Adsa Logo White Adsa Title White

Preliminary studies on the use of fluorescence spectroscopy and chemometrics for classification of nonfat dry milk based on spore counts.

C. Qian

Events

06-22-2020

Join C. Qian on this page for a live text chat!
6:00 PM - 8:00 PM GMT

Abstract:

M53
Preliminary studies on the use of fluorescence spectroscopy and chemometrics for classification of nonfat dry milk based on spore counts.
C. Qian*, D. Vega, K. Bonilla, R. Phebus, J. Amamcharla. Kansas State University Manhattan, KS.

Nonfat dry milk (NDM) is a popular ingredient in a wide range of shelf-stable food products. However, high spore containing NDM can lead to ropiness and introduce unwanted lipase and protease activity. The thermophilic and mesophilic spores can enter into raw milk through the cow, feed, and beddings at the farm level. Further, spore counts can increase during the manufacture of NDM due to the concentration factor as well as contamination from the matured biofilms formed on the equipment. Therefore, the spore count is a critical quality indicator to be monitored during production. Previous research suggests that dipicolinic acid (DPA) is present in the core of endospore and can be used as a fluorophore of interest for rapid detection of spores. This objective of this study was to use DPA fluorescence spectra and chemometrics to develop classification models based on the spore levels. Commercial NDM samples (n = 40) were procured within the United States. The reference spore counts (cfu/g of NDM) were obtained by heating reconstituted NDM (10%) at 100�C for 30 min, plated on Tryptic Soy Agar, and incubated at 55�C for 48 h. To release all available DPA and to remove interferents, the reconstituted NDM (10%) was autoclaved at 121�C for 30 min followed by acidification and centrifugation. The terbium chloride was added to the supernatant buffered to pH 5.6 to enhance the DPA fluorescence signal. Emission spectra of terbium DPA complex were collected between 450 and 650 nm fixed at the excitation of 270 nm. Classification models were developed using partial least square quadratic discriminant analysis (PLS-QDA), forward selection quadratic discriminant analysis (FS-QDA), and random forest (RF). It was found that random forest provided the highest mean classification accuracy of 87% while FS-QDA and PLS-QDA showed the mean accuracy at 84% and 83%, respectively (validated using bootstrapping technique). The results suggest the potential of using fluorescence spectroscopy to classify the NDM based on spore counts.

Keywords: Bacillus endospores, classification models, dipicolinic acid.

Biography: Chenhao completed an undergraduate degree in food science at UC Davis and is currently pursuing a master degree at Kansas State University