Precision measurements of brain oxygen utilization with positron emission tomography

Abstract

Oxygen utilization is important for studies of brain metabolism, alongside other measurements such as for glucose metabolism. Oxygen and other measurements with [15O] tracers and PET, however, are significantly more challenging than measurements of [18F]fluorodeoxyglucose, the standard for probing tissue glucose metabolism in vivo, in part due to the much shorter radioactive half-life of [15O]. This work examines details of precision measurement of [15O] tracers and their kinetics. Investigations of arterial input functions (AIFs) and image-derived input functions (IDIFs) have figured prominently for PET, but [15O] tracers are rarely studied given the small numbers of PET facilities equipped to work with these tracers, particularly in their inhaled form. Estimates of IDIFs and AIFs for [15O] tracers have demonstrably distinct characteristics arising from instrumentation as well as circulatory physiology. To reconcile IDIFs and AIFs, we developed a generalizable model for bolus tracer transport, corrected for known effects of instrumentation for measuring IDIFs and AIFs, and found intravascular [15O]CO to be especially suited for constructing a robust recovery coefficient for IDIFs compared against AIFs. Within a Bayesian framework for posterior estimation and estimating data evidence, IDIFs provide parameter estimates compatible with AIFs in the setting of biological variability. IDIFs also provide data evidence that exceeds that of results from AIFs. These suggest that scalar recovery coefficients may be adequate to estimate partial volume effects, and that the circulatory consistency of internal carotid IDIFs with brain tissue perfusion provides greater precision than what can be estimated using radial artery AIFs, which exhibit greater variability of recirculation wave forms.