# [FieldTrip] fieldtrip Digest, Vol 69, Issue 14

K Jeffrey Eriksen eriksenj at ohsu.edu
Tue Aug 16 00:44:17 CEST 2016

```Johannes,

I understand now, at least about the dipole implementation. The rest should come as I work with FieldTrip some more. Thanks for your help,

-Jeff

Message: 11
Date: Fri, 12 Aug 2016 17:09:29 -0600
From: Johannes Vorwerk <j.vorw01 at gmail.com>
To: FieldTrip discussion list <fieldtrip at science.ru.nl>
Subject: Re: [FieldTrip] fieldtrip Digest, Vol 69, Issue 13
Content-Type: text/plain; charset=utf-8

I see. Your misunderstanding is that the dipoles are ?represented? through hexahedrons (and that you need to implement this ;-)). The idea that is usually used to derive FEM for EEG is to stick to the model of a dipole and to represent this dipole through the FEM basis functions. One method to do this (the ?Venant approach?) is already implemented in fieldtrip. The way this representation is achieved is described in the papers I mentioned. If you do not want to use a different approach here, no further work on your side is needed.

What you refer to as a ?source model? is what I would call ?source space? (since ?source model? can be easily mixed up?). The source space can be defined just as you would do it for spheres and BEM in fieldtrip. To apply FEM, you have to indicate to fieldtrip to generate a hexahedral head model and ?simbio? as method of forward computation.

Best,
Johannes

> Am 12.08.2016 um 16:53 schrieb K Jeffrey Eriksen <eriksenj at ohsu.edu>:
>
> Johannes,
>
> I an unfamiliar with the derivation you mention (Venant approach) and will read the papers you list here to try to understand it.
>
> In my naive view based on thinking about the spherical and BEM cases, the head model is simply a set of 3-4 nested volumes, and in the central volume one implements dipoles at particular locations, which can generally be located anywhere inside the innermost volume. In FEM the entire head volume is hexagons, and I see that FieldTrip/simbio uses a modified (distorted or warped) cubic voxel model for the hexagons. I would presume the simplest idea would be to equate particular gray matter (GM) hexagons with individual dipoles, and perhaps this is what the Venant approach does.
>
> I just do not know how to specify particular GM voxels to use as my set of dipoles. How do I address them since they have been distorted from their original cubic voxel shape? Maybe this will be more clear as I read these papers and do some more work with FieldTrip.
>
> Oh, I guess to me a "source model" is the set of generator locations and orientations. For spheres and BEM these could be point dipoles(1 or 3 per position) or perhaps 2D patches. For FEM these would be tetrahedrons or hexagons with 1 or 3 dipoles at each element used.
>
> Thanks,
> -Jeff
*************

```