Anatomical-functional correlation using an adjustable MRI-based region of interest atlas with positron emission tomography

Abstract

A procedure is described for combining anatomical information from magnetic resonance imaging (MRI) or computerized tomography (CT) and functional information from positron emission tomography (PET) in a rapid fashion. MRI data are combined with a procedure for the definition, storage, and recall of anatomically based regions of interest. An atlas of standard regions of interest, defined for a set of 18 parallel planes spaced at 6-mm intervals, provides an initial region of interest template for each patient slice. Global adjustments to scale, orientation, and position are applied to obtain an initial match. Individual regions of interest may then be moved, deleted, or redrawn as needed. The ability to store region of interest templates ensures reproducibility of analysis over long periods and introduces a standardization of analysis technique. In 25 brain structures, the mean coefficient of variation in cerebral glucose utilization rate (CMRGlc) measurements among five neuroanatomically trained observers was reduced from 8.1% for manual region of interest definition to 4.0% using the template approach with MRI. Template analysis for space-occupying lesions such as tumors or infarcts is illustrated with PET data from a stroke study, emphasizing the facility for rapid, reproducible analysis of multifunctional studies. MRI-PET matching for a structurally intact caudate nucleus having reduced CMRGlc in Huntington's disease emphasizes the accuracy of anatomical localization required to quantify small structures.