FWIW, FreeSurfer uses 2 curvature calculation engines. The newer one is based on the Gauss-Bonnet theorem and more robust to noise, albeit slightly slower than the default engine.
I'd suggest comparing the indices using 'mris_curvature_stats' which uses the Gauss-Bonnet engine:
mris_curvature_stats -G <subj> <hemi> <surface>
e.g.
mris_curvature_stats -G subject1 lh smoothwm mris_curvature_stats -G subject1 rh smoothwm
You'll notice that the final measure reported is the so called "natural" intrinsic curvature index. This should be, ideally, 1 for FS surfaces. The distance away from 1 is an indication of how accurate the measures are.
You can run 'mris_curvature_stats' with a '--continuous' flag to force it to revert to the default curvature engine, so
mris_curvature_stats --continuous -G subject1 lh smoothwm mris_curvature_stats --continuous -G subject1 rh smoothwm
will give you mris_anatomical_stats type numbers. It would probably be a good idea to compare the '--continuous' results with the Gauss-Bonnet ones.
On 3/30/11 15:30 , Bruce Fischl wrote:
Hi Karl
it's hard to say without knowing why. Curvature in general is a 2nd derivative, so it is usually a noisy measure.
cheers Bruce
On Thu, 31 Mar 2011, Liukarl wrote:
Hello, Freesurfer experts:
We are extracting the average cortical thickness of each hemisphere for every subject using mris_anatomical_stats. We found that for one subject (attached below), the thickness measures are quite similar but the ¡°intrinsic curvature index¡± differs drastically between left/right hemispheres. Will this curvature index affect the validity of our thickness measure?
***the left hemisphere mris_anatomical_stats -l lh.cortex.label -f subject1l.txt subject1 lh limiting computations to label lh.cortex.label. reading volume /usr/local/freesurfer/subjects/subject1/mri/wm.mgz... reading input surface /usr/local/freesurfer/subjects/subject1/surf/lh.white... reading input pial surface /usr/local/freesurfer/subjects/subject1/surf/lh.pial... reading input white surface /usr/local/freesurfer/subjects/subject1/surf/lh.white... structure is "lh.cortex.label" number of vertices = 157421 total surface area = 102165 mm^2 total gray matter volume = 290073 mm^3 average cortical thickness = 2.488 mm +- 0.694 mm average integrated rectified mean curvature = 0.170 average integrated rectified Gaussian curvature = 0.105 folding index = 4980 intrinsic curvature index = 592.3
***the right hemisphere mris_anatomical_stats -l rh.cortex.label -f subject1r.txt subject1 rh limiting computations to label rh.cortex.label. reading volume /usr/local/freesurfer/subjects/subject1/mri/wm.mgz... reading input surface /usr/local/freesurfer/subjects/subject1/surf/rh.white... reading input pial surface /usr/local/freesurfer/subjects/subject1/surf/rh.pial... reading input white surface /usr/local/freesurfer/subjects/subject1/surf/rh.white... structure is "rh.cortex.label" number of vertices = 158401 total surface area = 102285 mm^2 total gray matter volume = 291519 mm^3 average cortical thickness = 2.482 mm +- 0.707 mm average integrated rectified mean curvature = 0.173 average integrated rectified Gaussian curvature = 0.299 folding index = 5756
Thanks in advance
Karl
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer