Authentication of cocoa (Theobroma cacao) bean hybrids by NIR-hyperspectral imaging and chemometrics

• Published in: Food control Vol. 118; p. 107445
• Main Authors: Cruz-Tirado, J.P., Fernández Pierna, Juan Antonio, Rogez, Hervé, Barbin, Douglas Fernandes, Baeten, Vincent
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01-12-2020
• Subjects: Computer vision; Genotype; Imaging; Near-infrared imaging; Computer vision; Genotype; Imaging; Near-infrared imaging
• ISSN: 0956-7135; 1873-7129
• DOI: 10.1016/j.foodcont.2020.107445

The hybridization of cocoa generates new varieties with the aim of opening new flavors, textures, and disease resistance. The objective of this study was to develop and validate classification models based on NIR hyperspectral imaging and chemometrics for the discrimination of five valuable cocoa bean hybrids. The chemometrics tools, PLS-DA and SVM, showed comparable results for two-class (hybrids) models, but SVM (3.8–23.1% prediction error) was superior to PLS-DA (4.4–34.4% prediction error) when all five classes (hybrids) were included in a model. PLS-DA maps showed a simple and informative way to discriminate hybrids, allowing a correct classification in 50–100% of cases. Finally, it can be concluded that the models created in this work could be a good and reliably alternative to the actual visual method for the discrimination of cocoa bean hybrids.
[Display omitted]
•NIR-HSI coupled to chemometrics were used to identify cocoa bean hybrids.•SVM and PLS-DA models showed similar performance to identify two-classes hybrids.•SVM was superior to PLS-DA to identify 5-classes hybrids.•Spectral information from the bean shell was dependent on the mother's genetics.