Applying near infrared spectroscopy to the detection of fruit fly eggs and larvae in intact fruit

The objective of this work was to investigate the potential use of near infrared (NIR) spectroscopy for non-destructive detection of fruit fly eggs and larvae in intact fruit. Mangoes, the major export fruit of Thailand, were used as model samples. The NIR spectra acquired under interactance mode in the short wavelength region of 700 nm to 1100 nm provided the best classification results, compared with spectra taken under the reflectance mode in the long wavelength region from 1100 nm to 2500 nm. The dominant factor in correct classification was the incubation period after infestation. The best classification was achieved using spectra of green mangoes obtained 48 h after infestation, with an error rate of 4.2% (two out of 48) for infested fruit and 0% for the 48 control fruit. Comparing regression coefficient plots of various partial least squares discriminant analysis (PLS-DA) models, it was determined that the most important classification wavelength was near 730 nm, coinciding with a unique peak previously observed in the spectra of dried fruit fly larvae. A universal calibration developed from two mango cultivars produced similar error rates. The results justify development of an automatic classification and sorting system based on NIR imaging technology.