Imaging specialists say TSA scanners can miss underwear bombs
Are TSA used at US airports actually effective? According to two imaging specialists who were once faculty at the University of California, San Francisco, the answer is no. Based on their modeling of the scanners' performance, they conclude that an appropriately shaped piece of plastic explosive will be effectively invisible if placed against a passenger's body.
Article location: link to arstechnica.com
From Ars Technica
On the face of it, this article has crackpot written all over it. Both professors are emeritus, or retired, which is often a sign of a faculty member who has stopped participating in the scientific community and started to pursue pet projects. They don't even have any hardware to test. The journal in which their report is published, the Journal of Transportation Security, is only a few years old, extremely specialized, and only publishes four very small issues annually. None of these are good indicators of a rigorous scientific analysis.
But the story is a bit more complex than that. The two former faculty members, Leon Kaufman and Joseph Carlson, were imaging specialists. Kaufman ended up switching to emeritus status because he left UCSF to help start a medical imaging company. And the paper spells out in some detail just how the two managed to model the behavior of the scanners without getting their hands on the hardware.
The actual figures needed to calculate the exposure and effectiveness of these scans are unavailable, according to Kaufman and Carlson. There is, however, enough published information to understand what they term the "performance parameters." So, for example, the National Council on Radiation Protection and Measurements has a publication that indicates the energy spectrum of the X-rays used. "The solid angle for the detectors can be reasonably deduced from photos and floor plans available on the web from the manufacturer of the Rapiscan 1000," according to the paper.
Other information comes from the images of the devices' output that have been made available. These provide an indication of the pixel density and signal to noise ratio of the scanners themselves, which the authors use to calculate the amount of exposure needed for an effective scan.
Combined, these figures provide either hard numbers or informed estimates on the systems' key parameters. To convert these into exposure and performance information, the authors turned to a physics simulation called GEANT4. This software package, put together by a large international team, models the interactions of high energy particles and photons as they pass through matter. The site on which the software is hosted has a set of pages dedicated to describing the testing and validation of its output; it has apparently been used to help model the Large Hadron Collider's CMS detector.
Using this model, the authors calculate the exposure and depth for two of the scanners, based on the need to generate about 55 counts per pixel in the detectors. The results are given in nanoGrays (nGy), a measure of the amount of energy deposited in the material being exposed—in this case, the human body (typically, health problems aren't obvious until the exposure gets into the single-digits of Grays). The low energy scanners deposit most of their energy within the first five centimeters of the surface, but leave over 100nGy there. The higher energy equipment puts much less energy near the surface—about 50nGy—but, by 10cm in, they're depositing more energy than the lower-powered versions.
Overall, these echo the thoughts of their former peers at UCSF: the total exposure is pretty low, but it's concentrated near the surface and goes much deeper than the layers of the skin that are occupied by dead cells.
On the detection side, they simulated the images that could be produced using the backscattered photons. Here, the big problem was explosives like TATP and PETN. They do have different densities than human tissue, but the differences are not that easy to spot. As a result, detection of these substances is easiest if they're molded into a form with sharp edges—which happens to be what most of the demonstration photos show. If the materials are shaped as a thin pancake with tapered edges, however, they essentially vanish. The authors calculate that up to 320g of PETN could be slipped by security this way—eight times the amount carried by the shoe bomber.
They also note that the front/back scanning process creates a bit of a problem. Metal-rich items hanging at an individual's side will appear dark against a dark background, and might be very difficult to spot. A blade taped to the side of the torso "will be invisible."
The authors conclude that the penetration of X-rays is larger than might be expected, which influences the exposure risk, and the scanners may not do much good anyway. "The penetration not only distributes exposure throughout the body (this affecting the calculation of effective dose, which comprises a sum over all organs), but tends to diffuse the effects caused by contraband materials," Kaufman and Carlson wrote. "The calculated signal excursions at high kilovoltage are so small as to make it doubtful that at any reasonable exposure levels density differences will be noticeable unless the contraband is packed thickly and with hard edges."
The authors build a pretty compelling case that they're probably getting the physics right, and identify a series of cases where the systems are likely to be prone to failure. However, the real test of a model like this is how well it performs in the real world. Hopefully, someone will be able to obtain one of these scanners and find out.
Journal of Transportation Security, 2010. DOI: 10.1007/s12198-010-0059-7 (About DOIs).
contribute to this article
add comment to discussion
view discussion from this article