Detection of hazelnut rancidity inside chocolate tablets using near-infrared hyperspectral imaging
Year: 2026
Type: article
Abstract: Near-infrared hyperspectral imaging (NIR-HSI) was evaluated as a non-destructive tool to detect hazelnut rancidity directly inside commercial chocolate bars, without separating the nuts from the chocolate matrix. Chocolate bars containing fresh or deliberately oxidized hazelnuts were imaged and analysed using a spectral unmixing strategy to separate chocolate and hazelnut contributions, yielding a residual component dominated by the nut signal where rancidity-related differences became more evident. ASCA applied to the original preprocessed spectra showed that rancidity explained 13.09% of the total variance ( p = 0.001), confirming a significant and interpretable spectral effect despite the dominant chocolate background. A PLS-DA model built on the residual spectra successfully discriminated fresh from rancid hazelnuts and correctly classified all chocolate bars in an independent validation set, using fewer latent variables than a model based on the original data. These results demonstrate the feasibility of in situ detection of hazelnut rancidity within chocolate tablets and support NIR-HSI, combined with chemometric modelling, as a promising approach for non-destructive quality control in heterogeneous confectionery products. • NIR-HSI detects rancid hazelnuts within intact chocolate bars. • Spectral unmixing isolates hazelnut signal from chocolate matrix. • ASCA: rancidity explains 13.1% of spectral variance ( p = 0.001). • PLS-DA on residual spectra classifies bars with 100% accuracy. • Residual model needs fewer latent variables than raw-spectra PLS-DA.
Institutions: University of Copenhagen, Universitat Rovira i Virgili
Language: English
Subfield: Analytical Chemistry
Field: Chemistry
Domain: Physical Sciences
SDG: Zero hunger
Open Access status: hybrid
Estimated APC paid: $3,750
Awards: G1315835405, G6034141024, G6097747396
OpenAlex ID: W7124994775