Interferometric terahertz imaging for detection of lethal agents using artificial neural network analyses Aparajita Bandyopadhyay, Alexander M Sinyukov, Amartya Sengupta, Dale E Gary and John F Federici Dept of Physics, New Jersey Institute of Technology, Newark, New Jersey 07103, United States of America, Address for Correspondence: n Robert B Barat Otto York Department of Chemical Engineering, New Jersey Institute of Technology, Newark, New Jersey 07103, United States of America Zoi-Heleni Michalopoulou Department of Mathematical Sciences, New Jersey Institute of Technology, Newark, New Jersey 07102, United States of America Abstract-Interferometric imaging is a non-invasive, non- contact method to detect concealed lethal agents employing spectral imaging in the terahertz (THz) range. Parallel to the experimental testing over short range, extensive modeling simulates reconstructed images of lethal agents at different frequencies applying interferometric techniques. Near-field correction to such imaging has been accounted for and edge probe algorithm and consequent artificial neural network analyses identify the agents under consideration. This work addresses the issues related to THz imaging for rapid and successful recognition of lethal agents in security screening. Index Terms Terahertz imaging, interferometric arrays, artificial neural networks, kohonen self organizing maps, detection of concealed agents I. INTRODUCTION In the wake of the acute sense of vulnerability to concealed threats, a primary focus of national security is to develop non-obtrusive, yet highly reliable schemes, for monitoring and detection of different lethal agents. Lying between the microwave and infrared region of the electromagnetic spectrum between 0.1-10 THz, terahertz radiation bridges the gap between photonic and electronic devices and offers a large expanse of unused, unexplored bandwidth. Its ability to penetrate non-metallic common materials, such as clothing, plastic, ceramic, wood, and identify hidden objects, such as plastic explosives, chemicals and other materials beneath clothing and in packages[1,2,3], presents an opportunity to use THz technology in the areas of security and defense. In future, therefore it can be expected that, THz technology would be able to complement and enhance existing and emerging techniques to increase operational effectiveness in the areas of security screening [4,5,6]. In the present study, a novel screening method has been discussed as shown in Figure 1, which combines the spectral imaging capability of THz waves through characteristic transmission or reflection spectrum of explosive agents[7] and interferometric imaging technique[8] which provides spatial detection of such lethal agents using only a few number of detectors. In addition to the experimental testing[9] of this THz imaging system, extensive simulation yields spatial composite images of agents at different frequencies, providing spectral contrast. These images are simulated based on the spectral data obtained in THz time domain spectroscopic system and different interferometric detector configurations. The consequent processes of artificial neural network (ANN) analyses, to achieve positive identification of lethal agents out of the composite images, are carried out using two distinct methodologies, namely the Multilayer Perceptron (MLP) model and Kohonen Self Organizing Map (KSOM). Fig.l(Color online) Proposed THz imaging system for security screening II. INTERFEROMETRIC THz IMAGING The proposed imaging interferometer consists of an array of individual detectors arranged non-periodically. As the wavefront of reflected (or transmitted) THz radiation encounters the array, each pair of detectors measures one spatial Fourier component of the incoming THz radiation as determined by the separation of the detector pair, otherwise known as a base line. In order to obtain an interferometric