Raytheon's Trimode Imager for Nuclear Detection:
Merging Technologies to Defeat Radiological Threats
Nuclear or radiological terrorism is a growing concern for U.S. national security, driving a need for high performance (high probability of detection and low false alarm rate) standoff detectors for nuclear material. Under the Standoff Radiation Detection System (SORDS) program, Raytheon has developed a Trimode Imager (TMI) that employs three simultaneous modalities — Compton imaging (CI), code aperture (CA) imaging, and spatial information from a non-imaging shadow technology — in a wide field of view system to improve system performance, with an emphasis on driving down the false alarm rate. The research team included Raytheon, national labs, small businesses and universities. The system was designed to detect nuclear or radiological threats by enabling the rapid search of urban or suburban environments using a mobile threat imaging system with unique discrimination capabilities.
Detecting gamma rays from strategic nuclear materials or radiological materials is very difficult in a complex urban environment because of the natural background and environmental sources of gamma rays. To detect a threat (a point source of radiation) in the presence of background (a distributed source of radiation) in a large field of view requires that the field of view be pixilated to achieve a detectable signal to noise. To identify a threat and eliminate natural sources of radiation requires very effective imaging for low and high energy gamma rays. The CA imaging capability of the TMI is very effective for lower energy gamma rays, and the CI capability is very effective for higher energy gamma rays.
Two Technologies Merge for the First Time
The Raytheon TMI for the first time merges these two imaging technologies to achieve an unsurpassed effectiveness in defeating nuclear and radiological threats. The non-imaging shadow technology utilizes the shielding of the detectors by the superstructure of the truck and mounting hardware to give the operator a rough idea of where in the search area an increase in radioactivity may be located.
The development of the Raytheon TMI is the result of a collaboration composed of Raytheon, Los Alamos National Laboratory (LANL), Massachusetts Institute of Technology, University of Michigan, and Bubble Technologies Inc. (BTI). The major components of the TMI are shown with their responsible organizations in Figure 1.
The CA and CI designs are being headed by LANL and BTI, respectively; the Shadow imager is the responsibility of BTI. BTI is also leading the development of the data acquisition and data analysis systems. Data visualization design and development is headed by Raytheon. The digital camera system design and development is also headed by Raytheon. Modeling and simulation of the system is headed by LANL, which is also in charge of the development of all imaging and analysis algorithms and how they interact to yield a fused nuclear and visual image with alarm protocols suitable for display by the data visualization system. BTI contributes to this analysis system with the isotope identification algorithm. Raytheon is responsible for overall system integration, with company directorates leading design, integration and test efforts associated with system architecture, system software, system utilities (power, HVAC, lighting, etc.), and vehicle transportation.
How the TMI Works
Figure 2 shows a rendition of an exploded view of the TMI, with the TMI instrument package mounted in a panel truck. The cutaway shows the two arrays that provide for the TMI’s imaging capability. The front array of square sodium iodide (NaI) detectors serves as the mask for CA imaging and as the first scattering center for the Compton imaging. The back array of rectangular NaI detectors is the location-sensitive plane for both the coded and Compton imaging modes. The location measurement is accomplished using the difference between the amplitudes of the signals in the photomultiplier tubes on each end of the NaI detectors. The energy measurement is accomplished from the sum.
The signals from the photomultiplier tubes are collected by a data acquisition system in the truck and then time-stamped, digitized and labeled with additional orientation and position information. These preprocessed signals are fed into the data analysis system, where the algorithms for constructing the cabinet assembly and CI images operate, seeking a point source of radiation in the moving field of view (FOV) on an event-by-event basis. The isotope identification detector (ID) algorithms operate in conjunction with the imaging algorithms to determine if a point source in the FOV is a threat and is presented on a display in the cab of the truck.
An example of the data visualization system is shown in Figure 3. A Cobalt-60 (Co-60) source inside the building near the window is detected at 25 meters with the TMI traveling at 30 mph. The nuclear images from the CA and the CI, shown in the upper left of the figure, are fused using an algorithm developed specially by the Raytheon team to form a combined nuclear image shown to the right.
Simultaneously, range data is used to determine that the point source is 25 meters from the TMI, which in this example is traveling at 30 mph. The combined nuclear image is used in the isotope ID algorithm, which in this case determines that the point source is Co-60. The isotope ID spectrum is shown in the figure, along with the color code options for display. Co-60 is considered a threat. The color-coded crosshair labeling this as a threat is overlaid on the appropriate digital camera image determined from the geolocation and orientation data and presented to the TMI operator. The TMI user interface adds confidence level, geolocation data, and alarm status along with system health data.
Development of a Unique Algorithm
The TMI team has developed a unique algorithm for fusing the two nuclear images. This fusion algorithm enables unexpected improvement in the performance of the TMI instrument for both lower and higher energy gammas.
Conventional wisdom holds that CA imaging works well for low-energy gammas, whereas Compton imaging works well for high-energy gammas. In a conventional CA system, the mask elements eventually become transparent to higher and higher energy gammas, thus defeating the utility of the mask. In the TMI active mask system, the elements of the mask detect the gammas that strike them. If the gammas are of low energy, they are stopped in the mask element and their energy measured. If the gammas are of higher energy they will Compton scatter in the first array and be stopped in the back array. This enables the Compton scattering imager to operate.
Analysis has revealed a far richer role of these two imaging modes, particularly when their images are fused to form a common nuclear image.
Equipped with validated simulations of imaging in the presence of background, and a TMI system and an algorithm for fusing the two nuclear images, the receiver operation characteristics (ROC) for the active aperture system can be evaluated. To calculate the ROC, one takes a figure of merit (FOM) that reveals the performance of the system — the presence of a point source radiation peak in the field of view. The operational system will have further FOMs such as the shape of the peak. This peak finding FOM is a starting point. The ROC curves for three radiation sources at 100 meters, with the TMI moving at 30 mph, are shown in Figure 4. The left plot is for Cesium-137 (Cs-137), the center plot is for Co-60 and the right plot for the H(n,γ) line at 2.23 MeV. All sources are 1mCi in strength. For each ROC curve, many hundreds of test cases were run with backgrounds randomly varied. For each CA and CI image, a simple peak finding routine searched the FOV, resulting in true positives and false positives. The percentage of true positives is plotted vertically and the percentage of false positives is plotted horizontally.
Results for the CI images are plotted in red, CA results are plotted in blue, and results for the fused image are plotted in green. A random peak finding result would be plotted as a diagonal line from the origin to the upper right-hand corner; an ideal result would be a step function rising from the origin to the upper left-hand corner. In the plot for Cs-137, the CI results shown in red almost mimic this random result. This behavior is not so surprising, given that CI is not expected to do very well at lower energies. The CA results shown in blue are clearly better. However, the surprising thing in the research is that even at these lower energies, the hybrid image results shown in green are better than either imaging mode taken alone.
For higher energy gammas (Co-60), the CI and CA have traded roles as the preferred approach, as expected, and the hybrid image results are almost the ideal step function. Moving higher in energy, the results for H(n,γ) show the CA performance is falling still further behind the CI, while the hybrid image results are nearly ideal. These results demonstrate analytically, for the first time, the superiority of the TMI aperture system over CA or CI imaging systems alone.
This instrument is being developed for the Standoff Radiation Detection System (SORDS) program being conducted by Domestic Nuclear Detection Office (DNDO) of the Department of Homeland Security (DHS) under contract HSHQDC- 08-C-00001.