Noise properties of simultaneous dual tracer PET imaging

The Cramer-Rao lower bound (CRLB) for one dimensional dual PET imaging of equilibrated tracers was calculated. The CRLB gives us a fundamental limit on the variance one can obtain in simultaneous dual tracer PET imaging when the two tracers are separated based on their difference in tracer half-life. The CRLB depends on the acquisition time, the two tracer half-lives and the ratio of the initial activities of the two tracers. The dual tracer bounds were compared with the bounds found in the single tracer case. Convergence properties of a 2D dual tracer tomographic reconstruction were also investigated. The images were reconstructed using a parametric iterative reconstruction (PIR) algorithm and image quality was compared to the separate mono tracer ML-EM reconstructions. The PIR algorithm was found to recover the two components, however, convergence rates were slower than the mono tracer convergence rates, thus requiring more iterations to reach the same error level. The convergence rate improved drastically for increasing acquisition times. Our analysis can be extended to evaluate the limits of any number of tracer mixtures