Analisi microbiologiche del pesce con spettroscopia infrarossa in trasformata di Fourier ed immagini multispettrali

Gilthead seabream stored at different temperature was analized with Fuorier-transform infrared spectroscopy and multispectral imaging technology. Furthermore, traditional microbiological techniques were used to assess the total viable count, Pseudomonas spp., Brochotrix thermosphacta and Lactic acid bacteria population. On the sample was also assessed the pH and a sensory evaluation was performed. The spectral data collected from the FTIR and from the Multispectral Imaging devices were then correlated with the total viable count in order to find a model suitable for TVC prediction using spectral data as input. Spectral data were pretreated with four different techniques and then used for the construction of Partial Least Square Regression models. With FTIR spectral data, a good PLS regression model was found using SNV pretreatment on data. Multispectral imaging data did not performed well in the construction of the PLS regression model. Further studies may be required in order to find a model with better indices of performance both for FTIR and multispectral imaging data.