grey matter thickness measurment

[MSh]

Hello Brainvisa experts

I'm interested to measure cortical thickness and based on the related discussions on forum, there seem to be two options:
1. cortical surface parameterization (which apparently is not recommended)
2. RIC toolbox (by Kochunov)

Having looked at the documentation of RIC toolbox, I had the impression that the thickness is calculated as the normal distance between white matter mesh and hemi mesh. If my understanding is correct, given the issues with hemi mesh reconstruction (e.g the problem described in http://brainvisa.info/forum/viewtopic.php?f=2&t=1428) how accurate would the thickness measurement be? What's the best way to get reasonably accurate measure of the thickness of grey matter or sulci? Would "maxdepth" (in the reconstructed graph) give accurate estimation of the thickness of each sulcus.

Thanks
MSh

[riviere]

Hi,
There are several ways to get thickness measurements in BrainVisa:
- Peter Kochunov's RIC toolbox. You may discuss of it with him, but if the normal to the mesh may locally be inaccurate at some points on the mesh, I guess the mean effect is quite OK.
- Grégory Operto's "cortical thickness texture" process in the cortical_surface toolbox (you need to be in "expert" userLevel), which I guess works in a quite similar way, trying to build a median surface between the G/W and G/CSF surfaces
- Pauline Roca's thickness calculations, which are integrated in the cortical folds graph construction 3.1 variants. This one is based on distance maps in the voxels of the cortex segmentation. Measurements are averaged at folds level and included in the cortical measurements in the graphs. In the graph construction process, you may optionally write an image of the median voxels layer with the thickness measurements in each voxel of this layer. The values should probably be averaged or smoothed to be really meaningful, because they are measured on voxels and the cortex is only 2-3 voxels thick.
Denis

[MSh]

Hi Denis

Thanks very much for your prompt reply - much appreciated!

I tried options 2 and 3 but have questions:

With “Create cortical texture” there are two output texture files: “white.thickness.tex” and “hemi.thickness.tex”. How are these two different and which one relates to the cortical thickness? (thickness at each vertex?) I used fusion of these texture files with the inflated mesh to get a visual estimation of thickness values. Is there any way to extract numerical values for the cortical thickness instead?

For option 3 (using 3.1 version of cortical graph reconstruction), there are fields called “thickness_mean” and “GM_volume” among other attributes of each labeled sulcus. Does “thickness_mean” mean average cortical thickness along that sulcus? How about “GM_volume”?

Many thanks
MSh

[riviere]

Hi,

- The cortical thickness textures are actually arrays of values, each value is mapped onto a mesh vertex. The ordering of the values in the texture array must match the ordering in the mesh (and, of course, numbers of elements in both the mesh and the texture must match). As he have two meshes for the cortex (inside and outside the cortex: G/W and G/CSF interfaces), the two meshes have different structures and different vertice numbers. So the process outputs two thickness textures, one for each of those meshes.
The numerical values of the thickness are the texture data, you can extract them using our libraries (pyaims in python, aims in C++), or convert them to an ASCII format (using AimsFileConvert -a with a .tex format extension for instance). But you still have to make the link between these values and the mesh vertice positions and ordering.

- in cortical folds graphs, each fold (or part of sulcus) is attached a cortical area definied by a voronoi diagram in the cortex. In 3.1 graphs, the thickness_mean is the average of the thickness over the corresponding cortical area. The GM_volume is the volume of grey matter of this area.

Denis