Abstract
FTIR absorbance spectra of four foodborne pathogens suspended in four common food matrices at three different concentrations were used with artificial neural networks (ANNs) for identification and quantification. The classification accuracy of the ANNs was 93.4% for identification and 95.1% for quantification when validated using a subset of the data set. The accuracy of the ANNs when validated for identification of the pathogens studied at four different concentrations using an independent data set had an accuracy range from 60% to 100% and was strongly influenced by background noise. The pathogens could be identified irrespective of the food matrix in which they were suspended, although the classification accuracy was reduced at lower concentrations. More sophisticated background noise filtration techniques are needed to further improve the predictions.
Original language | English (US) |
---|---|
Pages (from-to) | 1249-1255 |
Number of pages | 7 |
Journal | Transactions of the ASABE |
Volume | 49 |
Issue number | 4 |
DOIs | |
State | Published - Jul 2006 |
Externally published | Yes |
Keywords
- ANNs
- Differentiation
- FTIR spectroscopy
- Food matrices
- Food pathogens
- Quantification
ASJC Scopus subject areas
- Forestry
- Food Science
- Biomedical Engineering
- Agronomy and Crop Science
- Soil Science