Maximum-likelihood expectation maximization for dual-tracer PET imaging

• Main Authors: Pfaehler, E., Niekamper, D., Scheins, J. J., Shah, J. N., Lerche, C. W.
• Format: Conference Proceeding
• Language: English
• Subjects: Attenuation; Maximum likelihood detection; Maximum likelihood estimation; Microwave integrated circuits; Neurological diseases; Neurotransmitters; Semiconductor detectors
• ISSN: 2577-0829
• DOI: 10.1109/NSS/MIC/RTSD57108.2024.10658021

Conventional PET imaging is restricted to image a single radiotracer at time. Knowledge about two metabolic processes can be acquired by performing two PET scans sequentially. However, the simultaneous imaging of metabolic targets could give new insights about the interplay of neurotransmitter systems and the causes for neurological disorders. For the simultaneous reconstruction of PET images of two radiotracers, a \beta^{+}-emitter emitting only annihilation photons and a \beta^{+}-\gamma-emitter emitting annihilation photons and an additional prompt \gamma photon are combined. The \gamma-photon is used to identify events originating from the \beta^{+}-\gamma-emitter. However, due to e.g. attenuation the \gamma-ray is often not detected and not all events can correctly be associated with the \beta^{+}-\gamma-emitter. This makes an accurate separation challenging. In this work, we concentrate on tracers emitting a high-energy \gamma-photon (\gt625 \ \mathrm{keV}). In order to separate both tracers, a triple-to-'false double' ratio of detected events belonging to the \beta^{+}-\gamma- emitter is estimated. This ratio is dependent on the location of the annihilation in the field-of-view and the current attenuation map. In contrast to previous work, we estimate the triple-to-'false double' ratio including the attenuation of the prompt \gamma in the object and incorporate this estimation of 'false double' coincidences in the forward-projection of the ML-EM algorithm. For evaluation, we simulate two overlapping cylinders containing the \beta^{+}-emitter { }^{18} \mathrm{~F} and the \beta^{+}-\gamma^{-} emitter {}^{44}\mathrm{Sc}. To assess algorithm performance, we calculate the residual error of the \beta^{+}-\gamma-emitter in the reconstructed \beta^{+}-emitter image. The mean residual error remaining in the \beta^{+}-image is 7.2%. This result demonstrates that our approach can effectively separate both tracers and can reconstruct two separate images iteratively.