Detection measures for visual inspection of X-ray images of passenger baggage

Abstract

In visual inspection tasks, such as airport security and medical screening, researchers often use the detection measures d' or A' to analyze detection performance independent of response tendency. However, recent studies that manipulated the frequency of targets (target prevalence) indicate that d a with a slope parameter of 0.6 is more valid for such tasks than d' or A' . We investigated the validity of detection measures ( d' , A' , and d a ) using two experiments. In the first experiment, 31 security officers completed a simulated X-ray baggage inspection task while response tendency was manipulated directly through instruction. The participants knew half of the prohibited items used in the study from training, whereas the other half were novel, thereby establishing two levels of task difficulty. The results demonstrated that for both levels, d' and A' decreased when the criterion became more liberal, whereas d a with a slope parameter of 0.6 remained constant. Eye-tracking data indicated that manipulating response tendency affected the decision component of the inspection task rather than search errors. In the second experiment, 124 security officers completed another simulated X-ray baggage inspection task. Receiver operating characteristic (ROC) curves based on confidence ratings provided further support for d a , and the estimated slope parameter was 0.5. Consistent with previous findings, our results imply that d' and A' are not valid measures of detection performance in X-ray image inspection. We recommend always calculating d a with a slope parameter of 0.5 in addition to d' to avoid potentially wrong conclusions if ROC curves are not available.

Traditional Visual Search vs. X-Ray Image Inspection in Students and Professionals: Are the Same Visual-Cognitive Abilities Needed?

“…Wolfe et al (2007) proposes a z ROC slope of 0.6, which indicates that the noise (target-absent) distribution has a smaller standard deviation than the signal-plus-noise (target-present) distribution. Further publications (Van Wert et al, 2009; Godwin et al, 2010) have reported z ROC slopes similar to those reported by Wolfe et al (2007) while a study reported by Wolfe and Van Wert (2010) found a slope of 0.56 and a study by Sterchi et al (2019) a slope of 0.5 to fit the data more accurately. In our study, data from the basic visual search task (L/T-letter search task) were analyzed under the assumption of an equal variance model using d ′, whereas data from the X-ray image inspection task SBST were analyzed under the assumption of an unequal variance model with a z ROC slope of 0.5 using d a 1 .…”

Visual search behavior and performance in luggage screening: effects of time pressure, automation aid, and target expectancy

“…To complete the performance picture, we also calculated the corresponding performance in terms of signal detection theory measures, i.e., sensitivity (d a , as specified in Sterchi et al 2019 ) and the response criterion C (for calculation see Stanislaw and Todorov 1999 ). As some condition x participant combinations produced perfect hit rates, we loglinear corrected all hit and false alarm rates (Hautus, 1995 ).…”

Assessment of the Effect of Cleanliness on the Visual Inspection of Aircraft Engine Blades: An Eye Tracking Study

“…Pilot training is another well-studied area in eye tracking research [ 77 , 78 , 79 ]. Only a few studies applied eye tracking to scanning and decision-making tasks outside the cockpit, including air-traffic control [ 80 , 81 ] and airport luggage screening [ 82 , 83 , 84 ].…”