Hi Denny,
in general the lines won't be perpindicular to both tangent planes. Using the surface normal to compute thickness fails dramatically in some cases (which is why we don't use it). Consider a flat bottomed U for the pial surface with a larger flat bottomed U for the white surface. Near the edges of the larger (white) U the surface normal is parallel to the pial surface and hence never intersects it.
cheers, Bruce
On Wed, 27 Jan 2010, Milakara, Denny wrote:
How different is your method from using lines perpendicular to tangent planes touching both pial and white surface vertices?
Cheers, Denny
-----Ursprüngliche Nachricht----- Von: Bruce Fischl [mailto:fischl@nmr.mgh.harvard.edu] Gesendet: Mittwoch, 27. Januar 2010 01:36 An: Milakara, Denny Cc: FreeSurfer Forum Betreff: Re: [Freesurfer] Pial surface of both hemispheres as specific ASCII output
Hi Denny,
the thickness we use is the shortest distance to the white surface from each point on the pial surface averaged with the shortest distance to the pial surface from each point on the white surface.
cheers Bruce On Wed, 27 Jan 2010, Milakara, Denny wrote:
Hi Doug,
I'm a bit confused because the thickness values from the
thickness file is not identical to the euclidean distances I calculated myself. Calculated value is always a bit greater.
Example #1 Pial left: -16.679081 -87.523453 30.427801 Thickness left: 000 -16.95796 -85.12147
30.10204 2.06924
Thickness value from file: 2.06924 Thickness value calculated: 2.44
Example #2 Pial left: -17.137926 -87.548676 30.134594 Thickness left: 001 -17.27066 -85.21771
29.78851 2.13066
Thickness value from file: 2.13066 Thickness value
calculated: 2.3603
If the thickness values from l/rh.thickness are correct and
not the pure 3d euclidean distances, what factor makes the difference between them?
Cheers,
Denny
-----Ursprüngliche Nachricht----- Von: Bruce Fischl [mailto:fischl@nmr.mgh.harvard.edu] Gesendet: Mittwoch, 30. Dezember 2009 22:15 An: Milakara, Denny Cc: FreeSurfer Forum Betreff: Re: [Freesurfer] Pial surface of both hemispheres as specific ASCII output
Hi Denny,
the number of vertices for the pial surface should be identical to the # in the white surface. The thickness is a property of both surfaces, and can be associated with either.
cheers Bruce On Thu, 31 Dec 2009, Denny Milakara wrote:
I'm sorry that my message was confusing. The command line in your answer produces WM surface and thickness information per WM-vertex. I need thickness info per
GM-vertex. So to
say, the thickness 'below the surface' (because GM) and not
above it
(because WM). Since the number of vertices for the grey matter is somewhat greater than for the white in the same brain, it
is necessary
to map the 'thickness' to the pial surface. The command
line I tried, is related to that question:
mri_surf2surf --hemi lh ?srcsubject bert --srcsurfval
lh.thickness
--hemi lh --trgsubject bert --trgsurfval lh.pial --hemi lh
THX! Cheers Denny
Bruce Fischl wrote:
try doing:
mris_convert -c lh.thickness lh.white lh.thickness.asc
(note that this is in the mris_convert help)
cheers, Bruce
On Thu, 31 Dec 2009, Denny Milakara wrote:
> yes, without any switch it is outputed with vertices and
faces. The
> conversion to .off is trivial with matlab. If there is a
need I can
> provide the code needed for conversion. I didn't though
converted it
> from RAS to xyz, since it works well with MeshLab and
RoffView in MSWindows.
