Positron emission tomography (PET) allows quantitation of human biochemistry that is difficult to accomplish with other noninvasive techniques. However, the spatial resolution of PET is not high enough to resolve many neuroanatomic structures, and therefore central nervous system (CNS) PET studies are subject to quantitative inaccuracies due to large partial volume type effects. In addition, significant differences in spatial resolution capabilities exist between current PET tomographs. In spite of the widespread use of PET, little attention has been given to the influence of PET resolution on image quantitation and study interpretation. Our preliminary studies suggest that neglect of partial volume type effects can lead to errors of up to 100% in quantitative analysis techniques commonly in use with PET. Since inaccurate PET estimates of physiologic parameters hinders understanding of disease or pathology, it seems prudent to establish the importance of this physical factor in PET. Our project goals are: the increased understanding of the effects of resolution on CNS PET studies, improved study design and quantitative accuracy, and the development of general tools for assessing and correcting resolution-induced errors. We will address resolution-induced errors in two of the most important applications of PET: 1) measurements of 18F-2-deoxy-D-glucose (FDG) metabolic rates in normal and tumorous brain tissue and 2) D2 dopamine receptor density measurements. We will develop and validate computer software that will allow us and others to estimate the effects of PET resolution on radiotracer concentration measurements and enable one to perform quantitative corrections. Using this tool, we will correlate errors in measured glucose metabolic rates of normal brain tissue and supratentorial glioblastoma multiforme with different tomograph resolution. This will allow us to establish whether spatial resolution effects are responsible for conflicting PET results of high grade glioma metabolic rates. We will test, for tomographs of different spatial resolution, the accuracy and acceptability of using the controversial """"""""hot area"""""""" method of DiChiro of the NIH to visually assess cerebral tumor malignancy. We will correlate errors in D2 receptor density estimates with PET spatial resolution. For both the FDG and D2 receptor density studies we will define experimental methods (e.g., single scan versus dynamic FDG measurement, region of interest sizing and placement strategies, and subject positioning strategies) that minimize resolution induced errors in PET quantitation. Finally, we will determine whether compartmental modeling and tracer kinetic data can be used, as we recently proposed, as an alternative method to evaluate the effects of limited PET resolution.
