F3-H: AIT Ground Truth Effort

X-ray backscatter (XBS) and millimeter wave (mmW) whole body imaging are currently deployed worldwide to screen individuals at security screening checkpoints. The detection of impermissible objects carried by passengers via XBS and mmW screening shares many commonalities with medical radiological screening and diagnosis. A primary objective of this project is to leverage radiological imaging science onto the whole-body screening problem Further, application of radiological imaging science may also lead to metrics which could be applied to all XBS and mmW equipment. An assumption which underlies this study is that improvements in the ease with which objects of interest can be visualized in XBS and mmW datasets will lead to a corresponding increase in the probability of detection and a decrease in the probability of false alarm. For this study we utilized XBS and mmW datasets acquired by Sandia National Laboratory for this purpose. The dataset was specifically acquired to permit comparison of coupons present on given individuals as imaged by equipment from four vendors. The dataset consists of ~1500 distinct combinations of (1) person, (2) coupon (i.e. object from a defined list), and (3) placement of coupon on body (again, from a defined list of placements); each combination is designated as a “case” (neither the coupon list nor coupon placement list is contained in this report.) Each case was acquired in prompt sequence on each of four TSA-certified systems (two XBS-type and two mmW-type) that were employed in the manufacturer-recommended workflow. Within this overall dataset, we examined 30 selected cases in detail, which contained 30 separate placements of 74 total distinct coupons. This data was examined by four independent research scientists in Massachusetts General Hospital’s Department of Radiology. This work was performed at the SSI level, and an SSI final report was generated and sent to the ALERT Program Manager.