DocumentCode :
3759695
Title :
A maximum likelihood gamma positioning method with accurate calibration of the mean detector response function in a clinical SPECT detector
Author :
Zhenlei Lyu; Si Chen; Xinqiang Wang; Lilei Gao; Shi Wang; Tianyu Ma; Yaqiang Liu
Author_Institution :
Key Laboratory of Particle & Radiation Imaging, Ministry of Education and Department of Engineering Physics, Tsinghua University, Beijing, China
fYear :
2014
Firstpage :
1
Lastpage :
4
Abstract :
The objective of this study was to develop a maximum likelihood gamma positioning method (MLGPM) with accurate calibration of the mean detector response function (MDRF) and the statistics of output from PMTs and the next stage data acquisition electronics in a clinical SPECT detector. MLGPM have capability to improve the intrinsic detector performances, e.g. spatial resolution, linearity. One of the key components is accurate calibration of MDRF and statistical model for PMT output. In this study, we employed pencil beam with 5-10mCi 99mTc source to scan the FOV of detector at a grid of 55 × 43 points with 9.9mm separation. At each point, we acquired 200k γ events and generated the spectrums of signals from PMTs and electronics. The photopeaks were analyzed to compute mean and variance of PMT outputs. We optimized aperture size and source to aperture distance to minimize penumbral effects. The mean of PMT output were combined to generate a MDRF. We applied linear regression to mean and variance of PMT output to validate Poisson model and estimate Poisson parameter. We implemented 3 optimization algorithms to achieve MLGPM. Experimental data of two slit phantoms were acquired to validate and evaluate the MLGPM. The results showed that linear correlation coefficients between-the mean and variance of PMT output were in the range of 0.93-0.99, with an average of 0.98. This correlation indicated that Poisson was an accurate approximation of the noise model for PMT output. The Poisson parameters for all PMT were around 400. The MDRF of central PMTs were accurately modeled by truncated 2D Gaussian functions with FWHM ranging from 75mm-90mm. The result of the slit phantom show that MLGPM is comparable with the Anger method. In conclusion, with accuarate calibration of the MDRF and Poisson model, Maximum-likelihood gamma positioning method improves intrinsic performances for a clinical SPECT detector.
Keywords :
"Detectors","Single photon emission computed tomography","Apertures","Maximum likelihood estimation","Calibration","Mathematical model","Maximum likelihood detection"
Publisher :
ieee
Conference_Titel :
Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC), 2014 IEEE
Type :
conf
DOI :
10.1109/NSSMIC.2014.7430928
Filename :
7430928
Link To Document :
بازگشت