高光谱成像
化学计量学
丙烯酰胺
化学成像
化学
红外线的
食品科学
遥感
色谱法
光学
地质学
共聚物
物理
有机化学
聚合物
作者
Carlos Miguel Peraza-Alemán,Ainara López-Maestresalas,Carmen Jarén,José Ignacio Ruíz de Galarreta,Leire Barandalla,Silvia Arazuri Garín
出处
期刊:Food Chemistry
[Elsevier BV]
日期:2025-03-09
卷期号:479: 143794-143794
被引量:20
标识
DOI:10.1016/j.foodchem.2025.143794
摘要
This study investigated the potential of near-infrared hyperspectral imaging (NIR-HSI) for the prediction of acrylamide content in potato chips. A total of 300 tubers from two potato varieties ( Agria and Jaerla ) grown in two seasons and processed under the same frying conditions were analysed. Partial Least Square Regression (PLSR) and Support Vector Machine Regression (SVMR), combined with a logarithmic transformation of the acrylamide levels, were applied to develop predictive models. The most optimal outcomes for PLSR yielded R 2 p : 0.85, RMSEP: 201 μg/kg and RPD: 2.53, while for SVMR yielded R 2 p : 0.80, RMSEP: 229 μg/kg and RPD: 2.22. Furthermore, the selection of significant wavelengths enabled an 87.95 % reduction in variables without affecting the model's accuracy. Finally, spatial mapping of acrylamide content was conducted on all chips in the external validation set. This method provides both quantification and visualization capabilities, thus enhancing quality control for acrylamide identification in processed potatoes. • Acrylamide content was effectively predicted by PLSR. • The combination of 1D-SNV-PLSR in the full spectrum yielded the best outcomes. • Wavelength selection algorithms maintained the accuracy of the models. • Maps of acrylamide content were obtained.
科研通智能强力驱动
Strongly Powered by AbleSci AI