Non-destructive prediction of eggshell strength using FT-NIR spectroscopy combined with PLS Regression

Eggshell strength is crucial for ensuring high-quality eggs, reducing breakage during handling, and meeting consumer expectations for freshness and integrity. Eggshell strength information is also important in efficient egg processing, laying hens disease prevention, and ensuring regulatory compliance. Conventional eggshell strength detection methods are typically destructive, labor-intensive, and time-consuming. The existing limitation of conventional techniques emphasizes the increasing need for faster, non-destructive assessment of eggshell strength. This study assessed the suitability of near-infrared (NIR) spectroscopy as a fast, non-destructive, and non-invasive eggshell strength detection technique. This study used a benchtop NIR spectroscopy system to acquire eggshell spectra (1200-2500 nm) and a texture profile analyzer for corresponding eggshell puncture strength data. Partial least squares regression (PLSR) was employed to develop calibration models, while various spectral pre-processing and band selection techniques were explored to enhance the model's performance. The developed models were evaluated internally by the leave-one-out cross-validation method and with an independent validation set. The PLSR models with ten selected wavelengths predicted eggshell strength with RMSE, RPD, and R2 of 0.82 N, 5.62, and 0.90, respectively, on the independent validation set. The study highlighted the efficacy of NIR spectroscopy combined with machine learning tools in detecting eggshell strength, demonstrating their potential as green tools to enhance quality control and resource optimization for sustainable development in the egg industry.