Title: Comparison of Three Near Infrared Spectrophotometers for Infestation Detection in Wild Blueberries Using Multivariate Calibration Models Authors
|Peshlov, B - UNIV OF MAINE|
|Drummond, F - UNIV OF MAINE|
|Donahue, D - UNIV OF MAINE|
Submitted to: Near Infrared Spectroscopy Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: July 18, 2009
Publication Date: August 5, 2009
Repository URL: http://www.ars.usda.gov/SP2UserFiles/Place/54300520/411Comparisonofthreenirsforinfestationdetectioninwild.pdf
Citation: Peshlov, B.N., Dowell, F.E., Drummond, F.A., Donahue, D.W. 2009. Comparison of Three Near Infrared Spectrophotometers for Infestation Detection in Wild Blueberries Using Multivariate Calibration Models. Journal of Near Infrared Spectroscopy. 17:203-212. Interpretive Summary: There is an increased consumer interest in fresh and processed blueberry products because of their potential health benefits, in particular as a good source of antioxidants. The blueberry maggot fly is the most important pest of commercially grown lowbush and highbush blueberries in eastern North America and can infest a significant amount of the blueberry crop. Domestic and international markets for fresh, canned, and frozen fruit have a “near zero to zero” tolerance for infested fruit. There is a potential to change the way in which this fly is managed on-farm by developing technologies that rely upon detection of the fly within the fruit at the processing plant using optical detection systems such as near-infrared spectroscopy (NIRS). We studied the application of three NIRS instruments for detecting internal larvae and showed infested blueberries could be detected with accuracies up to 80%. This technology could be used to detect infested blueberries online and lead to an automated means of detecting and removing defective berries. This technology will help the blueberry industry meet the needs of domestic and foreign markets and thus lead to expanded markets for their product.
Technical Abstract: A near-infrared spectroscopy (NIRS) method for automated non-destructive detection of insect infestation internal to small fruit is desirable because of the zero-to-zero tolerance of the fresh and processed fruit markets. Three NIRS instruments: the Ocean Optics SD2000, the Perten DA7000 and the Oriel MS-257 were compared based on their capacity for larva detection in wild blueberries using partial least squares (PLS) regression models. The compared parameters included instrument wavelength range, scanning configuration, signal-to-noise ratio, and infestation prediction ratio of the PLS models. Factors that were considered were scatter and baseline effects in the NIR spectra and functional groups related to infestation identified by the PLS factor weights. The Perten spectrophotometer achieved highest signal-to-noise ratio and infestation prediction accuracy of 82%. The Ocean Optics had lower signal-to-noise ratio and achieved infestation detection accuracy 6% lower than Perten. Significant level of noise in the Oriel spectra led to less accurate detection. By analyzing the factor weights in the models it was concluded that the wavelength range of the Perten DA7000 includes more overtones of N-H, O-H, CH2 and CH3 which are possibly related to larvae infestation. Based on these results, it was concluded that the research grade Perten DA7000 provides best infestation detection results. However, the Ocean Optics SD2000 can be preferred in applications where high portability and low cost are required as it achieved comparable infestation detection accuracy.