Detection of Mold-Damaged Chestnuts by Near-Infrared Spectroscopy

•NIR spectroscopy is a feasible method for the detection of mold damage in chestnut.•QDA allows reliable detection of infected chestnut.•Best results depend on the side of the chestnut analyzed.•Classification accuracy depends on the severity of infection.•The suggested approach provides the basis for a rapid online detection system.

Mold infection is a significant postharvest problem for processors of chestnuts (Castanea sativa, Miller). Fungal disease causes direct loss of product or reduced value due to the lower-quality grade of the chestnut lot. In most cases, fungal infection is not detectable using traditional sorting techniques. In this study, the feasibility of using Near-Infrared (NIR) spectroscopy to detect hidden mold infection in chestnut was demonstrated. Using a genetic algorithm for feature selection (from 2 to 6 wavelengths) in combination with image analysis grading and Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA) or k-Nearest-Neighbors (kNN) routines, classification error rates as low as 2.42% false negative, 2.34% false positive, and 2.38% total error were achieved, with an Area Under the ROC Curve (AUC) value of 0.997 and a Wilk's λ of 0.363 (P < 0.001). A Savitzky–Golay first derivative spectral pretreatment with 33 smoothing points was used. The optimal features corresponded to Abs[1118 nm], Abs[1200 nm], Abs[1626 nm], and Abs[1844 nm].The results represent an important step toward the development of a sorting system based on multi-spectral NIR bands, with the potential to rapidly detect and remove chestnuts contaminated by fungi and reduce the incidence of hidden mold in chestnut lots.