Spectral analysis of biological agents to implement a tool for fast biological detection

Biosecurity and biosafety are key concerns of modern society. A plethora of biological agents (BAs) like toxins, bacteria, viruses, fungi and parasites, are able to cause damage to humans, animals and plants, and some of them can even be used as non-conventional weapons. Indeed, the use of BAs for military or terroristic purposes has been recorded from the ancient Roman Empire until present days and, today, it is considered one of the four main threat of the CBRN (Chemical, Biological, Radiological, Nuclear) group. One of the most remarkable characteristics of BAs is that they are able to spread unnoticed among the population for days or weeks before the first symptoms of the disease the cause become evident. Moreover, BAs are almost colorless and odorless and their sizes (that ranges from few nanometers for protein toxins to tens of micrometers for bacteria) makes them hardly noticeable to the human eye. Finally, different BAs may cause similar symptoms but require different treatments, so it is paramount to gain the capability to detect and characterize BAs at the early stage of their diffusion. Although the ultimate goal would be the development of a quick, field portable, user friendly tool for the unambiguous identification of BAs, together with the possibility to perform stand-off detection to reduce risks for operators and population, to date, no single technique has proved to achieve these results leading to a constant search for a compromise between rapidity, specificity, and stand-off capabilities. In this framework, UV-Light Induced Fluorescence (UV-LIF) is considered one of the most promising techniques to perform detection from medium to long distances (to date, from hundreds of meters to few kilometers), although photoluminescence from BAs is very weak and day time measurements may represent a real challenge; moreover, background aerosol UV signals poses further issues that are still to be soundly addressed. The PhD research activity consisted of a preliminary analysis (of the characteristics) of a system for the quick classification of BAs released in the atmosphere through the use of PhD in Industrial Engineering - Final Dissertation – Mariachiara Carestia SPECTRAL ANALYSIS OF BIOLOGICAL AGENTS TO IMPLEMENT A TOOL FOR FAST BIOLOGICAL DETECTION 7 UV-LIF techniques, to perform detection to warn and treat. In particular, this thesis reports on new insights on the standardization of the methodology to build and test a training database of BAs signatures, the multispectral analysis of BAs optical signatures, the constitution of a multispectral database, and the statistical and mathematical tools for the automatic classification of the spectral signatures. Moreover, quantum efficiency tests have been performed on biological samples to acquire information useful for the definition of the structural features of a tool for BAs detection. Measurements of different classes of BAs (toxins, vegetative bacteria and bacterial spores), some of them simulants of Biological Warfare Agents (BWAs), have been performed by conveniently selecting six excitation wavelengths in the UV spectral range (namely, 266, 273, 280, 300, 340 and 355 nm). The spectral signatures have been analyzed and different solutions for their automatic classification have been proposed and compared. In particular, Support Vector Regression based methods have been applied and an algorithm has been proposed with the aim of analyzing a restricted database of BAs. This work provides valuable information both for setting the basis for the set-up of a laboratory demonstrator that will be the first step toward the implementation of a tool for quick biological detection able to be installed on an Unmanned Air Vehicle (UAV), and for selecting the criteria for the implementation and analysis of a BAs spectral signatures’ database.