A Weighted-LSM Method to Improve Classification and Concentration Evaluation from Laser-Induced Fluorescence Spectra

Abstract

The detection of biological agents using optical systems is an open field of research. Currently, different spectroscopic techniques allow to detect and classify chemical agents while a fast and accurate technique able to identify biological agents is still under investigation. Some optical techniques, such as Laser-Induced Breakdown Spectroscopy (LIBS) or Laser-Induced Fluorescence (LIF), are already used as classification methods. However, the presence of background, spectrum similarities and other confounders make these techniques not very specific. This work shows a new method to achieve better performances in terms of classification and concentration evaluations. The method is based on the Weighted Least Square Minimization method. In fact, by using ad hoc weights, the LSM looks at specific features of the spectra, resulting in higher accuracy. In order to make a systematic analysis, numerical tests have been conducted. With these tests, the authors were able to highlight the various advantages and drawbacks of the new methodology proposed. Then, the method was applied to some LIF measurements to investigate the applicability of the method to preliminary experimental cases. The results show that, by using this new weighted LSM, it is possible to achieve better classification and concentration evaluation performances. Finally, the possible application of the new method is discussed.

An alternative SNR-based weighted-LSM algorithm to classify and measure the concentration of Biological Agents from Laser-Induced Fluorescence

“…To perform numerical tests comparable with the cited work of Gabbarini et al [1] it was decided to generate the synthetic spectra with the same mathematical methods. Also in this case, two reference spectra (wavelength simulated range from 190 nm to 785 nm with a discretization of 1044 values) have been generated for the database.…”

“…The use of Laser-Induced Fluorescence (LIF) is largely investigated as optical technique to classify the Biological Agents (BAs) in a sample [1][2][3][4]. The fluorescence phenomena are electromagnetic emission originated by a molecule through its excitation with a specific wavelength [4,5].…”

“…LIF is a helpful technique to classify and measure the concentration of BAs. The different combination of fluorophores composing each specific biological agent results in different emitted fluorescence spectra [1]. In this way, the collected spectra contain much information about the composition of the biological sample, and therefore the identities of agents contained inside [1].…”

“…Recently, Gabbarini et al [1] had investigated the possibilities to use the Least-Square Minimisation (LSM) method to classify and measure the BAs in a mixture sample using a reference database. This new method has the advantage that it requires only reference spectra and not mixture ones.…”

“…As in the cited work [1], synthetic spectra to test the performances of the proposed algorithm (W SNR -LSM) in comparison with the other (Classical-LSM and W DIF -LSM) is performed.…”