Neural Network Analysis of Interferometric Terahertz Images for Detection of Lethal Agents

F. Oliveira and R. Barat, Otto York Department of Chemical Engineering
B. Schulkin, F. Huang, J. Federici, and D. Gary, Department of Physics
New Jersey Institute of Technology

Abstract

A non-invasive means to detect and characterize concealed lethal agents employs spatial imaging of their characteristic transmission or reflection wavelength spectrum in the Terahertz (THz) electro-magnetic range. Artificial neural network (NN) analyses of these THz spectral images provide specificity of agent detection at reduced false alarm rates. Published THz spectra are utilized to generate simulated interferometric images of bioagent contained within an envelope, and a suicide bomber. Both multilayer perceptron and radial basis function NN architectures are used to analyze these spectral images. Positive identifications are generally made, with radial basis function NNs generally yielding superior results.

Introduction

A growing national security challenge is the development of novel methods to monitor, detect, and characterize hidden lethal agents such as plastic explosives strapped to a person or bioagents in envelopes. The method under study here is based on spatial imaging of the characteristic transmission or reflection wavelength spectrum in the Terahertz (THz) electro-magnetic radiation range. Artificial neural network (NN) analyses of these THz spectral images are used to distinguish the hidden agents at low false alarm rates from the backgrounds.

Terahertz radiation spans the far infrared range with wavelengths ~ 0.3 mm. It typically transmits through plastics, paper products, etc. - suggesting its utility for probing inside of clothing, luggage, and parcels. Reflective or transmissive THz spectra are claimed for several explosives [1,2] and bioagent simulants and DNA [3]. The THz spectra for the explosives appear to be distinct from the spectra of human skin and materials such as plastics and cloth [1].

Two potential THz scanning scenarios are envisioned: a transmission mode for screening of mail or parcels and a reflection mode for people. In both cases, multiple THz sources illuminate the target area, object, or person. A novel detection scheme under development [4] and discussed briefly below yields spatial images in the 0.2-3 THz range. These images are based on a series of spatial arrays – one at each characteristic wavelength. Trained NNs then process these image arrays in order to achieve positive identification of hidden lethal agents.

Neural networks have been used to classify strains of microorganisms based on Fourier Transform Infrared (FTIR) absorption spectra [5]. Individual organic components 3 in multi-species mixtures were identified with high accuracy from low signal-to-noise Raman spectra through the use of NNs [6]. In each case, some pre-processing of the complex spectra is needed prior to NN use for positive species identification. This paper specifically discusses our simulated THz interferometric imaging and NN interpretive analyses to positively identify target lethal agents at low false alarm rates.

Click here to download the full paper

Looking for predictive data analytics and neural network software to solve your own problem? Try NeuroSolutions Infinity free for 14-days! NeuroSolutions Infinity is the easiest, most powerful neural network software of the NeuroSolutions family. It streamlines the data mining process by automatically cleaning and preprocessing your data. Then it uses distributed computing, advanced neural networks, and artificial intelligence (AI) to model your data.