Simultaneous Detection of Multiple Adulterants in Ground Roasted Coffee by ATR-FTIR Spectroscopy and Data Fusion

This paper proposes a novel screening method for the simultaneous detection of four adulterants (spent coffee grounds, roasted coffee husks, roasted corn, and roasted barley) in ground roasted coffee using partial least squares discriminant analysis (PLS-DA) with mid-infrared spectroscopy. Two diffe...

Full description

Saved in:
Bibliographic Details
Published in:Food analytical methods 2017-08, Vol.10 (8), p.2700-2709
Main Authors: Reis, Nádia, Botelho, Bruno G., Franca, Adriana S., Oliveira, Leandro S.
Format: Article
Language:English
Subjects:
Adulterants
Analytical Chemistry
Attenuation
Barley
Chemistry
Chemistry and Materials Science
Chemistry/Food Science
Coffee
Corn
Data integration
Discriminant analysis
Food Science
Fourier transforms
Infrared analysis
Infrared spectra
Infrared spectroscopy
Least squares method
Microbiology
Multisensor fusion
Reflectance
Spectroscopic analysis
Spectroscopy
Spectrum analysis
Test sets
Training
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper proposes a novel screening method for the simultaneous detection of four adulterants (spent coffee grounds, roasted coffee husks, roasted corn, and roasted barley) in ground roasted coffee using partial least squares discriminant analysis (PLS-DA) with mid-infrared spectroscopy. Two different acquisition modes (attenuated total reflectance, ATR, and diffuse reflectance, DR) are compared. Two recent chemometric approaches, hierarchical models (HM) and data fusion (DF), were employed in order to improve model performance. First level models provided discrimination between unadulterated and adulterated coffee samples, whereas second level models were able to identify the presence of each specific adulterant. The use of DF decreased the percentage of misclassified samples for the first level models from 19.6/14.7% (DR) and 7.5/14.5% (ATR) down to 2.5/4.5% considering the training/test sets. The percentage of misclassified samples in the second level models went as low as 0% (DF—spent coffee, training set). The proposed method is simple, fast, reliable for detecting adulteration in coffee samples, and capable of identifying these adulterants, even when in complex mixtures containing other adulterants.
ISSN:1936-9751
1936-976X
DOI:10.1007/s12161-017-0832-3