Rapid determination of free amino acids and caffeine in matcha using near-infrared spectroscopy: A comparison of portable and benchtop systems

Abstract

Matcha is gaining popularity around the globe, because of its worthy compositions and health benefits. This work proposed to predict free amino acids (FAA) and caffeine contents in matcha using a portable near-infrared (NIR) spectroscopy system combined with two different acquisition techniques (integrating sphere, IS; and fiber probe, FP) in comparison to a benchtop Fourier transform-NIR (FT-NIR) spectroscopy system. The prediction models were built using multi-variable selection algorithms such as synergy interval (Si), competitive adaptive reweighted sampling (CARS), and bootstrapping soft shrinkage (BOSS). Results in the current work showed that BOSS-PLS (BOSS-partial least squares) models yielded satisfactory results for both NIR spectroscopy types. In addition, the portable NIR-IS system has a prediction ability comparable to the benchtop FT-NIR, with coefficients of prediction (Rp) of 0.8920 and 0.8992 for FAA and caffeine, respectively. Consequently, the portable NIR-IS system has great compatibility for the rapid, accurate and on-site prediction of matcha compositions.