EEG source reconstruction using fewer number of EEG channels

Hi Mike,

I am planning of source based analysis on my EEG data recorded via Emotiv Epoc headset which has 14 EEG channels. As I am new to source reconstruction problem and studying one some literature, I am not very confidant how to start with. So, I am writing here for a starting up help. I hope this will not be boring for those having better experience of the said analysis.
My first question is about the number of EEG electrodes. I would like to know your thoughts. Is it plausible to do source reconstruction analysis with 14 EEG sensors data.
Secondly, I would like to know which toolbox you recommend for computing the head-model (e.g. brainstorm or fieldtrip etc). I can see your work here in which you said you used brainstorm for leadfield. If I choose a toolbox (let say brainstorm) I can choose their database MRI images to co-register? (because i dont have MRI acquisition for my subjects).
Finally, I would like to know if there is any recommended course for source localization on Udemy or somewhere else (since your course solved challenges in NTSA is very helpful. once again thank you for that).

Rab Nawaz

Hi Rab. Technically, yes, you can do anatomical localization with 14 Emotiv electrodes. The question is whether it’s a good idea. Unfortunately, the answer is no. There are too few electrodes and they are not distributed across the scalp. That means that the level of uncertainty about the source reconstruction is unacceptably high. 64 electrodes is a decent minimum, and >100 would be better.

I recommend sticking to the electrode-level analyses and topographical plotting. Or if you are doing it just to learn about source imaging, then it’s probably better to use a public dataset with higher density.

As for creating the MRI headmodel, yes, I’ve used the Brainstorm toolbox, which in turn calls the OpenMEEG algorithm. You can also use fieldtrip or eeglab or SPM. It’s no problem to use a template MRI for EEG – usually we don’t know the exact electrode locations anyway (only approximate), and people have shown that the difference between subject-specific and template anatomy is negligible in most cases.

Finally, as for courses for learning source imaging, I don’t know of any. There are tutorials for implementing the analyses on the fieldtrip website, and overview videos about it (also from fieldtrip) that you can find on youtube. There are also detailed math books.

Hope that helps!

Hi Mike,

Thank you so much for your quick reply.
Yes, of course your suggestion clarify my thoughts and it helped.

Yeah, while searching about source localization I came across of those fieldtrip and brainstorm tutorials (& that fieldtrip YouTube videos about beamforming).

Anyway, I found the topic interesting and learning worthy.

(Btw, as you don’t prefer some formal addresses like Dr. , Sir. I also prefer to be called Nawaz. Although my first name is Rab, but I like to be called Nawaz. And the reason of that is my friends would always call me Nawaz back in my childhood. I always like to remind about those ungreedy days and missing my friends. :slightly_smiling_face::slightly_smiling_face:)

Again thank you so much for your guidance and help.