Evaluation of false alarm rates of a walkthrough detection portal designed for detecting triacetone triperoxide (tatp) vapour from field test results and receiver operating characteristic (roc) curves

Abstract

We have been developing a “walkthrough detection portal system” with the aim of preventing terrorist attacks or criminal bombings in crowded public places. The portal system consists of a vapour sampler, an atmospheric-pressure chemical ionisation (APCI) ion source, and an explosives detector based on ion-trap mass spectrometry (ITMS). The system can detect triacetone triperoxide (TATP) vapour at a high throughput (1,200 persons per hour). We tested the two-portal systems at the automated ticket gate areas of a train station to obtain background signal data from passengers. The fi eld test results for 9,951 passengers indicated that the false positive rate of the portal system for TATP detection was below 0.01% when multi-marker detection logic was used, although it was about 0.5‐2.0% with single marker detection (m/z 75, 77 or 91). We used the receiver operating characteristic (ROC) curves obtained from the fi eld test results at the train station and the calibration curves of the walkthrough detection portal system to evaluate the relationship between a true positive rate and a false positive rate. The ROC curves for the TATP vapour indicated the true positive rate of the walkthrough detection portal exceeded 99% with the false positive rate below 0.1%, even when only 1 mL TATP vapour was tested. It is concluded from the ROC curves that this detection portal system has suffi cient sensitivity and selectivity for detecting TATP in places where many people come and go.