[Eeglablist] Face skin with head-plot

arno arno at salk.edu
Sat Apr 22 05:47:50 PDT 2006


Borís BURLE wrote:
> Dear all,
>
> I would like to have some information on how to get the  skin color on 
> the face generated by headplot().
There are usually two files with head mesh information, one for the 
bottom of the head (skin color) and one for the top (interpolated color 
based on electrical potential). The first is contained in the variable 
"TRI2" in the "mheadnew.mat" file and the second in the variable "TRI1" 
of the same file. These two variables describe sets of triangles and 
index the variable "POS" in "mheadnew.mat". For example the two rows

1 2 3
3 5 7
....

would indicate that the first triangle links point 1,2 and 3 (which 
coordinates are given in  the "POS" variable) and the second triangle 
links point 3,5 and 7.

The trick is that if you have a standard head mesh, it is not 
straightforward to divide it in two (with one part for the skin where 
you estimate that the color interpolation is meaningless). If you simply 
select the points above a specific height for instance, the mesh 
separation will not be straight and smooth (it will look like sawtooth 
because the mesh points do not lie on the selected boundary). You should 
thus write a function that will add points at the boundary and recreate 
new triangles to make the edge smooth (if you do it, please let me 
know). Other software (like BESA) do interpolate on the whole head, but 
I agree this would be welcomed. Also, the best thing would be to compute 
the 3-D convex hull that would determine the outer limit of the 
electrodes on the head-mesh in 3-D (so there would be no interpolation 
beyond the outer limit). This is more simple that it seems since Matlab 
does have a 3-D convex hull function.

Best,

Arno

ps: the file "mheadnewtri2.txt" (and other text files) were made 
available because Windows computer could not read the "mheadnew.mat" 
matlab file (this is because Winzip uncompresses it as text by default 
so it becomes corrupted). This has been fixed with EEGLAB 5.0x since we 
are now creating ".zip" files that are fully recognized both by Windows 
and Unix (including Mac of course).
> It seems that one needs a special mesh file (mheadnewtri2.txt) to 
> import in the mheadnew.mat. Where can we find this file ? Or is there 
> any other way of getting the face in skin color (through options in 
> the command line ?)
>
>    Thanks in advance,
>          B. Burle
>




More information about the eeglablist mailing list