TIV-E is a semi-automatic brain extraction technique involving the removal of the skull and non-brain tissue using a tri-feature approach (i.e. using a PD, T2, and T1). The TIV segmentation is automatically created and manual editing is applied where necessary. Using this segmentation, an accurate measure of total intracranial capacity can be determined.
The brain is automatically rotated into Talairach space by aligning the anterior and posterior (ACPC) commissure, the eyeballs, and the longitudinal fissure. This automatic adjustment is checked manually to ensure accurate alignment.
Fifteen landmarks are placed on the ACPC aligned T1 image (e.g. AC, PC, sylvian fissure, etc.) and are used to locate the boundaries for regions of interest. This method allows for an individualized parcellation which acknowledges the variability of the human brain, which is particularly pronounced in elderly and diseased populations.
Ventricular CSF is reassigned in order to differentiate it from the subarachnoid CSF. Using this segmentation, volumetrics can be obtained for grey matter, white matter, ventricle CSF and subarachnoid CSF. The cerebellum is then manually removed to get a measure of the supratentorial space (TIV-ST).
The parcellation segmentation is visually inspected to confirm that the brain has been properly segmented into the 26 SABRE regions (13 per hemisphere).