I suspect it might be problematic but it also depends on how your sources are connected. For instance if you have V1 as your non-hidden source that receives input and is connected to hippocampus and amygdala, this is not a good model because there is nothing that constraints the estimation of your hidden source activity. Usually the model will only be meaningful if you have non-hidden sources that receive inputs via hidden sources. Best, Vladimir 2011/6/14 飞鸟 <[log in to unmask]>: > Dear Vladimir, > Thanks for your quick and detailed reply, I understand that much. How do > you think if I have 1 non-hidden source and 2 hidden soruces? > Thanks! > Haoran. > > At 2011-06-14 23:11:18,"Vladimir Litvak" <[log in to unmask]> wrote: > >>Dear Haoran, >> >>2011/6/14 飞鸟 <[log in to unmask]>: >>> Dear Vladimir, >>> With regard to hidden source, I still have several simple questions. In >>> the paper(David O, etc. 2011) you recommended to me the last time, David >>> specify the hidden source's coordinate as [0 0 0]. However, you suggest that >>> we can specify the position further that 200mm from >>> the origin (like [300 300 300]). Are there any differences between [0 0 0] >>> and [300 300 300]? >> >>That research was done with older SPM. The [0 0 0] trick works only >>with single sphere model for MEG, but setting coordinates to a large >>number works for any head model and the effect is the same. >> >>> I don't know whether this is ture or not: as for any subcortical >>> regions, if I want to specify them as hidden sources, I just need to plus >>> 200mm to their coordinates respectively(e.g. Cingulate [0 36 28] , change to >>> [200 236 228] )? >> >>You just need to set the coordinates to a large number, like [300 300 >>300]. There is no need to use different coordinates for different >>sources, the effect will be the same. >> >>> Further, could I specify two or more hidden sources in my model? If so, >>> how can I specify the coordinates for them? Set both of them to [0 0 0] or >>> others? >> >>You can use [300 300 300] for all your hidden sources. You can have >>more than one but remember that you must also have non-hidden sources >>with some meaningful connectivity to the hidden sources. Otherwise you >>won't be able to get something meaningful from the model. I'd also not >>exaggerate with the hidden source. 5 non-hidden, 1 hidden is OK, but 5 >>hidden 1 non-hidden is unlikely to work. >> >>Best, >> >>Vladimir >> >>> Thank you! >>> Best wishes to you! >>> >>> Haoran. >>> >>> At 2011-06-11 00:56:11,"Vladimir Litvak" <[log in to unmask]> wrote: >>> >>>>Dear Haoran, >>>> >>>>There are two answers to that. The first answer is that whether there >>>>is any signal in the MEG from amygdala or hippocampus is subject to >>>>debate and there is some evidence in favor of that but it's not >>>>conclusive. The second answer is that you can include sources in your >>>>DCM model even if there is no signal in the MEG/EEG from these sources >>>>and learn something about those 'hidden' source from their >>>>interactions with sources that are observed. As an example you can >>>>look at http://www.ncbi.nlm.nih.gov/pubmed/21325540 . You can either >>>>just place sources e.g. in the hippocampus the usual way or (in an >>>>up-to-date SPM) if you specify the position further that 200mm from >>>>the origin (like [300 300 300]) the lead field of the source will be >>>>set to zero and then it will be truly hidden source. You can compare >>>>the two ways with model comparison to see if there is really any >>>>evidence for the source being at particular location or not. >>>> >>>>Best, >>>> >>>>Vladimir >>>> >>>>2011/6/9 飞鸟 <[log in to unmask]>: >>>>> Dear spm's users, >>>>> Thanks for your concern! As DCM for EEG/MEG is based on nerual mass model, >>>>> I wonder that can I apply DCM for EEG/MEG to cerebral limbic system( such as >>>>> cingulate, hippocampus and amygdala etc). In other words, is it valid to >>>>> specify my model not only include cortex regions(e.g. frontal lobe, temporal >>>>> lobe etc.), but also include cingulate or other limbic system's regions? >>>>> Any help will be appreciated! >>>>> >>>>> -- >>>>> Haoran LI (MS) >>>>> Brain Imaging Lab, >>>>> Research Center for Learning Science, >>>>> Southeast University >>>>> 2 Si Pai Lou , Nanjing, 210096, P.R.China >>>>> >>>>> >>> >>> >>> >>> -- >>> Haoran LI (MS) >>> Brain Imaging Lab, >>> Research Center for Learning Science, >>> Southeast University >>> 2 Si Pai Lou , Nanjing, 210096, P.R.China >>> >>> > > >