Hyperspectral signature characterization and detection of Escherichia coli (E. coli)

Date

2014-12

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Using a non Shiga toxin producing Escherichia coli, or E. coli such as the lab safe K12 strain I will demonstrate a method to develop signatures and automated detection schemes suitable for hyperspectral data searches. Conventional and reliable laboratory methods remain the mainstay to isolate and identify suspected sessile and planktonic bio-material. however the FDA Food Safety Modernization Act (FSMA), Jan 04, 2011, is requiring additional proactive identification and monitoring. These process cost are passed onto the consumer and not always kept to a competitive minimum. These proposed methods could allow vendors an opportunity for higher screening coverage without the costly laboratory time. These Standard test are laborious and lengthy at times taking a couple days to complete. With recent outbreaks, reported by the CDC, the need for faster identification is again emphasized. Optical, noninvasive techniques such as hyperspectral remote sensing technology has been adapted for microscopic sensing. Many applications have pursued this avenue with varying degrees of success. As the cost of hyperspectral detectors falls, the promise of an optical detection solution is within reach. Distinctive bands between 450 - 900nm are reported in current literature as offering promise. Hyperspectral imagery can consist of tens to hundreds or thousands of discrete bands measured at various wavelengths and widths. Spectral band selection is a fundamental problem as outside laboratory or uncontrolled environments are ripe with sample contamination and unconstrained blending of sample materials.

The goal of this research is to develop a detection method based on hundreds of cells still in their planktonic stage before the damaging effects of their more colonized form, sessile where the biofilm attaches to a solid surface, with cell counts in excess of 10^6 CFU/ml. Once colonies, biofilm, have reached these concentrations their removal is much more difficult as environmental coping mechanisms are fully developed. Their often toxic excretions have yet to be formed while in a planktonic state thus offering a benign infection if detection and eradication is effective prior to colonizations. Our low planktonic count detection will offer higher success treatments and prophylactic actions. The overarching objective is to determine a HSI signature that has a low false alarm rate (Fa) from unstained (low contrast) and stained (high contrast) bacterial samples.

While there are many attempts to clarify the definition of the data set. Most require some domain understanding to follow however one presentation describes HSI as "... sensors acquire reflectance values at many different wavelength bands, to cover a complete spectral interval." The hyperspectral sensor or camera collects imagery with many bands (frames), saving them as individual layers or reflectance bands. This concept has proven to be the most confusing aspect of technical conversation with respect to hyperspectral sensors and data sets and in conservation requires a lengthy and careful explanation.

The objectives to accomplish the stated goals are the following:

  1. determine a HSI signature that has a low false alarm rate from unstained (low contrast) samples.
  2. determine a HSI signature that has a low false alarm rate for stained (high contrast) samples. 3)integrate the two primary algorithms into a single algorithm which uses the strengths of both to create a stronger sum of the two.
  3. propose a software architecture that will accommodate the above and facilitate both testing and production work flows.

I will both demonstrate signature development and deliver a working signature based on data collected November 2012. Signature development and search algorythms are discussed and presented in the form of software source code along with empirical evidence attesting to functional abilities.

Description

Keywords

Hyperspectral, Bacteria, E. coli, Artificial intelligence, Spectral angle mapper

Citation