I've been trying to determine cerebellar volumes using FIRST but I believe
the volumes are being underestimated.
The volume seems to be underestimated around the perimeter (grey matter) of
the cerebellum. Slicesdir appears to show a reduced volume mask over the
brain (at the cerebellum). Similarly, first_roi_slicesdir illustrates the
cerbellum region (and boundary region) being smaller than they should be. I
viewed my T1 3D brain with the brain mask used in first_flirt and the mask
does appear to cover the cerebellum, ie no brain tissue appears to be
outside the mask.
Does anyone have any advice on how i might modify my processing steps to
improve the precision of the cerebellar segmentation (my data has been bias
corrected already)? My T1 3D is high resolution, having been acquired with
0.51x0.51mm voxels (512 x 512 matix), is it possible this could be
detrimental in trying to segment the finer extremities of the cerebellar volume?
Best wishes
John McLean
|