Sensors based on laser-induced fluorescence (LIF) enable the rapid detection of micron-size airborne pathogens. As with any sensor, the central design issue is the trade between sensitivity and selectivity. In the case of a LIF bio-particle sensor, the objective is to best distinguish a small concentration of "threat" particles against a potentially much larger concentration of harmless "background" particles, without an excessive rate of falsely alarming when threat particles are absent. In this paper, we characterize sensor performance using four inter-related metrics -- sensitivity, probability of detection, false positive rate (FPR) and response time. We develop several sensor design principles and present a new approach to signal processing called the "degree of threat" algorithm. We describe a recent experiment quantifying the performance of a BioLert testbed in distinguishing a biological agent (Bacillus globigii spores) from a mineral dust (kaolin), using a receiver operating characteristic (ROC) curve to show the trade between sensitivity and FPR.
© (2004) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
