A Compton Camera for Spectroscopic Imaging from 100keV to 1MeV

Abstract

Compton cameras are a particularly interesting gamma-ray imaging technology because they have a large field of view and rely on electronic rather than mechanical collimation (lead). These systems produce two dimensional, spectroscopic images using data collected from spatially separated detector arrays. A single acquisition contains data to produce image signatures for each radionuclide in the field of view. Application of Compton cameras in field of astrophysics has proven the systems capability for imaging in the 1 to 30MeV range. Other potential applications, in the 100keV to 1MeV range, include nuclear material safeguards and nuclear medicine imaging. A particularly attractive feature for these applications is that the technology to produce a portable camera is now available due to improvements in solid state room-temperature detectors. The objective of this work is to investigate Compton camera technology for spectroscopic imaging of gamma rays in the 100keV to 1MeV range. To this end, accurate and efficient camera simulation capability will allow a variety of design issues to be explored before a full camera system is built. An efficient, specific purpose Monte Carlo code was developed to investigate the image formation process in Compton cameras. The code is based on a pathway sampling technique with extensive use of variance reduction techniques. In particular, the technique of forcing is used make each history result in a partial success. The code includes detailed Compton scattering physics, including incoherent scattering functions, Doppler broadening, and multiple scattering. Detector response functions are also included in the simulations. A prototype camera was built to provide code benchmarks and investigate implementation issues. The prototype is based on a two-detector system, which sacrifices detection efficiency for simplicity and versatility. One of the detectors is mounted on a computer controlled stage capable of two dimensional motion (14x14cm full range with ±0.1mm precision). This produces a temporally encoded image via motion of the detector. Experiments were performed with two different camera configurations for a scene containing a 75Se source and a 137Cs source. These sources provided a challenging test of the spectroscopic imaging capability of the Compton camera concept. The first camera was based on a fixed silicon detector in the front plane and a CdZnTe detector mounted in the stage. The second camera configuration was based on two CdZnTe detectors. Both systems were able to reconstruct images of 75Se, using the 265keV line, and 137Cs, using the 662keV line. Only the silicon-CdZnTe camera was able to resolve the low intensity 400keV line of 75Se. Neither camera was able to reconstruct the 75Se source location using the 136keV line. The camera has a low energy limit imposed by the noise level on the front plane detector's timing signal. The timing performance of the coplanar grid CdZnTe detector design was improved, resulting in a reduction in the full width half maximum of the coincidence timing peak between two detectors from 800ns to 30ns. The energy resolution of the silicon-CdZnTe camera system was 4% at 662keV. This camera reproduced the location of the 137Cs source by event circle image reconstruction with angular resolutions of 10° for a source on the camera axis and 14° for a source 30° off axis. The source to camera distance was approximately 1m. Typical detector pair efficiencies were measured as 3x10-11 at 662keV. The dual CdZnTe camera had an energy resolution of 3.2% at 662keV. This camera reproduced the location of the 137Cs source by event circle image reconstruction with angular resolutions of 8° for a source on the camera axis and 12° for a source 20° off axis. The source to camera distance was 1.7m. Typical detector pair efficiencies were measured as 7x10-11 at 662keV. Of the two prototype camera configurations tested, the silicon-CdZnTe configuration had superior imaging characteristics. This configuration is less sensitive to effects caused by source decay cascades and random coincident events. An implementation of the expectation maximum-maximum likelihood reconstruction technique improved the angular resolution to 6° and reduced the background in all the images. The measured counting rates were a factor of two low for the silicon-CdZnTe camera, and up to a factor of four high for the dual CdZnTe camera compared to simulation. These differences are greater than the error bars. The primary reasons for these discrepancies are related to experimental conditions imposed by source decay cascades and the occurrence of random coincidences which are not modeled by the code.

Description

Keywords

Citation

Degree

PhD

Discipline

Nuclear Engineering

Collections