Simulation studies of a phoswich PET detector design with a two-fold improvement in spatial sampling

Xuezhu Zhang ,Jonathan D Thiessen ,Andrew L Goertzen ,Christopher J Thompson

doi:10.1109/nssmic.2013.6829093

Xuezhu Zhang , Jonathan D Thiessen + Show 2 more

https://doi.org/10.1109/nssmic.2013.6829093

Copy DOI

Export

Save

Cite

Abstract
Full-Text
Similar Papers

Abstract

Listen

Phoswich detector designs, which combine multiple scintillator materials each having unique light emission properties, have been extensively studied for improving spatial sampling in PET. In this work we propose a detector for PET imaging based on a 2D scintillator array in which each scintillator element in the array is a phoswich pair consisting of two scintillator materials (with different emission decay times) optically coupled together so that scintillation light can freely pass between the two halves of the phoswich pair. A conventional reflector is placed between the scintillator elements. When an N×N crystal array of this design is imaged with a conventional detector (i.e. no decay time discrimination) it will appear to be simply an N×N array, however when the decay time discrimination is enabled, the left/right half of each element can be identified and the sampling becomes 2N×N. In this way, a detector can have spatial sampling of X/2×Y while only needing to resolve crystal elements of size X×Y. This design can readily be extended to multi-layer detectors, in particular dual-layer offset (DLO) designs. The potential improvement in spatial resolution achieved by a DLO detector using this phoswich design was studied using an analytical simulation and fully 3D LOR based reconstruction with system response modeling for a 16 detector small animal PET system with inner ring diameter of 60 mm and a 1.6 mm crystal pitch (0.8×1.6 mm phoswich sub-element size). Four versions of the DLO design were simulated, corresponding to the 4 possible combinations of the top and bottom crystal layer alignments (i.e. parallel and perpendicular to the axial direction). In each case a 20% event misidentification in the phoswich pair was simulated. The best spatial resolution was obtained for the case of both top and bottom layer phoswich elements having their long side aligned with the axial direction, giving the best transaxial sampling. For this case, 0.7 mm hot rods were well resolved.

Full Text