Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 1/31/2005
Publication Date: 5/23/2005
Citation: Smith, D.P., Mauldin, J.M., Lawrence, K.C., Park, B., Heitschmidt, G.W. 2005. Detection of fertility and development of hatching eggs with hyperspectral imaging [abstract]. European Symposium on the Quality of Eggs Meeting Abstract.
Technical Abstract: Manual candling to detect defects of hatching eggs (non-development due to infertility or early embryo death) is a laborious process. Therefore only a small percentage of eggs are candled to determine fertility rates of the producing farm. Screening all eggs prior to or during the first days of incubation to determine fertility or early development could benefit the hatchery as only fertile, developing eggs would be set or allowed to remain in the incubators. Incubation space could be minimized, or eggs set to produce better airflow, or potential 'exploders' culled. A hyperspectral imaging system (CCD detector, spectrograph, lens assembly, and software) has been developed to detect defects on chicken carcasses and meat. It has been modified, along with a light source, to detect external and internal defects of eggs. Hatching eggs are now being studied to determine if the system can detect fertility and early embryo development. In two replicate trials, 48 unincubated SCWL (white shell) eggs were obtained from a commercial hatchery, incubated, and then 12 eggs were removed and imaged on days 0, 1, 2, and 3 (n=96). Hyperspectral images were collected on each egg (wavelengths from approximately 400 to 900 nm) using tungsten-halogen backlighting with a 30 millisecond exposure time. Eggs were then broken out for confirmation. A ratio of transmittance images at two different wavelengths was used to determine fertility/development. The system as configured was unable to accurately determine fertility or development on Day 0 or 1, but did detect development on Day 2 (60% accuracy) and Day 3 (91% accuracy). Further testing has been conducted on broiler-type brown shell eggs, and detection of development again exceeded 90% by Day 3 of incubation. The hyperspectral imaging system appears capable of detecting fertile and developing eggs at Day 3 of incubation. Additional modifications to lighting and further image processing algorithms for hyperspectral images are being tested to improve system accuracy during earlier stages (from Day 0 to Day) of embryo development.