Dear Frederic,<div><br></div><div>I see that you have computed the 'noise' (from cfg.lcmv.projectnoise='yes'). You can often use this output to remove the center of sphere artifact to which you refer. Have you tried this (i.e. described here
<a href="http://fieldtrip.fcdonders.nl/tutorial/beamformer#neural_activity_index">http://fieldtrip.fcdonders.nl/tutorial/beamformer#neural_activity_index</a> ) and does this help?<br><div><br></div><div>Also, how much is the rank of the covariance matrix (i.e. cov_data.cov) reduced relative to number of sensors you have? In other words, if length(indx) is much less than the number of sensors, then you may need a cfg.lcmv.lambda greater than 5%.</div>
<div><br></div><div>Best,</div><div>Johanna</div><div><br><div class="gmail_quote">2012/5/8 Frederic Roux <span dir="ltr"><<a href="mailto:fredericroux@hotmail.de" target="_blank">fredericroux@hotmail.de</a>></span><br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div><div dir="ltr">
Dear all,<br><br>I was looking at my lcmv-beamformer maps of ~200 Sec of <br>eyes closed resting state MEG activity and wondering if<br>what I was seeing is the center of sphere artefact.<br><br>The code I used is:<br><br>
%filtering of the data<br>cfg = [];<br>cfg.bpfilter = 'yes';<br>cfg.bpfilttype = 'but';<br>cfg.bpfreq = [5 45];<br>cfg.bpfiltord = 4;<br>cfg.bpfiltdir = 'twopass';<br><br>[meg_data] = ft_preprocessing(cfg,meg_data);<br>
<br>%PCA to extract components with max explained variance<br>[cf,pcs,vexp] = princomp(meg_data.trial{1},'econ');<br>pexp = 100*vexp/sum(vexp);<br>indx = find(cumsum(vexp) <=90);<br>meg_data.trial{1} = (pcs(:,indx)*cv(:,indx)')';<br>
<br>% computing the covariance matrix<br>cfg = [];<br>cfg.channel = {'MEG'};<br>cfg.covariance = 'yes';<br>cfg.pad = 'maxperlen';<br>cfg.sgncmb = {'MEG' 'MEG'};<br>cfg.removemean = 'yes';<br>
<br>[cov_data] = ft_timelockanalysis(cfg,meg_data);<br><br>%LCMV beamformer<br>cfg = [];<br>cfg.channel = {'MEG'};<br>cfg.grid = grid_data;<br>cfg.vol = hdm;<br>cfg.method = 'lcmv';<br>cfg.grid.dim = [Nx Ny Nz];<br>
<br>cfg.lcmv.fixedori = 'no';<br>cfg.lcmv.lambda = '5%';<br>cfg.lcmv.projectnoise = 'yes';<br>cfg.lcmv.keepfilters = 'yes';<br>cfg.lcmv.projectmom = 'no';<br>cfg.lcmv.keepmom = 'no';<br>
cfg.lcmv.reducerank = 2;<br>cfg.lcmv.normalize = 'yes';<br><br>[bf] = ft_sourceanalysis(cfg,cov_data);<br><br>%make power unit invariant<br>bf.avg.pow = bf.avg.pow./max(bf.avg.pow);<br><br>Since I don't have enough experience to judge about that<br>
I wanted to ask if anybody out there with experience <br>could tell me their opinion.<br><br>The grid was computed using an inwardshift of -0.5 and a grid<br>resolution of 2.5 mm. The head model using the ft_prepare_singleshell.<br>
<br>Any help,advice, comments or suggestions would be highly appreciated.<br><br>Best,<br>Fred<span class="HOEnZb"><font color="#888888"><br><br><br><br>-- <br>Frédéric Roux, PhD student<br>Department of Neurophysiology<br>
Max Planck Institute for Brain Research<br>D-60529 Frankfurt am Main<br><a href="mailto:Frederic.Roux@brain.mpg.de" target="_blank">Frederic.Roux@brain.mpg.de</a><br><a href="tel:%2B49%280%2969630183225" value="+4969630183225" target="_blank">+49(0)69630183225</a><br>
<br><br> </font></span></div></div>
<br>_______________________________________________<br>
fieldtrip mailing list<br>
<a href="mailto:fieldtrip@donders.ru.nl">fieldtrip@donders.ru.nl</a><br>
<a href="http://mailman.science.ru.nl/mailman/listinfo/fieldtrip" target="_blank">http://mailman.science.ru.nl/mailman/listinfo/fieldtrip</a><br></blockquote></div><br></div></div>