7 Oct
2008
7 Oct
'08
3:36 a.m.
Hi all,
We are doing a test-retest study of FreeSurfer segmentation on our dataset. We found reliable estimations for most of the subcortical structures (1% difference) except palladum(3% to 4% difference) while in Fisch's 2002 Neuron paper, pallidum is one of the best segmented structure using manual tracing as gold standard. Any comment? By looking at the aseg, there is no intensity change from pallidum voxels and neighborhood white matter voxels. Is the definition of boundary purely based on the probabilistic atlas?
Thanks, Zheng Hui