Forum

snovotill · Joined: Feb 8, 2013 Posts: 80

Hi Jarek, In BioEra we have the FFT based bh-Coherence element which implements the famous Thatcher algorithm. BioExplorer on the other hand seems to implement Coherence and Phase on top of simple bandpass filters. Do you have any idea how the latter could be approximated in BioEra for comparison?

I've seen two bandpass filters feeding into (A+B)-ABS(A-B) but that's more like a Synchrony measure. I can't imagine how one derives Coherence and Phase from bandpass filters and I'm not having any luck googling for it.

thanks,
Stepan

jarek · Joined: Oct 22, 2007 Posts: 1073

I also don't know how that works. And I also spent some time googling. All definitions of coherence I found are based on complex spectral output, like here: http://en.wikipedia.org/wiki/Coherence_(signal_processing)
Which is exactly what BioEra does.

snovotill · Joined: Feb 8, 2013 Posts: 80

I came across an old post from Larry citing that BioExplorer Coherence Object calculates as per section 2.1 of the paper \"The contribution of EEG coherence to the investigation of language\" which explains an FFT approach. Therefore it appears the methodology is very similar to what BioEra does except it seems that bandpass filters are being used as a front-end to the FFT.

jarek · Joined: Oct 22, 2007 Posts: 1073

If it was FFT, then there would be some way to configure it, e.g. number of bins like in all other BioEx objects which are clearly FFT based. That is why I doubt it.

I heard about other (than FFT) way to calculate coherence, but I have never found an algorithm. I only had some ideas, which might be what is being done, but I am not sure.