Krill oil contains eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), long chain omega-3 polyunsaturated fatty acids with essential roles in human health, and astaxanthin, a naturally occurring keto-carotenoid that protects EPA+DHA against oxidation. Here, we assess Raman and IR spectroscopy (as stand-alone techniques and paired using three different data-fusion approaches) as methods for simultaneous quantitation of EPA+DHA and astaxanthin in krill oil. Raman spectroscopy could accurately (RMSE_P = 40 μg g^–1, r²_p = 0.98) quantitate astaxanthin in krill oil despite its low concentrations (212–693 μg g^–1). This analysis could be performed directly through gelatin capsules with no loss of prediction accuracy (RMSE_P = 27 μg g^–1, r²_p = 0.99). Fusing IR and Raman data did not improve the astaxanthin quantitation models. EPA+DHA quantitation was more accurate using “mid-level” fusion (RMSE_P = 1.2%, r²_p = 0.99) than models from either Raman (RMSE_P = 4.5%, r²_p = 0.90) or IR (RMSE_P = 7.3%, r²_p = 0.73). Data fusion also significantly improved quantitation accuracy for quantification of other fatty acid classes.

中文翻译：

---

用于磷虾油成分快速，多组分定量的拉曼光谱和红外光谱数据融合策略

磷虾油包含二十碳五烯酸（EPA）和二十二碳六烯酸（DHA），对人体健康至关重要的长链omega-3多不饱和脂肪酸和虾青素（一种天然存在的酮类胡萝卜素，可保护EPA + DHA免受氧化）。在这里，我们评估拉曼光谱和红外光谱（作为独立技术并使用三种不同的数据融合方法进行配对），作为同时定量磷虾油中EPA + DHA和虾青素的方法。拉曼光谱法可以准确地（RMSE _P = 40μgg ^–1，r ² _p = 0.98）定量分析磷虾油中的虾青素，尽管其浓度较低（212–693μgg ^–1）。该分析可以直接通过明胶胶囊进行，而不会损失预测准确性（RMSE _P = 27μgg ^–1，r ² _p = 0.99）。融合IR和拉曼数据并不能改善虾青素的定量模型。使用“中级”融合（RMSE _P = 1.2％，r ² _p = 0.99）进行的EPA + DHA定量比来自拉曼（RMSE _P = 4.5％，r ² _p = 0.90）或IR（RMSE _P）的模型更准确。= 7.3％，r ² _p = 0.73）。数据融合还显着提高了定量精度，可用于其他脂肪酸类别的定量。