Non-invasive prediction of mango quality using near-infrared spectroscopy: Assessment on spectral interferences of different packaging materials

• Published in: Journal of food engineering Vol. 357; p. 111653
• Main Authors: Ding, Fangchen, Zuo, Changzhou, García-Martín, Juan Francisco, Ge, Yan, Tu, Kang, Peng, Jing, Xiao, Hongmei, Lan, Weijie, Pan, Leiqing
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.11.2023
• Subjects: firmness; food contamination; food quality; Mango quality; mangoes; near-infrared spectroscopy; NIR spectroscopy; Non-destructive analysis; Packaging materials; poly(vinyl chloride); polyethylene; prediction; Spectral filtering; support vector machines; titratable acidity; Variable selection; Spectral filtering; Packaging materials; Mango quality; NIR spectroscopy; Non-destructive analysis; Variable selection
• DOI: 10.1016/j.jfoodeng.2023.111653
• ISSN: 0260-8774

Different packaging materials pose a challenging work for the non-invasive determination of contaminated food quality using near infrared (NIR) spectroscopy. This study investigated the effects of polyvinyl chloride (PVC), polyethylene (PE) and expandable polyethylene (EPE) packaging materials on the prediction of mango firmness (FI), dry matter (DMC), soluble solids (SSC), and titratable acidity (TA) using NIR. Obvious spectral interferences resulting from the three packaging materials were particularly located at 1150–1250 nm and 2320–2400 nm, and significantly reduced NIR prediction accuracy. Spectral filtering methods had the potential to reduce the NIR spectral interferences of packaging materials for contaminated mangoes’ quality assessment. Besides, least squares support vector machine (LS-SVM) models with the combination of spectral filtering and variable selection method can further improve the FI, SSC, DMC and TA prediction of packaged mango, with RPD values from 2.31 to 3.05. Consequently, it is crucial to consider suitable spectral filtering and variable selection methods to improve NIR prediction of quality of packaged food. •NIR spectral reflectance and shapes were affected by different packaging materials.•Spectral filtering can reduce the interferences of packaging materials on mango.•Variable selection can improve the NIR prediction of packaged mango quality.•Spectral filtering coupled with variable selection can improve the NIR prediction.