Omega-3 fatty acids (FAs) are essential for fish health and growth, especially during early development, making them critical indicators of egg spawn quality. The current gold standard for measuring FAs is gas chromatography-mass spectrometry (GC-MS), which offers high sensitivity but requires complex sample preparation, skilled analysts, and slow turnaround times. In contrast, Raman spectroscopy allows for the acquisition of a chemical fingerprint directly from an egg sample within seconds. While Raman technology has been used in quality control for fish fillets, its application in fish eggs remains unexplored.
In this study, we applied Raman spectroscopy, coupled with machine learning algorithms that leverage existing GC-MS data, to develop a rapid, in-situ method for obtaining FA data in fish eggs. California Yellowtail (Seriola dorsalis) eggs were used in this study with samples comprised of floating, neutral, and sinking eggs across twenty spawns and three spawning seasons. These samples were homogenized and analyzed for FAs via GC-MS and their corresponding Raman spectra were collected. Our preliminary model successfully predicted DHA concentration with an R2 of 0.896 and a mean square error (MSE) of 4.31%, demonstrating the ability to reliably predict egg FAs (Figure 1).
Importantly, this model can be shared to build larger databases, further enhancing predictive accuracy and reducing the need for in-house GC-MS analysis in aquaculture. This novel application introduces a cost-effective, time-efficient tool for egg quality assessment and can provide producers with immediate insights into spawn viability and quality—an application that was previously unattainable. Additionally, our research can potentially be expanded to include different fish species and enable rapid detection of other nutrients of interest such as vitamins, amino acids, and lipid classes.