Eric P. Rubenstein

Low cost, pervasive detection of radiation threats

The recent nuclear crisis at Fukushima, Japan is a stark reminder that radiation emergencies can and do happen. In addition to accidents, the potential use of radioactive materials by terrorists has raised serious concerns. While the primary concern has been with preventing these materials from entering the United States, thousands of dangerous radiological sources are already here within our borders, located in vulnerable locations in hospitals, food processing plants, and industrial sites. These sources pose a risk for use in two terrorist threats described by the Department of Health and Human Services (DHHS): the Dirty Bomb and the Silent Source. In a Dirty Bomb attack, radioactive material is dispersed using a conventional explosive. In a Silent Source attack, radioactive material is hidden in locations where people congregate (restaurants, airports, subway stations, shopping malls, etc.). Both scenarios can injure or kill people and cause significant political, social and economic disruption. This paper will describe the GammaPixTM technology, which has the potential to provide low cost, pervasive detection of, and warning against, radiation threats. The GammaPix technology is based on software analysis of the images produced by a surveillance or smartphone camera to measure the local gamma-ray radiation exposure at the device. The technology employs the inherent gamma-ray sensitivity of CCD and CMOS chips used in the digital image sensors of these devices. This paper describes the use of the technology in calibration and testing scenarios using installed video cameras and smartphone cameras.

Crowd-sourced calibration of uncontrolled radiation detectors

Image Insight Inc. recently released an Android application (GammaPix Lite™) that uses smartphone cameras for radiation detection and measurement. Given the wide diversity of smartphones in use, and the frequent release of new models, traceable calibrations of all the devices on which the app could possibly run is clearly not practical. In this paper, we discuss the methods we have developed to address this problem by using data gathered from multiple instances of each device in uncalibrated use. The quality control mechanisms, reference material, and resulting calibrations are discussed in this paper.