> > > I find it very interesting to include the cortical thickness > information in surface information, ideally a thickness
value per
> vortex. I found several postings about this issue, but I
obviously
> didn't understand that process correctly. > > If I try in 'bert/surf' this: > mris_convert lh.thickness bert_thickL.asc I get this: > Segmentation fault > > So, I tried to map thickness to pial surface, by doing this: > mri_surf2surf --hemi lh ?srcsubject bert --srcsurfval
lh.thickness
> --hemi lh --trgsubject bert --trgsurfval lh.pial --hemi
lh but I got
> that: > ERROR: Option ?srcsubject unknown > > What am I doing wrong? > > > Cheers, > Denny > > > > Bruce Fischl wrote: >> no, and I don't think you should use -p. Try having the
output be
>> R.asc and see how it works (-p is to convert patches,
and -c is for
>> curvature files. The default is for surfaces) >> >> cheers, >> Bruce >> >> >> On Wed, 30 Dec 2009, Denny Milakara wrote: >> >>> yes, I also expected integers - indices of vertices
that describe
>>> the faces (in this case triangular faces). I guess the
following
>>> command produced the incomplete result: >>> mris_convert -p rh.pial R.txt >>> >>> Should I have used switch '-c'? >>> >>> >>> >>> Bruce Fischl wrote: >>>> the index is implicit by position (e.g. the first face
is 0, the
>>>> 2nd is 1, etc...). Each face just lists the vertices
(by index)
>>>> that are in it On Wed, 30 Dec 2009, Denny Milakara wrote: >>>> >>>>> Shouldn't the faces be integer numbers? There are no
integers at
>>>>> the end of the output file. Is it possible to
generate ASCII
>>>>> file without faces? >>>>> >>>>> I need brain surface only, so I assume that 'lh.pial'
and 'rh.pial'
>>>>> are OK for that purpose. >>>>> >>>>> >>>>> Cheers, >>>>> Denny Milakara >>>>> >>>>> >>>>> >>>>> Bruce Fischl wrote: >>>>>> Hi Denny, >>>>>> >>>>>> the faces are in the 2nd half of the file. If you
scroll (way)
>>>>>> down you'll find them. The columns are the x,y,z
coords in what
>>>>>> we call surface RAS, which is documented on our wiki. >>>>>> >>>>>> cheers, >>>>>> Bruce >>>>>> >>>>>> p.s. what surfaces you need are need is dependent on
what you
>>>>>> want to do (e.g. inflated, white, etc...) >>>>>> >>>>>> On Wed, 30 Dec 2009, Denny Milakara wrote: >>>>>> >>>>>>> Thank you for your answer Bruce, >>>>>>> >>>>>>> >>>>>>> Nobody expects here default solutions for exotic
needs. So let
>>>>>>> me be more specific: >>>>>>> >>>>>>> I used 'mris_convert' to convert 'lh.pial' in
ASCII format.
>>>>>>> The output file containes 3 colummns of real numbers with >>>>>>> 3xwhitespace between them. >>>>>>> >>>>>>> 1) Are 'lh.pial' and 'rh.pial' the proper and only needed >>>>>>> source files for converting and subsequent
rendering of cortex
>>>>>>> surface? >>>>>>> >>>>>>> 2) Does each row contain x y z of vertices in RAS
space? If
>>>>>>> not, what information then contain those columns? >>>>>>> >>>>>>> 3) I don't see the faces in the output file. How can I >>>>>>> generate them? >>>>>>> >>>>>>> >>>>>>> Regards, >>>>>>> Denny Milakara >>>>>>> >>>>>>> >>>>>>> >>>>>>> Bruce Fischl wrote: >>>>>>>> nothing by default that will do it, but it
should be easy
>>>>>>>> enough in matlab >>>>>>>> >>>>>>>> >>>>>>>> On Mon, 21 Dec 2009, Denny Milakara wrote: >>>>>>>> >>>>>>>>> Dear Freesurfers, >>>>>>>>> >>>>>>>>> >>>>>>>>> I wonder if there is a way to output the
triangulated pial
>>>>>>>>> surface for both hemispheres into a single ASCII file >>>>>>>>> according to OFF-file >>>>>>>>> schema: >>>>>>>>> ---------------------------------- >>>>>>>>> OFF numVertices numFaces numEdges x y z x y z ... >>>>>>>>> numVertices like above NVertices v1 v2 v3 ... vN
MVertices
>>>>>>>>> v1 v2 v3 ... vM ... numFaces like above >>>>>>>>> --------------------------------- ...where the
number of
>>>>>>>>> edges is to be ignored. >>>>>>>>> >>>>>>>>> >>>>>>>>> Thank you for your suggestions! >>>>>>>>> Denny Milakara >>>>>>>>> _______________________________________________ >>>>>>>>> Freesurfer mailing list >>>>>>>>> Freesurfer@nmr.mgh.harvard.edu >>>>>>>>>
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
>>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>> _______________________________________________ >>>>>>> Freesurfer mailing list >>>>>>> Freesurfer@nmr.mgh.harvard.edu >>>>>>>
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
>>>>>>> >>>>>>> >>>>>>> >>> > >