|Mauldin, J. - UGA-DEPT. OF POULTRY SCIE|
|Heitschmidt, G. - UGA-BIO AND AGRI ENG|
Submitted to: European Symposium on Quality of Poultry Meat
Publication Type: Proceedings
Publication Acceptance Date: April 20, 2005
Publication Date: May 25, 2005
Citation: Smith, D.P., Mauldin, J.M., Lawrence, K.C., Park, B., Heitschmidt, G.W. 2005. Detection of fertility and early development of hatching eggs with hyperspectral imaging. Proceedings 17th European Symposium on Quality of Poultry Meat. p.139-144. Interpretive Summary: Many of the eggs put into commercial hatcheries will not hatch due to being infertile or early embryo death. Not putting these eggs into incubators, or removing them after 3 days will benefit the hatchery by reducing crowding (which lowers production), possibly reducing utility costs, and preventing contamination (eggs that don’t develop often spread bacteria and molds in the incubator). It is currently too labor intensive to manually check the eggs, as there are typically more than 200,000 per day entering the hatchery. Therefore an automated system capable of detecting these eggs would be a major benefit to the poultry industry. A hyperspectral imaging system was adapted to screen white shell (laying-type chickens) and brown shell (broiler chickens) before incubation and each during the first 3 days of incubation. Results show the imaging system was capable of accurately detecting positive embryo development on day 3 of incubation at a 91% rate for white shell eggs, and an 83% rate for brown shell eggs. Further testing and modifications to the procedure are expected to increase the accuracy rate of the system.
Technical Abstract: Detection of fertility prior to incubation or recognition of absence of development during the first days of incubation could allow hatcheries to cull infertile or problem eggs, reducing incubator space, utility costs, and contamination from exploder eggs. This study was conducted to determine the feasibility of using a hyperspectral imaging system to detect changes in eggs related to fertility and development for both white shell eggs and brown eggs. In Experiment 1, 48 single comb white leghorn (SCWL) white shell eggs were obtained unincubated from a commercial hatchery in each of two replicate trials (n=96). On days 0, 1, 2, and 3 of incubation, 12 eggs were imaged using a hyperspectral camera system (wavelengths detected from 400-900nm) and a tungsten-halogen backlight. Each day the eggs were broken out for visual confirmation after imaging. Exposure time was 30 ms, and a ratio of transmittance images at wavelengths 576 and 655 nm was used to detect fertility and embryo development. On Day 3, 23 of 24 eggs were determined fertile at breakout; the imaging system accurately identified the one infertile and 21 of 23 fertile eggs. One egg was determined fertile at Day 0, none at Day 1, and 12 at Day 2. One infertile egg was incorrectly identified as fertile on Day 2. In Experiment 2, in each of two replicate trails 12 brown shell broiler chicken eggs were obtained unincubated, and imaged on days 0, 1, 2, and 3 (same eggs kept and imaged throughout each trial; n=24). Embryo development was then visually confirmed on Day 6. All eggs used were fertile, so the earliest point of embryo development was determined. Exposure time was increased to 250 ms and a transmission ratio of wavelengths used at 576 nm, and a range of wavelengths (682 +/- 13 nm, depending on the light transmission per egg). Detection rates by Day were: Day 1, 13/24 (54%); Day 2 14/24 (58%); and, Day 3, 20/24 (83%). Overall, the hyperspectral imaging system appears capable of detecting developing eggs on Day 3 of incubation at a 91% rate for white shell eggs, and an 83% rate for brown shell eggs.