1. Count Rate Analysis of GSO-DOI PET Scanners
Keishi Kitamura and Hideo Murayama
Keywords: count rate, Monte Carlo simulation, positron emission tomography (PET), nuclear medicine
A detector made of thick crystals with a depth-of-interaction (DOI) capability can maintain high spatial resolution while increasing system sensitivity in 3D PET scanners with high geometrical efficiency, that is small detector ring diameter and large axial field of view (FOV). Along with sensitivity and spatial resolution, count rate performance is an important factor to affect the image signal-to-noise ratio in clinical PET studies. High light yield and fast decay time properties of Gd2SiO5:Ce (GSO) are suitable for improving energy and time resolution, which result in the decrease in scatter and random coincidences. However, these contributions may be canceled by the dead time losses at detector modules using a large area PS-PMT placed close to the patient. We therefore investigated the count rate properties of GSO-DOI PET scanners with the large PS-PMT using a Monte Carlo simulation and event loss model.
In this work, Monte Carlo simulation programs based upon EGS4 were used to calculate photon interactions in scintillation crystals and phantoms. Photons which have escaped from the phantom were tracked within the scintillator, and their position and energy were recorded if interactions occurred. A pair of photons having the same annihilation tag was counted as a coincidence event, and single events were used to generate random events. Dead-time factors at each stage of the data acquisition system were estimated using a conventional count rate model with the product of paralyzable and non-paralyzable dead time losses for the front-end circuit and non-paralyzable dead time losses for the other circuits. Noise equivalent count rates NECR = T2/(T+S+R) were then calculated as a function of activity concentrations, where T, S and R are the total true, scatter and random coincidence rates, respectively, and is the ratio of the lines of response passing through the object.
The proposed detector unit consisted of four stages of 16 x 16 GSO crystal arrays with a total depth of 30 mm coupled to a 52 mm squire PS-PMT having 16 x 16 multi-anodes. The scanner consisted of 5 ring detector blocks with 24 detectors per ring with a diameter of 38.2 cm and an axial FOV of 25.8 cm. All detector elements were assumed to have the same energy resolution of 20% with a pulse integration time of 250 ns. The outputs of the front-end circuits with an energy window of 400-600 keV were grouped into buckets and presented to the coincidence processors with a time-window of 6 ns. The relative gains of count rates by processing subset signals of the PS-PMT anodes in parallel with additional front-end electronics were also calculated. The results shown in Fig. 1 predict that the scanner will have higher sensitivity compared to current PET scanners mainly due to the high geometrical efficiency, and higher NECR despite the large size of the detector block. Using the 2x2 subdivision can increase the maximum NECR approximately 20% with a slight loss of sensitivity. However, using the 4x4 subdivision reduces sensitivity substantially because there are many escaping incident gamma rays from one segment to adjacent segments by Compton scattering in the scintillator.
Publications:
1) Kitamura, K., et al.: KEK Proceedings 2001-11, pp.108-114, 2001.
2) Kitamura, K., et al.: 2001 IEEE NSS & MIC Conf. Rec., M5A-7, 2001.
|