I don't have a big experience, but I've heard about the the FMRIB plug-in for EEGLAB. It seems to be quite reliable.

http://fsl.fmrib.ox.ac.uk/eeglab/fmribplugin/

Dear David

We usually use ICA (as implemented in EEGLAB) to remove the components that correlate with the EKG channel (recorded simultaneously during the EEG/fMRI).

Best

João 

The optimal basis set method in the FMRIB EEG plug in is good for removal. If you have brainvision analyser 2, that also does a good job. ICA on its own is not reliable though i can be useful to remove some residuals after you have run OBS or another template based subtraction method. Just as important as the actual removal is that you are accurately detecting the heart beats and ensure that you are marking exactly the same point in the beat consistently. Make sure that you have a good ECG/VCG signal as this is crucial.

Dear David,

Correction of the cardioballistic artifact can be tricky but it is possible to retrieve good data. If you are not happy with the pulse artifact correction in Analyzer, you need to track down why it did not work for you. Here are a few things to consider:

 - Have you checked the accuracy of the R peak detection in the datasets where you observe a poor correction (poor ECG, problems with the detection algorithm)? These temporal reference points (R-markers) are critical for the success of the correction.

- Are there problems with the pulse artifact itself (e.g. obscured by head movements, problems with the preceding gradient correction)?

If you would like some additional help or further info on using the CB Correction in Analyzer please feel free to contact the Brain Products Technical Support team: [email protected]

Techsup will be more than happy to track down the problems.

I hope that helps!

Tracy

Thank you all very much for these very helpful comments! We will soon acquire more data and then also try out and compare different methods/options. Once we have converged on a successful solution, I will shortly summarize our experience here in case others struggle with similar problems. 

Regarding Tracy's comment, yes we indeed noticed that the algorithm failed to reliably detect many of the R peaks and thus probably also struggled with removing the BCG artefact, as you have suggested? We noticed nothing in particular in the ECG trace, but will keep an eye on this and might contact the tech support of Brain Products.

We were also not very convinced by the Gradient removal, so this might have worsened the performance of the BCG algorithm (and possibly masked R peaks that were not detected)? Do you maybe have further practical tips about this (I could not find online documentation)?

Many thanks again!

David

Dear David,

yes, if the R peaks were not properly detected the correction won't be successful. This could be because of a poor ECG signal or it could be that the parameters selected for the peak detection were not appropriate. If you have a good signal it is likely to be the latter, in which case I would suggest running the cardioballistic correction transformation in semiautomatic mode so you can check and adjust the peak detection.

If you are not convinced by the gradient artifact correction you should try to fix that first since this will have an impact on any further corrections and analyses. The success of the correction can be influenced by the recording parameters and physical setup of your equipment so you should check this in addition to trying out different correction methods.

It is definitely a good idea to get these problems fixed during your pilot testing so I would suggest contacting Brain Products technical support so you can be up and running with good data quality as soon as possible.

Best wishes,

Tracy

Dear David,

Hi

Did you finally find the most reliable toolbox for Artifact removal?

I tried both FMRIB EEG plugin and Analyzer, but none of them worked as i expected. I have problems with different kinds of error in FMRIB plugin which i don't have any idea about how to fix them, Also i can't clear all the data with Analyzer and finally there remain some Gradient and BCG Artifacts.

Best

Nafiseh