Hi Marco, Thanks for sharing your observations. My reply is more an attempt to keep this thread active, until an expert decides to answer, but here are a couple of thoughts. I suppose it is feasible that the torso of a subject may have moved (rotated), even in the theoretical case of no head movement between runs. That would lead to a different ECG projection on the sensors, and thus a smearing out of the average SSP vector relative to the ones optimised for each run separately. Could the explanation for the blinks have to do with the fact that the eyes are close to the "white zone" (SSS no-mans-land, between inner and outer convergence zones), and thus signals from there not being totally accurately coregistered to the initial position? In a similar vein, perhaps the blink-related field is spatially under-sampled, and thus not accurately represented in the SS basis. (Does the blink-field induce high spatial frequency components in the sensors at the edge of the helmet that don't get dealt with?) In any case, thanks for pointing this out. Very interesting and relevant even if my logic fails :) /Chris -- Christopher Bailey, MSc MEG Engineer, MINDLab Core Experimental Facility Center of Functionally Integrative Neuroscience (CFIN) Aarhus University, Denmark tel. cell: +45-2674-9927 tel. office: +45-7846-9942 On Jan 24, 2014, at 2:05 PM, Marco Buiatti <[log in to unmask]<mailto:[log in to unmask]>> wrote: Dear all, I'd be happy to have some feedback on my question, which I think is quite a general issue. If however you think it's trivial, I'd be happy to know the obvious solution :-) Thanks, Best, Marco On 17 January 2014 15:47, Marco Buiatti <[log in to unmask]<mailto:[log in to unmask]>> wrote: Dear Neuromag users, I have a question concerning what is the best strategy to compute SSP projectors for blink/cardiac artifact removal in Neuromag data. Typical datasets from my lab contain 4 to 8 separate runs from each subjects, and co-registration across runs is typically computed with Maxfilter (Maxmove), based on the head position measurement performed at the beginning of each run. In principle, since co-registration has already been done, it should make sense to compute SSPs for all runs together. However, from my experience, results are not always as good as by computing and applying SSPs to each run separately. To test this, I have compared the two approaches on one of my datasets: 1) Local: Compute SSPs for both blinks and cardiac artifacts separately on each run (here the first run of my dataset) 2) Global: Compute SSPs for both blinks and cardiac artifacts on all runs together. My dataset consists in 8 runs of 6 minutes each, already co-registered by MaxMove. I did this test by using Brainstorm (actually, I have posted a similar message on the Brainstorm forum a few weeks ago, but I did not get any feedback from the Neuromag users). Here I attach: * The blink and cardiac artifact for MAG (top row) and GRAD (bottom row, topography is the norm of the grads), first run. * The SSP projectors for MAG and GRAD for blinks (top row) and cardiac (bottom row) for approach 1) (local) and 2) (global). They look very similar, though blink topographies are more localized around the eyes for the 'local' approach, and cardiac topographies look also slightly cleaner. * Comparison of residual blink for local (left side) vs global (right side) approach. * Comparison of residual cardiac for local (left side) vs global (right side) approach. As you see, though most of the artifact is removed with both approaches (note the change in scale of the amplitude), the local approach is more successful. Of course, it is just an example, I cannot generalize. Is this due to an inefficient co-registration, or is it possible that blinks/cardiac artifacts change during the recordings? What is your typical procedure? Thanks in advance for your feedback, Best, Marco -- Marco Buiatti, PhD CEA/DSV/I2BM / NeuroSpin INSERM U992 - Cognitive Neuroimaging Unit Bāt 145 - Point Courrier 156 Gif sur Yvette F-91191 FRANCE Ph: +33(0)169.08.65.21 Fax: +33(0)169.08.79.73 E-mail: [log in to unmask]<mailto:[log in to unmask]> http://www.unicog.org/pm/pmwiki.php/Main/MarcoBuiatti *********************************************** -- Marco Buiatti, PhD CEA/DSV/I2BM / NeuroSpin INSERM U992 - Cognitive Neuroimaging Unit Bāt 145 - Point Courrier 156 Gif sur Yvette F-91191 FRANCE Ph: +33(0)169.08.65.21 Fax: +33(0)169.08.79.73 E-mail: [log in to unmask]<mailto:[log in to unmask]> http://www.unicog.org/pm/pmwiki.php/Main/MarcoBuiatti ***********************************************