PET Imaging

Following administration of a positron-emitting radiopharmaceutical (tracer), PET permits the direct measurement of the four-dimensional radioactivity profile throughout a 3D object over time. Depending on the characteristics of the tracer, physiological parameters can be estimated, such as blood flow, metabolism, and receptor concentration. These measurements can be made with subjects in different states (e.g., stimulus or drug activation), used to compare patient groups to controls, or to assess the efficacy of drug treatment.

The goal of PET tracer kinetic modeling is to devise a biologically validated, quantitatively reliable, and logistically practical method for use in human PET studies. Animal studies are typically performed to characterize the tracers, followed by initially complex human studies, typically leading to the development of simplified methods, e.g., using continuous tracer infusion. Mathematical methodology includes linear and non-linear differential equations, statistical estimation theory, methods to avoid the needs for arterial blood measurements (the input function) such as blind deconvolution, plus the development of novel rapid computational algorithms.

Proper characterization of the PET image data is essential for modeling studies. This requires accurate and carefully characterized corrections for the physics and electronics of coincident event acquisition. Studies of these effects are performed with phantom measurements made on the scanner.

A critical component in the application to real data is the correction for subject motion, particularly as the resolution of modern machines has improved (better than 3-mm in human brain machines). Both hardware and software approaches are employed to address these issues. To produce accurate images with minimum noise, a statistically-based iterative reconstruction algorithm is necessary. Developments in this area include the mathematical aspects of algorithm development, the computer science issues associated with a large cluster-based algorithm, the incorporation of the physics and motion correction, the use of prior information provided from MR images, and the tuning and characterization necessary for practical application for biological studies. The ultimate goal is the combination of the tracer kinetic modeling and image reconstruction to directly process a 4D dataset into parametric images of the physiological parameters of interest.

In whole-body applications, corrections for respiratory and cardiac motion become important. Lack of correction for these motions will blur tumor activity and dramatically affect quantitative interpretation of cardiac uptake of tracers. Thus, full analysis of dynamic PET data can be viewed as a 6-D problem, with cardiac and respiratory cycles as the 5th and 6th dimensions.

PET neuroreceptor studies have focused on determining changes in receptor concentration as a function of disease or measurement of receptor occupancy by drugs. A more recent approach provides an estimate of changes in synaptic neurotransmitter concentration. This method determines the change in tracer binding levels after administration of behavioral or pharmacological stimuli that affect neurotransmitter levels. With careful experimental design and appropriate mathematical modeling techniques, the change in radiotracer binding can be attributed to changes in the level of synaptic neurotransmitter that competes with the radiotracer for receptor binding. Such changes have been successfully demonstrated in the dopaminergic, muscarinic, and serotonergic systems.