Monte Carlo Simulations of a Non-Invasive Positron Detector to Measure the Arterial Input Function for Dynamic PET

This work presents Monte Carlo (MC) study of a novel non-invasive positron detector, hereinafter called NID, designed to measure the arterial input function (AIF) through the wrist of a patient for use with dynamic positron emission tomography (PET). The goal of the study was to optimize a previously developed NID prototype, to determine its efficiency and ability to distinguish between the arterial and venous portions of the signal escaping a wrist phantom. A user code based on the Geant4 MC toolkit was developed to model the NID and a wrist phantom. The scintillator based detectors were modelled as 64 polystyrene cylinders, 0.97 mm in diameter, 10 cm long and capped with cylindrical photomultiplier tubes. The scintillator fibers were arranged in a single band around a 64.13 mm polyethylene cylinder representing a model of the wrist. Two cylinders, 2.30 mm in diameter were placed 6 mm apart, 2 mm below the surface of the wrist phantom representing the radial artery and vein. Two simulations were performed by placing 100 million decay events of oxygen-15 (¹⁵O) or fluorine-18 (¹⁸F), randomly distributed between the artery and vein. Visible wavelength photon tracking was enabled, and Photomultiplier tubes were simulated to collect the visible photons. Deposited energy per event and location of energy deposition were calculated. A python algorithm was used to analyse the results. The arterial signal produced a 5.28 mm and 10.32 mm FWHM for ¹⁵O and ¹⁸F respectively. The algorithm could determine the correct location of interaction 98% of the time. The NID can resolve the arterial and venous signal and is thus suitable for determining the AIF for dynamic PET.