Uso da espectroscopia de infravermelho próximo para avaliação da qualidade de tomate

Abstract

Traceability and innovation in tomato production are essential to ensure food safety and enhance the valorization of agricultural products, while quality is a key determinant for both the food industry and consumers. In this context, near-infrared spectroscopy (NIRS) stands out as a rapid, nondestructive technique applicable to both geographical origin classification and the prediction of quality parameters. In experiment 1, tomatoes (Solanum lycopersicum) from the Brazilian states of Goiás, São Paulo, and Minas Gerais were analyzed using the portable F-750® spectrometer. Spectral pre-processing (Baseline, standard normal variate - SNV, and SNV combined with detrending) was applied, and chemometric models were developed in MATLAB® (PCA-LDA, PLS-DA, SPA-LDA, and GA-LDA) to discriminate geographical origin. The SPA-LDA model built from absorbance spectra showed the best performance for classifying Caniati® tomatoes from São Paulo, achieving 97% sensitivity, 100% specificity, and 99% overall accuracy, highlighting the strong potential of NIRS for geographical traceability. In experiment 2, tomatoes cultivated in Minas Gerais were collected from different plant positions, with spectral acquisition at the equatorial region and soluble solids (SS) determined as °Brix; a comparative evaluation was also conducted between two portable NIR devices (F 750® and MyNIR®). Data were explored using PCA and modeled by partial least squares regression (PLSR), with calibration and validation statistics used to assess model performance. The best F-750® model yielded R²c = 0.6644 and SDRcv = 1.31 (second-derivative, 5+5 window), whereas the best MyNIR® model achieved R²c = 0.6115 and SDRcv = 1.32 (smoothing + detrending). Overall, these findings demonstrate that portable NIRS is a promising approach for the tomato supply chain, providing high performance for origin classification and comparable predictive capability between devices for soluble solids estimation. In experiment 3, the influence of fruit position on the plant on the postharvest quality of Caniati® tomatoes was evaluated by analyzing soluble solids, titratable acidity, dry matter, vitamin C, and lycopene. Fruits harvested from higher positions on the plant showed higher dry matter and vitamin C contents, indicating a partial effect of fruit position and contributing to harvest heterogeneity.