Sequential probability ratio test using scaled time-intervals for environmental radiation monitoring

Sequential probability ratio test (SPRT) of scaled time-interval data (time to record N radiation pulses), SPRT_scaled, was evaluated against commonly used single-interval test (SIT) and SPRT with a fixed counting interval, SPRT_fixed, on experimental and simulated data. Experimental data were acquired with a DGF-4C (XIA, Inc) system in list mode. Simulated time-interval data were obtained using Monte Carlo techniques to perform a random radiation sampling of the Poisson distribution. The three methods (SIT, SPRT_fixed and SPRT_scaled) were compared in terms of detection probability and average time to make a decision about the source of radiation. For both experimental and simulated data, SPRT_scaled provided similar detection probabilities as other tests, but is able to make a quicker decision with fewer pulses at relatively higher radiation levels. SPRT_scaled has a provision for varying the sampling time depending on the radiation level, which may further shorten the time needed for radiation monitoring. Parameter adjustments to the SPRT_scaled method for increased detection probability are discussed.