Gross count multi-spectra analysis of weak activity sources amid elevated background and various shielded conditions

Abstract

Current methods to detect low activity radiological sources in areas with elevated or changing background may be insufficient for assessing real-time events. Background radiation fluctuates throughout the day, which thwarts the effectiveness of decision thresholds (y*) extrapolated from long background count times. However, decision thresholds calculated from a string of spectral measurements can be used to optimize the decision threshold (y*) at which count rates statistically exceed background. The goal of this project is to apply a frequentist string data analysis tool to real-time gross count sodium iodide (NaI) spectra under varying background radiation and shielding conditions. A sodium iodide scintillator is utilized with the ProSpect acquisition software to acquire gross count data for analysis of the source detection efficiency under varying background and shielding conditions. A series of measurements is taken using short count times, and assessed using Receiver Operating Characteristics (ROC) curves. The best combination of N measurements, with n events individually exceeding a discriminator threshold, is determined using true positive rates from the generated ROC curves. Ideally, the optimized combination of n exceedances of the discriminator threshold in N measurements discerns weak signals from elevated background, while minimizing false positive events.