Speciation of Listeria via pyrolysis/gas chromatography

Jeffrey P. Donohue, William J. Welsh

Research output: Contribution to journalArticle

1 Scopus citations

Abstract

Listeriosis is a serious food-borne illness. Traditional biochemical methodology requires 5-7 days to confirm the presence of Listeria monocytogenes, the causative agent. Pyrolysis/gas chromatography offers a rapid analytical technique, which generates a prodigious amount of data. Numerous studies characterizing bacteria by pyrolysis have been performed in various laboratories to date. Historically, many of these studies have targeted specific biochemical compounds or processes for subsequent research. The current study has taken a different approach. In this research, the pattern generated by the pyrolysis of various Listeria species (L. innocua, L. ivanovii, L. monocytogenes, and L. seeligeri) is analyzed without regard to the biochemical basis for this pattern. The goal is to speciate these organisms by computational analyses of the pyrolysis pattern. Eighteen pyrolyzates of L. innocua, 10 pyrolyzates of L. ivanovii, 10 pyrolyzates of L. monocytogenes, and 32 pyrolyzates of L. seeligeri were chromatographed on a 30 m DB-5 column, using flame ionization detection (FID). Fifteen pyrolyzates of L. innocua, L. monocytogenes, L. seeligeri, each and 16 pyrolyzates of L. ivanovii were chromatographed on a 30 m DB-17 column with FID. The resulting signal data were processed and then converted via a Daubechies 8 wavelet transform into a data set of significantly smaller size. These data sets were presented as input to an optimized artificial neural network (ANN) functioning as a pattern recognition tool. The ANN (a probabilistic neural network) achieved correct identifications for over 97% of the organisms chromatographed on the DB-5 column. While a similar ANN correctly identified the organisms chromatographed on the DB-17 column at a rate of over 80%.

Original languageEnglish (US)
Pages (from-to)221-228
Number of pages8
JournalJournal of Analytical and Applied Pyrolysis
Volume72
Issue number2
DOIs
Publication statusPublished - Nov 2004

    Fingerprint

All Science Journal Classification (ASJC) codes

  • Analytical Chemistry
  • Fuel Technology

Keywords

  • Artificial neural network
  • L. monocytogenes
  • Listeria
  • Pyrolysis
  • Wavelet transform

Cite this