Non-invasive in-ovo sexing of chicken eggs using Raman spectroscopy and hyperspectral imaging

In order to meet the demand for consumption eggs, billions of specially bred layer chickensare hatched every year. Serving no purpose in the industry, over 420 million one day oldmale chicks are culled every year in Europe. Current, accurate (>95%) commercial in-ovosexing techniques are unfit for sexing before day 9 of incubation (E9) and their invasivenature imposes a risk for bacterial infection. With upcoming new legislation aiming tooutlaw the culling of chicken embryos after E7, there is an urgent need for thedevelopment of non-invasive early in-ovo chicken sexing methods. Reports relying onspectroscopy techniques have been abundant in recent years, yet relatively few have beendemonstrated in practice. The most compelling of the explored research techniques relieson the analysis of extraembryonic blood vessels using Raman and fluorescencespectroscopies, enabling sexing with good accuracy at E4 without disrupting the eggmembranes. Other researchers turned to machine vision to determine blood vesselgeometry from conventional photographs to perform in-ovo sexing with good accuracy atE4. Hyperspectral imaging (HSI) has the potential for commercial deployment of in-ovosexing, as texture analysis of HSI in the near-infrared can facilitate sexing with excellentaccuracy, even before incubation. While these techniques have indisputable potential in inovosexing, most of these efforts failed to meet the high standards for sexing accuracy,while others were faced insurmountable challenges during upscaling or when generalizingtheir approach to other breeds. This PhD project envisions the determination of spectralsex-specific digital biomarkers in the eggshell using Raman imaging complementary tovisible-light HSI in enabling early and accurate non-invasive in-ovo sexing. Initialidentification of novel biomarkers in white and brown eggs will be performedusing Raman and elaborate chemometric analysis with sex determined using PCR asthe cross-validation method. Likewise, informative spectral, morphological and texturalfeatures will distinctively be identified in HSI data obtained for the same eggs. State of theart data fusion strategies will be employed to expedite joint analysis and training ofclassification algorithms on Raman imaging and HSI data for in-ovo sexing. Digitalbiomarker model evaluated by deep learning approaches as a tool will be explored as partof this PhD project in the screening for early detection of sex in the eggs.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.