It seems to me that vertex areas in the native subject space are not meaningful. The orig surfaces should be uniformly tesselated; however, topology correction and final surface finding should introduce complicated and probably uninteresting variation.
But, if a subject's surfaces are resampled to atlas (i.e. make_average_subject with only one subject as input), then the vertex areas should indicate the degree to which squishing or stretching is required to match the atlas (as Bruce said), which could be quite interesting. Or put another way, vertex area represents the relative size of a patch of cortex compared to the collection of subjects that went into the atlas. This measure of vertex area may vary between groups and may be correlated to certain genetic markers.
Since volume depends on both area and thickness, it seems redundant to use all three measures.
Date: Tue, 17 Feb 2009 23:16:33 -0600 Subject: Re: [Freesurfer] Qdec area measure From: mharms@conte.wustl.edu To: fischl@nmr.mgh.harvard.edu CC: d.gitelman@gmail.com; freesurfer@nmr.mgh.harvard.edu
Here is a question related to this whole discussion (prompted by what Darren wrote): As a practical matter, have you found the localized (i.e., vertex-based) measures of area and volume to be biologically useful and meaningful? Have they been used in any published studies? (I don't recall seeing any surface maps of "area" or "volume" differences between groups, whereas there are obviously many published studies showing surface maps of thickness differences).
cheers, Mike H.
Hi Darren
the curvature is related to the angles of the edges connected to each vertex (also called the "angle deficit"). The surface area should just be given by how much compression expansion the individual surface undergoes when it aligns to a spot in the atlas. We'll compute it and correct for it sometime soon.
cheers, Bruce On Tue, 17 Feb 2009, Darren Gitelman wrote:
Devdutta, Nick, Bruce, Mike
I'm trying to understand these measures in terms of how they would map to the brain physically.
Would it be correct that the size of the triangles touching a vertex would be related to the amount curvature near a particular vertex and therefore the amount of unfolding that takes place? Thus surface area would seem to relate to both brain size and the foldiness of the brain, and brain size could relate to a variety of factors including the cortical thickness but also the amount of underlying white matter. So overall surface area seems to be a complex metric that could reflect changes in underlying white matter, cortical thickness and foldiness. Is this correct?
I guess one would have to correct for overall brain size (using TIV or similar measure as a covariate) when examining surface area because of the effect of amount of gray and/or white matter.
Volume then would be the amount of cortex underlying a vertex and the triangles connected to it (estimated as an average of the pial and white matter surface areas).
Darren
I think the area probably should be corrected to preserve the native area
of the subject, but as you say we haven't gotten to the correction yet. On Tue, 17 Feb 2009, Nick Schmansky wrote:
Devdutta,
The 'area' of a vertex on a surface in freesurfer is just the average of the area of the triangles which touch that vertex .
The volume data associated with a vertex created for a subject is not buggy in how it is calculated now. it is the area of the mid-point between the white and pial surfaces times the thickness at the vertex. The question that has been bandied about internally is whether this volume calculation should be corrected when it is mapped to the fsaverage surface during a group analysis (qdec does this mapping of the volume calcs of each subject to the fsaverage surface, when you select 'volume' as one of the measures). At some point we will visit this issue (and determine if it is an issue) but other stuff seems to take priority. Maybe someone on the list can chime in.
Nick
On Tue, 2009-02-17 at 11:28 -0600, Devdutta W wrote:
Hi, In Qdec, what exactly does the area option measure? As Iunderstand, qdec does its calculations vertex by vertex, which makes sense in case of cortical thickness. But if area refers to surface area how does that change from vertex to vertex? Or is it calculating the area by ROI? Is my understanding of this correct?
Also, as regards to the volume option, I was told that it isnot working. But when I ran it it seems to work fine. Should I take that to mean that the calculation has bugs and that the results it displays are incorrect? Or has it been fixed?
I greatly appreciate any help in this matter. Thank you, Devdutta _______________________________________________ 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
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
_________________________________________________________________ See how Windows connects the people, information, and fun that are part of your life. http://clk.atdmt.com/MRT/go/msnnkwxp1020093175mrt/direct/01/