A directional algorithm for an electronically-collimated radiation detector (ECRD)

Publication Year:
Usage 24
Downloads 18
Abstract Views 6
Repository URL:
Lackie, Adam
back-projection; medical imaging; Compton imaging; imaging; Compton; backprojection
thesis / dissertation description
An electronically-collimated radiation detector (ECRD) is being developed to be used for locating radiation sources, e.g. for intraoperative localization of sentinel lymph nodes, or for public safety applications. The design emphasizes a compact, portable detector with a wide field of view. Typical probes provide either high sensitivity but no directional information when uncollimated, or directional information but poor detection efficiency when collimated. The ECRD design provides high sensitivity to the presence of radiation because it lacks physical collimation, and simultaneously provides directional information using electronic collimation. Intended to be a hand-held device, the ECRD front end comprises an array of cadmium-zinc-telluride (CZT) detectors. An incident gamma ray scatters in the primary detector; interaction of the scattered photon in a secondary detector is detected in coincidence. For each photon, Compton kinematics specifies a cone on which the source must be located. Localization is achieved by finding the intersection of many Compton-scatter cones. This paper reports on the development and evaluation of two directional algorithms for this device, a modified Compton telescope algorithm and an algorithm based on finding the intersections of rectangles circumscribing the Compton cones. The methods developed were evaluated using ideal simulated data from a point source as well as data from a Monte Carlo simulation of an ECRD device. The accuracy, precision and convergence of each directional algorithm were evaluated. It was found that for the modified Compton telescope technique, a useful field of view extending 60º from the forward direction was observed, an angular resolution better than 20º was achieved throughout the field of view, and the method converged to these values around 300 events; the results for the ideal data did not significantly differ from those using the Monte Carlo data. For the circumscribed rectangle technique, the useful field of view covered nearly the entire area in front of the detector for the ideal data but the more realistic physics of the Monte Carlo simulated data shrank the useful field of view to the region within approximately 30º of the forward direction, while the angular resolution was better than 20º and convergence was approached at approximately 50 events.