PEAKSEEK: A Statistical Processing Algorithm for Radiation Spectrum Peak Identification

An accurate analysis of radiation data is essential in many nuclear related applications. In the intelligent model assisted sensing system (iMASS) development, nuclear resonance fluorescence (NRF) spectra of radioactive isotopes are used for detection of nuclear material in cargo containers at US ports. The NRF spectrum of a particular radioactive isotope has a unique signature at unique energy levels. This paper presents a statistical processing algorithm, Peakseek, developed to identify radiation peaks in NRF spectra with a certain degree of confidence. The algorithm tracks the changes in the count rate (theta) of the NRF spectrum and identifies the point of abrupt change in the count rate, i.e., energy level. Identification of abrupt changes in the count rate is performed on the basis of a generalized likelihood ratio statistical test.