Mature-ripe tomato spectral classification according to lycopene content and fruit type by visible, NIR reflectance and intrinsic fluorescence

The human health benefits of carotenoids and phenolic compounds present in fruits and vegetables such as tomato, the second most consumed vegetable in the world, are well established. But characterisation of fruits is a tedious task. Tomato, with a thin, transparent and edible exocarp is interesting as a case-study for rapid characterisation by spectroscopy. An experiment was conducted to rapidly measure tomato lycopene content and classify mature-ripe tomatoes according to fruit-type (beef, vine or pink type) in a non-destructive way. A total of 1,560 fruits were collected from the market, retailers and producers. Three solid-state spectrometers were used; two for measurement of reflectance (visible; Vis: 450-750 nm and near-infrared; NIR: 600 to 1,150 nm) and one for capture of intrinsic fluorescence at either 275 or 400 nm excitation. Best results for lycopene content were obtained by Vis+NIR on beef-type fruits; coefficients of determination (r²) = 0.65, ratio of data standard deviation to root mean square error of prediction (SDR) = 1.80. A model based on the three fruit types yielded a r² of 0.64 and a SDR of 1.59. These results allowed for lycopene classification within three categories (<30, 30-50 and >50 mg/kg) with a correct classification rate of 73.6%, and no confusion between high and low lycopene content fruits. Fluorescence at EX 400 nm was also efficient at measuring lycopene (r²=0.59, SDR=1.51) but there was no advantage adding fluorescence to Vis or NIR data. The three types of fruits were well separated by Vis spectroscopy (80.3% correct classification). Adding EX 400 nm data to the model improved the classification rate to 87.9%. Intrinsic fluorescence at EX 275 nm clearly separated pink from the beef and vine-type fruits. Solid-state spectrometers thus have a potential for semi-quantitative classification of mature-ripe tomato fruits.