Detecting hidden fungal infection in tomato plants using pocket-sized near-infrared spectroscopy and chemometrics

Tomatoes are one of the most widely cultivated crops, but infections by pathogens like Fusarium oxysporum f. sp. lycopersici can lead to severe yield losses. This pilot study explores the use of SCiO©, a pocket-sized near-infrared spectral sensor operating in the 740–1070nm band, for non-destructive, early detection of Fusarium infections in tomato plants. Twenty greenhouse-grown plants were scanned at four branch locations using SCiO©, producing eighty spectra, separated equally over symptomatic and non-symptomatic plants. A chemometric model based on Orthogonal Partial Least Squares Discriminant Analysis was developed using spectra from fourteen plants for training and six for validation. The model achieved ∼92% classification accuracy. Visual inspection confirmed infection via internal browning. Results demonstrate the potential of low-cost spectral tools like SCiO© for automated early disease detection and sorting in greenhouse settings.