Detecting hippocampal activity through EEG components using VWM task


#1

I am currently conducting a pilot study for my MSc dissertation, investigating whether we can detect any of the hippocampal activity through EEG components through the use of a visual working memory task. The only problem is that I only know how to analyse for the P300 ERP. I am just so confused about the hippocampal bit. How can we identify if some activity is from the hippocampus? How can we differentiate between a hippocampal ERP and the P300? I am using EEGLAB for analysis. I was thinking we could see which components contributed to the P300 ERP, and then plot a 3D scalp map of those components to see if any hippocampal activity is coming from any of the regions around the cheekbones for example? Is that sufficient?

Can someone please help?? I am very confused.


#2

You might find these papers relevant,

This one is a Review paper, with more background.

Source localization would be required, in other words large numbers of channels and very sophisticated signal processing. It’s not clear that EEG would work; this paper uses MEG. Even then, it is also tricky separating cortical activity from the deep brain activity.

So in short, what you are requesting would require hugely expensive equipment.


#3

From my novice understanding, most discernable EEG signals stem from relatively large scale synchronization of cortical neurons - particularly on gyri. Even detecting signals in cortical folds can be difficult, as neigboring sources tend to cancel each other.

However, the following journal article may prove useful to your question:


#4

Here is an excerpt from Brain-Computer Interfaces, Chapter 3 “Electric and Magnetic Fields Produced by the Brain”, page 58:

The Inverse Problem

The classical inverse problem in EEG is concerned with finding the locations and strengths of the current sources… with discreet samples… on the surface of the volume conductor… In practice, dipole searches employing sophisticated computer algorithms are based on recordings at perhaps 20-128 or more surface locations… By contrast to the forward problem, the inverse problem has no unique solution… Scalp potentials can always be made to fit a wide range of distributed cortical sources… In the absence of additional information, the constraints (e.g. psysiological assumptions) required to obtain these fits cannot generally be expected to be accurate, and the inverse solutions (computed source locations) are generally no better than the psysiological assumptions and head-model accuracy… Sophisticated computer models are useful only if they are based on sound physiology; sophisticated mathematics can never compensate for unrealistic physical assumptions.

Further reading