Optical spectroscopy enhancing acrylamide sensing in French fries production increasing food safety

Carcinogenic contaminants in food are typically detected using sample-based and time-consuming chemical analysis. To offer a non-destructive alternative, we investigate the use of optical sensing methods for food safety evaluation, particularly considering broadband diffuse reflection spectroscopy and spatially resolved spectroscopy. As a case-study, the acrylamide sensing during potato processing is considered. For both spectroscopic methods, the broadband spectra (400-1700 nm) are recorded for different potato batches, showing acrylamide concentrations between 200 ppb and 2000 ppb, covering levels below and above the European regulation of 500 ppb. First, the raw spectra are evaluated enabling to identify the key spectral regions, being 400-525 nm, 750-1000 nm, 1220 nm, and 1440 nm, corresponding to different potato constituents such as riboflavin, starch, reducing sugar and water. Following, different data processing algorithms were applied maximizing the classification performance of the different batches. When considering the reflection spectroscopic data, a classification performance exceeding 92% could be obtained using Linear Discriminant Analysis, while with spatially resolved spectroscopy, a classification performance around 75% was obtained when considering the ratios of the diffuse and specular reflected light. Both techniques show a harvest-independent result, while coping with the large natural variation by considering the combined effect of different acrylamide precursors. In general, our results demonstrate that optical spectroscopy enables a successful acrylamide precursor sensing, allowing an accurate exclusion of potatoes unsuited for French fries production, complying with the European regulations.