|
|
||||||||
|
|
|||||||||
|
|
Instrumentation & Data Analysis: Data Analysis & ManagementData Analysis & Management Posters |
1 Department of Radiology, University of Utah, Salt Lake City, Utah
1606
Objectives: The ability to image several PET tracers in a single scan would enhance detection and characterization of disease. However, since each PET tracer gives rise to indistinguishable 511 keV photon pairs, differences in tracer kinetics and radioactive decay must be exploited to estimate the PET signal for each tracer. The feasibility of recovering certain measures for 2-3 tracers from a single scan has been established, though advances are necessary for the method to become practicable. This work establishes a theoretical formulation of the signal-separation problem, providing a foundation for the study of improved algorithms and techniques.
Methods: The premise of rapid multi-tracer PET is that tracers obey kinetic constraints which, when using dynamic imaging with staggered injections, provide sufficient information to estimate the relative contribution of each tracer at each timepoint. Two classes of signal-separation algorithms are identified: model-restricted and model-guided. The former directly implements the kinetic constraints, whereas the latter minimally uses the constraints to achieve tracer-separation. Various kinetic models can be used, e.g. compartment models, spectral analysis, or component analysis, and tracer-separation can be applied in projection or image space with various tradeoffs.
Results: Multi-tracer scans of 18F-FDG (metabolism), 62Cu-PTSM (perfusion), and 62Cu-ATSM (hypoxia) in dogs with pre-existing tumors provide example application of the algorithms. In general, SUVs and tracer uptake are robustly recovered, whereas washout differences are sensitive to the tracer combination, timing, and algorithm used.
Conclusions: Signal-separation algorithms based upon this framework are broadly applicable to many tracer combinations, and may potentially provide improved multi-tracer imaging performance.
| ||||||||||||||||||||||||||||||||||||||