Rapid estimation of lycopene concentration in watermelon and tomato samples by fiber optic visible spectroscopy

Scope and method of study. Rapid lycopene sensing techniques using fiber optics and visible spectroscopy of watermelon flesh, watermelon puree and tomato puree were explored. Visible spectra of samples were correlated with lycopene concentrations determined by hexane extraction and spectrophotometry. Fiber optic probes consisting of a light emitting fiber and a detecting fiber connected to a miniature spectrometer were used for acquisition of visible spectra. An Absorbance Index (AI) and a Normalized Absorbance Index (NAI), obtained from the absorbance values at 700 nm and 565 nm were correlated with lycopene concentration. Partial Least Squares (PLS) regression models were developed from the absorbance values in the spectral range 500–750 nm. For watermelon flesh, spectra in transmittance and reflectance mode were correlated separately with lycopene concentration. For pureed samples of watermelon and tomato, only reflectance spectra were used. To study the effect of distance between light emitting and receiving fiber optics on reflectance spectra from pureed samples, multiple linear regression models correlating AI and NAI as a function of distance between optical fibers and lycopene concentration were developed.;Findings and conclusions. For transmittance spectra of watermelon flesh, AI and NAI were linearly correlated with lycopene concentration (R2 = 0.67 and 0.66). The PLS model from reflectance spectra of watermelon flesh gave better correlation with an R2 = 0.97 and a standard error of prediction (SEP) of 5.2 micro g/g. The AI and NAI from reflectance spectra of pureed samples were linearly correlated with lycopene concentration (R2 = 0.90 for both AI and NAI from watermelon, and R2 = 0.62 for AI and 0.61 for NAI from tomato). The PLS model for watermelon puree predicted lycopene concentration with an R2 of 0.97 and a SEP of 3.4 micro g/g. The PLS model for tomato puree predicted lycopene concentration with an R2 of 0.88 and a SEP of 2.5 micro g/g. Thus the PLS models predicted lycopene concentration from pureed samples with the same accuracy as the hexane extraction method. The distance between the light emitting and the light receiving fiber affected the spectral response (AI and NAI) of the fiber optic probe significantly.