<div dir="ltr"><div>Dear Fieldtrippers,<br><br>I am currently doing source reconstructing for MEG data and I’ve run into an issue I’m hoping someone can help with.<br><br>Specifically, I am using the ft_virtualchannel function, but with cfg.method = 'none', so that the components of the virtual channels are not combined. For further analysis I need the sources to remain separated per parcel.<br><br>I closely followed the scripts from <a href="https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fmcvinding%2Fwarpimg%2Fblob%2Fmain%2Fbenchmarking%2Fsource_c_virtualchan.m&data=05%7C02%7Cfieldtrip%40science.ru.nl%7Ce0fe659ef5b84dd9d38008dd4f62ec13%7C084578d9400d4a5aa7c7e76ca47af400%7C1%7C0%7C638754008476955439%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=qb06dZ5LTlb9EXpt2yelEy5yZ8OvQM92GSvAjx7cL14%3D&reserved=0" originalSrc="https://github.com/mcvinding/warpimg/blob/main/benchmarking/source_c_virtualchan.m" shash="RPPiIfpulxMLU8Hg5+sy273TN9OegoAMXFx18z8GUlLPOddP6hvXHi5FzAiJN9/57nV4rlOT4I7ttt3J9AsI1yzz4wRzoJBNu8HNW9BPdQxhEcOkR9twclDEIKwLQntfeZhhM5bPZUhPeRH+jSMAEHtP5v1WKwAvFc8bHwHcncA=">Mikkel Vinding and Robert Oostenveld</a>: I created the spatial filters using the LCMV beamformer and interpolated the AAL atlas onto the sourcemodel, then used the ft_virtualchannel function.<br><br>The problem is that I'm seeing a highly uneven number of virtual channels per parcel. For example, Heschl_L has only 25 channels, while Frontal_Mid_R has 651. In total, the mean number of virtual channels per parcel is 200, but the standard deviation is 146. This seems quite strange to me, as I would expect the ROIs to be more comparable in size.<br><br>Is this expected behavior? Or have I made an error somewhere in the process?</div><div><br></div><div><br>My code is as follows:<br><br>%% Interpolate Atlas<br>load(fullfile(ftpath, '/template/sourcemodel/standard_sourcemodel3d4mm'));<br>atlas = ft_read_atlas([ftpath filesep 'template/atlas/aal/ROI_MNI_V4.nii']);<br><br>cfg = [];<br>cfg.interpmethod = 'nearest';  % Or another interpolation method if needed<br>cfg.parameter = 'tissue';      % or other relevant parameters based on the atlas<br>sourcemodel_atlas = ft_sourceinterpolate(cfg, atlas, sourcemodel);<br>sourcemodel_atlas.tissuelabel = atlas.tissuelabel;<br><br>atlas_grid = ft_checkdata(sourcemodel_atlas, 'datatype', 'source');<br>atlas_grid.inside = sourcemodel.inside;<br><br>%% Create Common Filters using LCMV beamformer<br>cfg = [];<br>cfg.method              = 'lcmv';<br>cfg.channel             = 'meg';<br>cfg.lcmv.keepfilter     = 'yes';<br>cfg.lcmv.fixedori       = 'yes';<br>cfg.lcmv.lambda         = '5%';<br>cfg.lcmv.kappa          = kappa;<br>cfg.lcmv.projectmom     = 'yes';<br>cfg.headmodel       = headmodel;<br>cfg.sourcemodel     = grid;  %leadfield<br>commonfilter_sub              = ft_sourceanalysis(cfg, tlckw);  %tlckw is time-locked data<br><br>%% add atlas to Filter<br>commonfilter_sub.tissue = atlas_grid.tissue;<br>commonfilter_sub.tissuelabel = atlas_grid.tissuelabel;<br><br>% Create virtual channels<br>cfg = [];<br>cfg.parcellation = 'tissue';<br>cfg.method = 'none';<br>virtualData = ft_virtualchannel(cfg, data_av, commonfilter_sub); 
%
data_av

is averaged data

over trials<br><br><br>
<div>Thanks in advance for your help!<br></div>Marisa Birk<br><div><br></div>

<br><br><br><br></div></div>