Feasibility of rapid multi-tracer PET tumor imaging

Abstract

Positron emission tomography (PET) can characterize different aspects of tumor physiology using various tracers. PET is usually limited to one tracer since there is no explicit signal for distinguishing multiple tracers. We tested the feasibility of rapidly imaging two PET tracers using dynamic imaging techniques, where the signals from each tracer are separated based upon differences in tracer half-life, kinetics, and distribution. Time-activity curves for FDG, acetate, ATSM, and PFSM were simulated using appropriate compartment models, and noisy dual-tracer curves were computed by shifting and adding the single-tracer curves. Single-tracer components were then estimated from dual-tracer data using two methods: principal component analysis (PCA)-based fits of single-tracer components to multi-tracer data, and multi-tracer compartment models estimating single-tracer rate parameters from multi-tracer time-activity curves. PCA analysis of single- and dual-tracer data found that there is information content present for separating multi-tracer data, and that tracer separability depends upon tracer kinetics, injection order and timing. Multi-tracer compartment modeling recovered rate parameters for individual tracers with good accuracy but somewhat higher statistical uncertainty than single-tracer results when the injection delay was >10 min. These approaches to processing rapid multi-tracer PET data may potentially lead to a robust method for imaging multiple aspects of tumor physiology in vivo.