Research On Compton Imaging Base On A 3-D Position Sensitive CdZnTe Detector

The three-dimensional position-sensitive CdZnTe detector can provide good energy resolution under room temperature operating conditions,and also provides threedimensional position information of multiple interactions simultaneously.Therefore,it is widely used in many fields related to radiation detection and gamma-ray imaging.According to the kinematics of Compton scatter interaction,a set of possible positions of the radiation source can be confined on the surface of a cone,whose axis determined by the opposite direction of scattered photon,and half angle determined by Compton scattering angle.Many cones overlap to obtain the image of source.In this dissertation,we built the first fully-integrated Compton camera prototype in China,the imager consist of a 2.0cm × 2.0cm × 1.5cm CdZnTe detector with 121 square pixel anodes and a planar cathode,the size of anode pixel is 1.72 mm.The energy resolution of the whole detector is 1.2%(FWHM@662keV,the same below)for single pixel events after depth correction,the best one among all pixels reaches 0.9%.The position resolution is 1.72 mm in the horizontal direction,and approximately 1mm in the depth direction.Compared to the traditional Compton camera design with two detectors,this camera has a full 4? field of view,higher detection efficiency,and ease of expansion for portable application scenarios.The analytical analysis of the camera's angular resolution shows that for this Compton camera prototype,the final angular uncertainty caused by the energy measurement resolution is less than 10°,the final angular uncertainty caused by the position measurement resolution is about 25°,the final angular uncertainty caused by Doppler broadening is about 10°,and the vertex error is less than 1° in the far field approximation.The imaging algorithm used in this work are the widely used direct back-projection(BP)and list-mode data maximum likelihood expectation-maximization algorithm(list-mode MLEM).The angular resolution of the reconstruction of a point source is 36° and 10°,respectively.Since the list-mode MLEM requires the progressive calculation of the system matrix as the measurement progresses,and real-time imaging cannot be achieved,we have developed a maximum likelihood method based on rearranging the detector response space(rebin-MLEM).Thereafter the calculation of system matrix can be performed by simulation in advance,we can perform the iterative processes directly during the collection process,and real-time iterative reconstruction of images is realized.Sometimes two-dimensional Compton imaging based on far-field approximation is not enough for localization of radioactive sources,especially in complicated application scenarios,so three-dimensional Compton imaging must be used.The key point of the application is to use the movement of the detector to provide reconstruction cones with different vertices at different positions.Based on the fact that if we use the method similar to two-dimensional imaging,there will be too many pixels in the image space and then the iterative memory and computation time consume too much,this dissertation proposes a changeable image space based on ordered subset maximum likelihood(OSEM).In this reconstruction method,a preliminary iteration is first performed on a sparse image space,and the sparse image space is divided into rough pixels to obtain approximate location information of the radioactive source,then some local image spaces where the hot spot of the radioactive source is located are selected and zoomed,a non-uniform image space is obtained for next reconstruction to achieve the target accuracy.This method was verified by simulation and experimental data base on our Compton camera system.