Near-infrared spectroscopy as a tool to assist Sargassum fusiforme quality grading: Harvest time discrimination and polyphenol prediction

Abstract

Seaweeds have attracted considerable attention as healthy and nutritious food. Harvest time is essential to ensure the quality and economic value of seaweeds. This study applied near-infrared (NIR) spectroscopy and chemometrics for quality grading of seaweed Sargassum fusiforme harvested at different times. S. fusiforme was identified according to harvest time by support vector machine (SVM), K-nearest neighbor (KNN), back propagation neural network (BPNN), and partial least squares-discriminant analysis (PLS-DA). NIR spectroscopy combined with PLS regression models was applied to quantify the vital quality indicator polyphenols. Herein, wavelength selection methods, including synergy interval (SI), genetic algorithm (GA), and competitive adaptive reweighted sampling (CARS), were used for model improvement. The SVM model performed better than the other models in the discrimination of harvest time, and the correct recognition rates in the calibration and prediction sets reached 100%. The CARS-PLS model achieved satisfactory results in predicting the contents of polyphenols, with a root mean square error (RMSEP) of 3.23 g kg−1 and a coefficient of determination (RP2) of 0.99 in an independent prediction set, respectively. Results suggest that S. fusiforme should be harvested at the middle maturity stage. The rapid and efficient NIR spectroscopy combined with chemometrics was proved to be an appropriate tool to assist the quality grading of S. fusiforme.