Aykas Didem Peren, Rodrigues Borba Karla, Rodriguez-Saona Luis E
Department of Food Science and Technology, The Ohio State University, 100 Parker Food Science and Technology Building, 2015 Fyffe Road, Columbus, OH 43210, USA.
Department of Food Engineering, Faculty of Engineering, Adnan Menderes University, Aydin 09100, Turkey.
Foods. 2020 Sep 15;9(9):1300. doi: 10.3390/foods9091300.
This research aims to provide simultaneous predictions of tomato paste's multiple quality traits without any sample preparation by using a field-deployable portable infrared spectrometer. A total of 1843 tomato paste samples were supplied by four different leading tomato processors in California, USA, over the tomato seasons of 2015, 2016, 2017, and 2019. The reference levels of quality traits including, natural tomato soluble solids (NTSS), pH, Bostwick consistency, titratable acidity (TA), serum viscosity, lycopene, glucose, fructose, ascorbic acid, and citric acid were determined by official methods. A portable FT-IR spectrometer with a triple-reflection diamond ATR sampling system was used to directly collect mid-infrared spectra. The calibration and external validation models were developed by using partial least square regression (PLSR). The evaluation of models was conducted on a randomly selected external validation set. A high correlation (R = 0.85-0.99) between the reference values and FT-IR predicted values was observed from PLSR models. The standard errors of prediction were low (SEP = 0.04-35.11), and good predictive performances (RPD = 1.8-7.3) were achieved. Proposed FT-IR technology can be ideal for routine in-plant assessment of the tomato paste quality that would provide the tomato processors with accurate results in shorter time and lower cost.
本研究旨在通过使用可现场部署的便携式红外光谱仪，在无需任何样品制备的情况下，同时预测番茄酱的多个品质特性。在美国加利福尼亚州，共有1843个番茄酱样品由四家不同的领先番茄加工商在2015年、2016年、2017年和2019年的番茄季期间提供。包括天然番茄可溶性固形物（NTSS）、pH值、博斯韦克稠度、可滴定酸度（TA）、血清粘度、番茄红素、葡萄糖、果糖、抗坏血酸和柠檬酸在内的品质特性参考水平通过官方方法测定。使用配备三反射金刚石衰减全反射（ATR）采样系统的便携式傅里叶变换红外光谱仪直接采集红外光谱。校准模型和外部验证模型通过偏最小二乘回归（PLSR）建立。模型评估在随机选择的外部验证集上进行。从PLSR模型观察到参考值与傅里叶变换红外光谱预测值之间具有高度相关性（R = 0.85 - 0.99）。预测标准误差较低（SEP = 0.04 - 35.11），并实现了良好的预测性能（RPD = 1.8 - 7.3）。所提出的傅里叶变换红外光谱技术对于番茄酱品质的工厂常规评估可能是理想的，这将为番茄加工商在更短的时间内以更低的成本提供准确的结果。
