ResearchPod Summary
Positron emission tomography (PET) typically relies on two-photon annihilation events. However, ortho-positronium (o-Ps) decays into three photons, which contain valuable information about the local tissue micro-environment. Current methods for imaging these three-photon events are limited by either poor spatial resolution or the requirement for a prompt photon, which restricts the choice of radionuclides. This paper introduces the TRIO algorithm to reconstruct three-photon events more accurately using standard radionuclides like 18F.
The authors formulate the image reconstruction as a Bayesian maximum a posteriori inference problem. The TRIO algorithm unifies three distinct sources of information: time-based trilateration (using photon detection times), energy-based reconstruction (using energy conservation laws), and a physics-informed prior derived from quantum electrodynamics (QED) describing o-Ps decay. By combining these into a single probabilistic framework, the algorithm estimates the most likely annihilation vertex. The authors use the Nelder-Mead optimization method to find the global maximum of the posterior probability density function, as the complexity of the model precludes simple derivative-based solutions.
Using Monte Carlo simulations modeled after the Siemens Biograph Quadra scanner, the authors demonstrate that TRIO significantly outperforms existing methods. The mean position error for TRIO is 1.62 cm, which is approximately twice as accurate as time-based trilateration (3.05 cm) and an order of magnitude better than energy-based reconstruction alone (18 cm). Because TRIO does not require a prompt photon, it is compatible with standard clinical PET tracers and existing time-of-flight PET scanners capable of registering three-photon coincidences.
This work provides a pathway to extract diagnostic markers—such as the three-to-two-photon yield ratio—from standard PET scans. By improving the spatial resolution of three-photon imaging, TRIO enables the potential characterization of tissue microstructure and oxygenation without needing specialized radionuclides or hardware, potentially expanding the functional capabilities of existing clinical PET infrastructure.
AI-generated third-party summary by ResearchPod. Not official content or an endorsement by the paper authors or affiliated organizations.