Non-destructive quantification and physiological interpretation of leaf carotenoids using raman spectroscopy

Carotenoids in plants serve not only as key indicators of physiological status but also as critical compounds for human nutrition due to their antioxidant properties. As such, there is growing demand for non-destructive, real-time methods to quantify carotenoids in crops. Raman spectroscopy offers considerable promise in this context, as it is unaffected by water interference and requires minimal sample preparation. However, leafy vegetables remain particularly challenging for Raman-based carotenoid analysis due to strong fluorescence from chlorophyll and the coexistence of complex biomolecules.
This dissertation presents one of the earliest attempts to apply a classification-based approach using Raman spectroscopy combined with Linear Discriminant Analysis (LDA) to assess carotenoid levels in chlorophyll-rich vegetables. Arabidopsis thaliana mutants with genetically controlled carotenoid content were used to establish and validate the model, which was subsequently applied to spinach (Spinacia oleracea) cultivated under varying abiotic stress conditions. Various spectral preprocessing techniques and Raman shift subsets were tested to optimize classification accuracy. The LDA model successfully categorized lutein and β-carotene concentration ranges, achieving classification accuracies of up to 95.45% in Arabidopsis and 90.91% in spinach. These results demonstrate superior practical applicability compared to continuous regression models such as Partial Least Squares Regression (PLSR).
Overall, LDA-assisted Raman spectroscopy provides a selective, robust, and field-adaptable framework for non-destructive carotenoid quantification and quality assessment in agricultural and food systems. Additionally, the study investigated the use of peak intensity ratios between chlorophyll and carotenoids, as well as among different carotenoid types, in Raman spectra as potential indicators of photosynthetic and photoprotective activity under stress conditions in spinach. While these physiological assessments yielded limited results, the findings suggest strong potential for future development.