NON-INVASIVE QUANTIFICATION OF EGG QUALITY BY TIME-DOMAIN NMR RELAXOMETRY

Eggs are one of the most nutritionally complete natural foods, offering a balanced combination of essential groups of amino acids and lipids. Consumed all over the world and present in a variety of cultures, eggs gain marked value due to their low production and sales costs. Their commercial value is strongly influenced by internal quality parameters traditionally assessed through destructive methods such as albumin height (Haugh unit) and Yolk index influenced by egg viscosity affected by chicken food nutrition, age and quality life management. Therefore, the growing demand for non-destructive and rapid analytical techniques has driven research toward magnetic resonance in time domain as a viable alternative due to its low cost and operational speed. TD-NMR allows quantification based on relaxation times related to molecular mobility within egg compartments, enabling correlation with destructive measurements to calibrate egg quality prediction models, supporting its application in large-scale processing units without further food waste. The goal of this study was to establish a relationship between the signal intensity obtained from T₂ relaxation time and the mass of egg content (albumen and yolk) to develop predictive calibration models for use as initial step to quality egg control.

Fresh chicken eggs from two different production systems (free-range and semi-intensive) were used. Sample were categorized by weight according to the Brazilian Ministry of Agriculture regulation¹ covering mass from 45g to above 68g. Relaxometry experiments were performed using a Spinlock HR-MQC device operation at 0.3T (13MHz). CPMG sequence was employed to obtain transverse relaxation time (T₂). Acquisition parameters included TA 32ms, echo time 500ms with 10000 echoes. Eggs were maintained at controlled temperature of 20 and 40ºC simulating the storage and incubation temperature. Intact eggs were positioned inside a thermal control unit composed of Styrofoam cup with cotton insulation and introduced into magnet. Following acquisition, all eggs were destructively analyzed to separate and weigh albumen, yolk and shell components. Signal were processed with bi-exponential fitting and Inverse Laplace Transform (ILT)² to decompose signals attributed to the lipid-rich content, with shorter time between 10-100ms – and aqueous protein solution content, the longer relaxation time around 600-1000ms). Signal areas were used as the analytic parameter for correlation with component mass.

Signal decomposition confirmed the clear distinction between yolk and albumen components based on relaxation profiles. A strong correlation was found between peak areas of the albumen at 20ºC (R = 0.91). These correlation curves were robust enough to generate a predictive linear model for estimate the egg components mass based on TD-NMR signal. For yolk, the model performed slightly less accurately (R= 0.88), likely due to higher variability in lipid distribution and lower signal intensity. Error analysis revealed a mean relative error of 4.21% for albumen mass prediction and an RMSE of 1.50g. Yolk performed with a relative error of 6.5% with an RMSE of 1.52g. At 40ºC, the correlations were less reliable, especially for yolk, likely due to increase water mobility and signal variability induced by thermal effects. However, the relationship between total signal area and total egg mass remained significant, indicating potential applicability under controlled conditions.

Temperature control was shown to be critical for reliability inducing prediction quality methods to ally with storage systems. This method offers promising applications for egg producers, hatcheries and food quality laboratories aiming for rapid, non-invasive assessment method capable of associating with destructive method. It also paves the way for future studies combining time domain NMR with other analytical approaches such as Haugh unit and yolk index measurements to create comprehensive and automated quality control methodology. Future work may focus on validating their models in larger scale, more diverse samples set, including variables such as hen age, egg storage time decay, which should influence relaxation properties.