Testbed T3: Whole Body Imaging

The objective of the Whole Body Imaging testbed is the creation of an unbiased, academic-oriented testbed for development and evaluation of multi-modal sensors and algorithms for whole body imaging. Specific objectives include: 1) enabling experimentation with new sensing modalities, via sensor configuration and scanning mode optimization; 2) exploring new algorithm concepts, such as model based vs. Fourier inversion, high resolution fused imaging, and automated anomaly detection; and 3) developing approaches to information fusion and adaptive multisensor processing.

This testbed is trying to improve on the current state-of-the-practice whole body imaging systems. An example of such a system is one that is currently capable of detecting many types of materials based on shape (metallic and non-metallic), such as liquids, gels, plastics, metals and ceramics. This system uses two linear antenna arrays and scans through 240 degrees. However, its limitations and potential areas for improvement include: "dead spots," lack of spectroscopic info, limited views, poor penetration through leather and metallic clothing, and no penetration through skin or into body cavities.

Testbed T3 will allow for the development of whole body imaging sensors with multimodal fusion potential. Examples of the distinguishing strengths and capacities of various modalities for imaging sensors are:

• Millimeter-Wave: Penetrates clothing, distinguishes body-worn objects other than flesh (i.e. metal, explosives, water, plastic), active system provides target contour info.
• Terahertz: Spectroscopic responses for explosives characterization, penetrates thin clothing.
• X-Ray Backscatter: Penetrates all concealing layers, dual energy distinguishes foreign materials, ionizing radiation but very low dosage.
• Infrared Thermography: Shows unusual heat patterns on body
• Nuclear Quadrupole Resonance: Non-localizing, but unique explosive discrimination, penetrates through body.

The ultimate goals of the whole body imaging testbed plan are to establish performance metrics for sensor modalities and to develop and evaluate novel inversion and multi-modal object detection algorithms. To that end, two benchmark goals have been identified. First, the acquisition of a precision portal multiaxis sensor array positioning system. This system’s design must accomodate various types of sensors, both separately, for analysis, and together, to test fused sensor information. Furthermore, the system should be built to be flexible for reconfiguration. Second, the testbed should provide access to raw measurement data that allows for specific, modality-based inversion and joint modality construction.

Testbed T3 will be established in Year 3 at Northeastern University.