Abstract
With the expanding egg processing industry and increasing demand for egg yolk powder, efficient non-destructive methods for detecting yolk percentage have garnered significant attention. Existing non-destructive testing techniques frequently exhibit limited accuracy for brown eggs. To establish the optimal setup for non-destructive yolk measurement, we compared magnetic resonance imaging (MRI) field strengths and found that 3.0 T provided the best performance. Building on this, we established a standardized imaging workflow using 3D Slicer software, enabling non-destructive measurement of yolk volume and other relevant parameters. To build a robust predictive model, we then scanned 360 white eggs and 750 brown eggs, isolating the yolk via image segmentation algorithms to calculate parameters such as yolk volume, surface area, and Feret's diameter. Using a 70/30 dataset split, the best-performing model achieved high coefficients of determination (r²) of 0.893 and 0.907 in the training and test sets, respectively, demonstrating excellent predictive accuracy. The model's utility was further demonstrated by its ability to accurately predict yolk weight and percentage under varying conditions, including different shell colors and storage times. Analysis using the model revealed significantly lower yolk weight and percentage in Rhode Island Red (RIR) brown eggs compared to White Leghorn (WL) white eggs (P < 0.001), and long-term storage significantly increased these parameters (P < 0.001). Genetic analysis of RIR eggs also yielded heritability estimates of 0.39 for yolk weight and 0.42 for yolk percentage. Regarding safety, MRI exposure had no significant effect on hatchability, with a rate of 93.3 % in the treated group compared to 86.7 % in the control group (P > 0.05). This study provides an effective solution for rapid, non-destructive measurement of yolk percentage, which will significantly benefit layer production and ultimately support the development of the egg processing industry.