Low Level Radioactivity Measurement using Bayesian Method

Abstract

The paper introduces original Bayesian algorithm developed by the CEA LIST for the measurement of low-activity uranium contaminations using high-resolution gamma-ray spectrometry based on a high purity germanium diode detector. Such measurement indeed provides access to an indirect estimation of surface activity, assuming that the ratio between the number of alpha particles to be quantified and the number of gamma-rays that are detected is known. The Bayesian approach allows to lower detection limits in low count rates and exploit a richer time-energy information structure than the algorithms used in conventional detection procedures. The performance evaluation and characterization of Bayesian statistical tests is performed using classical receiver operating characteristic curves by comparison to frequentist hypothesis tests. The results indicate that the Bayesian approach, in conjunction with HPGe detector has a superior detection performance of the low-activity uranium contamination up to 50% than that achieved within the frequentist tests. Furthermore, it ensure a significant compromise between the true detection rate, the false alarm rate and the response time.