MODELING AND PREDICTING SECOND and THIRD-ORDER FLUORESCENCE SPECTROSCOPY DATA AS A NOVEL QUALITY CONTROL STRATEGY ON MAYONNAISE

SIANO G G; AZCARATE S; CAMIÑA J; MONTEMURRO M; GOICOECHEA H C

Barcelona

Congreso; XVIII CAC 2016; 2016

Fluorescence spectroscopy has found in food analysis a large use due to that it is fast, gives directmeasurement, is not destructive and noninvasive [1].In this work, the potential of excitation emission fluorescence matrix (EEM) data recorded with a frontface system along with chemometric methods was investigated for the determination of spoilage onmayonnaise and for the rapid prediction of changes of microbial flora (counts of bacteria).Made in home mayonnaise and commercial mayonnaises samples were maintained at 25°C during fourdays, and stored at 37ºC for three days. The microbial load on mayonnaise samples was determined perday. Total viable count (TVC), aerobic mesophilic bacterias, Staphylococcus aureus, Escherichia coliand Salmonella were determined on mayonnaise at each step using culture methods.In parallel, mayonnaise samples were analyzed by 3D front-face fluorescence spectroscopy, withoutprior preparation, at 8 and 6 different times for the samples stored 25 °C, and 37°C, respectively. Thedata matrices were recorded varying the excitation wavelength between 230 and 400 nm each 10 nm,and registering the emission spectra from 300 to 600.5 nm each 0.5 nm. Thus, the EEMs were of size18×602. PARAFAC analysis allowed capturing the changes occurring in the fluorescence spectral data.The best PARAFAC models showed 3 components for data recorded both temperatures. Profiles ofmain compounds were extracted with this algorithm describing quality evolution on the time. In orderto meet those compounds that their concentration decreases and those that are produced (specificallyamino acids) a chromatographic analysis was performed [2]. The chromatographic analysis confirmsthe decrease of tyrosine and the production of tryptophan in the time.Partial least squared discriminant analysis (PLSDA) was applied to data set formed by concatenating ofall the fluorescence spectra at same temperature for testing the allocation of the spectra of theindividual samples within the five and four groups corresponding to the five and four investigatedstorage times. The results showed that 100% of good classifications were obtained using 3 PLS factors.These results allowed the classification of samples as a function of stored time. In addition, thisinformation can be related to the microbial counts.In order to evaluate other quality parameters, a study with optical fibers is being performed. EEMswere obtained through the degradation kinetics of mayonnaise produced by irradiation the sampleseach two minutes for a total of 20 minutes. Third order data analysis will be directed to the fat contentevaluation as well as at the content of different dressing of mayonnaise for obtaining information aboutthe quality mayonnaises.