Hi Johnatan

The BCG you observe is very typical artifact and there is plenty strategies related to it. Let me first relate to the theoretical issue you asked - why the frontal electrodes are more affected.

The BCG artifact is induced by 3 phenomena:

1) pulsating blood veasels

2) blood flow (changes in magnetic field susceptibility)

3) small movements due to heart beat and blood flow.

The last is the most important and also is the reason why the artifact is most prominent in frontal electrodes. The head lying on back (posterior channels are let's say stabilized by the MRI bed. So the movement (translation and rotation) is the effect of moving up and down of the whole body in the cardiac rhythm, however head is stabilized on the inion, so it rotates as pushed by neck so in result instead of direct vector you have rather circular momentum. I hope you get the point.

In terms of filtration I performed over 250 EEG-fMRI recordings and from my experience I found the best is the spatial filtration using PCA. The problem is how many components have to be filtred and which subspace projection choose. I generaly use all components which explain more than 1% of variance with Surrogate brain signal modeling.

If you have more questions, let me know

PS. Did you heard that new BESA Research has a whole EEG-fMRI processing pipeline?

Hello Mateusz,

Thank you for your answer. What we have observed is that the efficacy of the removal of the artifact is highly dependent on the electrode (BCG can be more easily removed from occipital electrodes than from frontocentral electrodes). We also have noted that variability in the morphology of the BCG is higher for the most affected electrodes which i guess is related with the phenomena you described.

My next question would be, according to your experience, have you found any effective strategy to reduce the artifact from the most affected electrodes during data acquisition? We have tried using a "spandage band", which is sort of a cloth net that envelopes the whole electrode cap to reduce movement of the electrodes. We have also tried positioning soft pillows between the forehead of the subject and the RF receiving coil, in order to minimize movement and apply preassure over these electrodes. Neither of these strategies have solved the problem, so I wonder if you have any idea of what else can we try.

Refering to the BESA research software I checked it out briefly at the EGI stand in the last SfN meeting, but we have not work with it nor do i know anyone that has, although it does seems to be very friendly software.

Once again, thank you for yor answer.

hello

the problem you report seems to be related with the choice of your reference electrode.

It may be solved using a different channel for reference.

best regards

Hi Jonathan, I am sorry that you did not stop at BESA booth - we would had chance to talk there :)

The reference electrode change, as Joao said, might be a good idea - try to recompute data using average reference.

As I wrote before I would try spatial filtration using PCA or SVD. OBS is not the best idea (since you need to find every R peak individualy). With ICA approach I had similar problems and the reason is that the ICA method is not the best for this - BCG generates few artifact topographies that are dependent so ICA (independent component analysis) seems not right too for this in my opinion.

Maybe you could PM me (mateusz.rusiniak(at)besa.de ) and send me a sample data so I can try to figure out what you can do with this...