hi yczhang,
see also my reply on Dec 2 2008 to frederic.andersson@univ-tours.fr re similar question -- search freesurfer archives on "comparing hemispheres"
Ultimately to look at the symmetry question properly, we will need Doug's tool. In the interim, below is one, admittedly imperfect approach.
Anyone else have a better suggestion in the interim?
Suppose you have a cluster of thickness difference (or bold signal) in one hemisphere between 2 groups. Create your own average subject (to increase accuracy of visualization (but fs average would work as well). Then hand-draw the analogous cluster in the other hemisphere. To do so, you would need to determine the min and max values that occur in x, y, z in talairach space and additional control points as required for the cluster or region where the primary finding occurs . Then flip the sign on the x coordinates and hand draw the cluster in the opposite hemisphere, guided by the tal coodinates. Determine the variable of interest (mean thickness, mean bold signal, etc) in the "mirror" cluster. Input that variable into a glm/anova (outside of freesurfer) where side, group, variable of interest and test for whether there is a side x variable interaction, suggestive of a truly lateralized finding. Remember that the test statistic/significance level for the PRIMARY hemisphere finding is not appropriate to report as it is inherently biased (it is defined as the region where there is a significant between group difference. It is the "mirror" contra-lateral and the presence or absence of a side x variable interaction that is the focus. You might also be able to simply analyze the mirror cluster alone without putting the other side into the anova since it, unlike the primary finding cluster, is not a biased metric.
best, carl
>>>>>>
no, the vertex index is essentially arbitrary and you can't use it to compare left and right. Doug had some tools for this, but I'm not sure whether they are ready for primetime yet.
cheers, Bruce
On Sun, 18 Jan 2009, Zhangyuanchao wrote:
Dear Expert,
I want to exame the symmetry property of two hemispheres,for example,
symmetry of cortical thickness or lgi, I want to know whether the same ID of the two hemispheres after resampling using the average pial surface corespond to the same anatomical region.
For example,
ID=1000, For the left hemisphere, the vertex 1000 corespond to 'medial_wall';
what about the vertex 1000 on the right hemisphere? Is it also corespond to the same anatomical region "medial_wall" so that we can compare their cortical thickness of vertex 1000 on both left and right hemisphere to exame the symmetry property.
Thanks a lot! yczhang
Carl E. Schwartz, M.D. Harvard Medical School Director, Developmental Neuroimaging & Psychopathology Laboratory Psychiatric Neuroscience Program Massachusetts General Hospital tel 617-726-8965 fax 617-726-4078