We describe a virtually automatic comprehensive parcellation of the human cerebral central white matter, which is based upon T1-weighted MRI scans. The system, which is "rule-based," is developed from prior anatomic studies of the human brain and experimental studies of connectivity in animals as elaborated in the companion manuscript. Boundaries which delineate anatomic subregions of the white matter are computed from the geometric features of anatomic landmarks visible in the imaging data. The fiber systems of the central white matter are ordered topographically into three compartments, reflecting the inferred arrangements of principal neural systems pathways. These include an outer radiate (fibers principally radially aligned), an intermediate sagittal (fibers principally sagittally aligned), and deep bridging (fibers bridging hemispheres or cortex and deep structures) compartments. Each of these compartments is secondarily parcellated into smaller units to increase the anatomic specificity and spatial resolution of the system. The principal intended uses for this system of anatomic subdivision are for the volumetric characterization of forebrain white matter in normal and abnormal brains and for precision and specificity of localization in focal lesion-deficit correlation studies.
Copyright 1999 Academic Press.