Point-source detection using 3D-position-sensitive semiconductor detectors with estimated background

In the detection of radiation sources, the combination of spectral and directional information can improve detection over using either one alone. Directional information found through probabilistic Compton-imaging methods helps to distinguish point sources from extended background sources. In this work, we use 3D-position-sensitive CdZn Te detectors to investigate the combination of these data for the automated detection, identification, and localization of point sources in a background where the spectral shape is known but its intensity may be unknown. If it is available, knowledge of the expected background energy distribution is shown to improve detection and identification performance compared with an unknown background. Such background estimates may be found from prior measurements or background tracking. Detection performance is compared with simple spectral methods as a function of false-alarm probability using experimental data from an 18-detector array system.