Performance of long rectangular semi-monolithic scintillator PET detectors.

Abstract

High-sensitivity and high-resolution depth-encoding positron emission tomography (PET) detectors are required to simultaneously improve the sensitivity and spatial resolution of a PET scanner so that the quantitative accuracy of PET studies can be improved. The semi-monolithic scintillator PET detector has the advantage of measuring the depth of interaction with single-ended readout as compared to the traditional pixelated scintillator detector, and significantly reducing the edge effect that deteriorates the spatial resolution at edges of the detector as compared to the monolithic scintillator detector if a long rectangular semi-monolithic detector is used. In this work, depth-encoding PET detector modules were built by using long rectangular semi-monolithic scintillators and single-ended readout by silicon photomultiplier (SiPM) arrays. The performance of the detector modules was measured. The rectangular semi-monolithic scintillator detector has an outside dimension of 11.6×37.6×10mm3 and consists of 11 polished lutetium-yttrium oxyorthosilicate (LYSO) slices measuring 1×37.6×10mm3 . The enhanced specular reflector (ESR) was glued on both cross-sectional surfaces of each crystal slice. For the face opposite to the SiPM array and the two end faces of the detectors, surface treatments with and without black paint were implemented for performance comparison. The bottom face of the semi-monolithic detector was coupled to a 4×12 SiPM array that was grouped along rows and columns separately into 16 signals. The four row signals were used to identify the slices, and the 12 column signals were used to estimate the y (monolithic direction) and z (depth direction) interaction positions. The detector was irradiated at multiple positions with a collimated 511keV gamma beam. The collimated beam was obtained with electronic collimation by using a 22 Na point source and a reference detector. The estimated width of the gamma beam is around 0.5mm. The flood histogram for crystal slices was measured by using the center of gravity (COG) method. The COG method and the squared COG method were used for y position estimation. The standard deviation of the column signals, the ratio of maximum to the sum of the column signals, and the sum of squared column signals were used for z position estimation. All slices were clearly resolved from the measured flood histograms for both detectors with different crystal surface treatments. The estimated y positions roughly linearly change with the true positions at the middle of the detector until ~5mm from both ends of the detector. The y and z spatial resolutions of the detectors were estimated for all middle positions located more than 5mm from both ends of the detector. The squared COG method provides better y position resolution than the COG method. The three z estimation methods provide similar depth of interaction (DOI) resolution. Surface treatment with black paint significantly improves both y and z position resolution but degrades the energy and timing resolution of the detectors. The average full width half maxima (FWHM) spatial resolution is improved from 1.77 to 1.07mm in the y direction by using the squared COG method and from 2.71 to 1.55mm in the z direction by using the standard deviation method. The slice-based average energy resolution degrades from 15.8% to 24.9%. The timing resolution of the entire detector module degrades from 596 to 788ps. The performance of rectangular semi-monolithic scintillator PET detectors with two different crystal surface treatments was measured. The detectors provide superior spatial resolution and depth-encoding capability and can be used to develop small animal and dedicated breast and brain PET scanners that can simultaneously achieve high spatial resolution, high sensitivity, and low cost.