Development of a decision threshold for radiological source detection utilizing Bayesian statistical techniques applied to gross count measurements

Abstract

Numerous studies have been published using Bayesian statistics in source localization and identification, characterization of radioactive samples, and uncertainty analysis; but there is a limited amount of material specific to the development of a decision threshold for simple gross count measurements using Bayesian statistics. Radiation detection in low fidelity systems is customarily accomplished through the measurement of gross counts. Difficulties arise when applying decision techniques to low count rate data, which are restricted by the fact that decisions are being made on individual gross count measurements alone. The investigation presented demonstrates a method to develop a viable Bayesian model to detect radiological sources using gross count measurements in low fidelity systems. An integral component of the research is the process required to validate a Bayesian model both statistically and operationally in Health Physics. The results describe the necessary model development, validation steps, and application to the detection of radiological sources at low signal-to-background ratios by testing the model against laboratory data. The approach may serve as a guideline for a series of requirements to integrate Bayesian modeling (specifically, an interaction model) with radiation detection using gross counts in low fidelity systems.