-,k:Ex: colorstrings(see also:vertandtimewarp above).default: nonesortvarpercent = float vector Plot percentiles for the sorting variablefor instance, 0.1 0.5 0.9 plots the 10thpercentile, themedianand the 90

43、th percentile.Plot options: noxlabelbottom x-axis ylabel= on|off Do not plot Time (ms) on the= string muVordinate axis label (default is). PruVwithavg_type used to smooth= boxcar|Gaussian The type of moving averagethe data. Boxcar smoothes the data by simplytakingthe mean ofa certain number of trial

44、s above and below eachtrial.Gaussian does the same butweights thetrialsaccording to a Gaussian distribution (e.g.,nearbytrialsreceive greater weight).The Gaussian isbettern theboxcarhat it rather evenly filters outhigh frequencyvertical components in theimage. Seeavewidth argumentdescription for mor

45、e information. default:boxcarimg_trialax_label = string The label of the axis correspondingto trialsheimage(e.g.,ReactionTime).Note,ifimg_trialax_label is set to somethingbesides Trials or , the tick marks on thisaxis will be set in unitsof the sorting variable. alternative to plotting theThis is a

46、usefulsorting variable when the sorting variable isnotilliseconds. Thisoption is not effective if sorting by phase, or EEG value. default: Trialslitude,img_trialax_ticks = Vector of sorting variable values at which tickmarks (e.g., 300350 400 450for reaction time image. Tick markvaluesshouldbe (e.g.

47、, Trials or msec).sec) will appear on thetrial axis of thegiveninunitsimg_trialax_labelThis option is not effective if sorting by or EEG value.default: automatic= low_boundary high_boundary a time window (in litude ineach trial will be removed from each trial (e.g.,litude, phase,baseline msec) whose

48、 mean-100 0) after filtering.Useful in conjunction with filt option to re-basline trials after they have beenfiltered. Not ne imagesary if data have alreadybeen baselined andpro baseliningsing does not affect baselinelitudedefault: no furtherof data.= low_boundary high_boundary a time window (in lit

49、ude ineach trial will be removed from each trial (e.g.,baselinedbmsec)whose mean-1000). Use basedBin limitsto remove a fixed value. Default is time before to use a0. Ifyou do not wantbaseline forlitude plotting, enter a NaNvalue.filtvector indicating the= low_boundary frequencycut-offs for ahigh_bou

50、ndary a two element3rd order Butterworth filtertwillbe appd to eachtrial of data.If low_boundary=0, the filteris aloss filter.Ifhigh_boundary=srate/2, then the filter is a filter.If bothhighpassboundaries are bet a bandpass filter.n 0 and srate/2, then thefilter isIf both boundaries are bet then the

51、 filter is a bandstopn 0 and -srate/2,filter (with boundaries equal to the absolute values of low_boundary andhigh_boundary).Note, using this option requires the srate option to bespecified and the signal pro function butter.m.You shouldsing toolboxprobably use the baseline option as well since the

52、mean prestimulus baselinemay no longer be 0 after the filter is appddefault: no filteringOptional outputs:outdata=outvar= smoothing).(times,epochsout) data matrix (after smoothing) (1,epochsout) actual values trials are sorted on (afterif sortvarpercent is used, this variable contains acell array with sorted_valuessorted_percent1.sorted_percentNouttrials limitsfrequency (Hz)axhndls=(