AI tool supports breeding for bone health in chickens

An Artificial Intelligence (AI) tool could help improve the welfare of chickens by predicting bone strength quickly and without the need for invasive testing, research has shown.

Researchers at the Roslin Institute have developed the automated tool that analyses X-ray images to estimate strength in leg bones, as an overall indication of skeletal health.

It offers an improvement on current methods, which are slow and labour-intensive, typically relying on manually analysing X-ray images or physically testing bones post-mortem from flock birds.

The tool is based on widely used AI systems adapted for this task, and can be implemented with standard computing infrastructure, making it accessible for use in breeding programmes.

Automated analysis

Researchers sought to assess the issue of bone damage, particularly in laying hens. Fractures can occur as a result of the physical demands of egg production and movement in housing systems, highlighting skeletal health as a key welfare priority.

The AI model works by identifying the chicken’s tibia bone from an X-ray image. It then analyses patterns in the image to predict how strong the bone is likely to be.

Developed using more than 900 X-ray images, the tool showed a strong correlation with bone strength measured using standard manual tests. It also outperformed manual scoring of X-rays, while requiring far less time and labour.

Because the method uses image data alone, predictions can be generated quickly once the system is trained, potentially enabling large-scale assessment of bone strength.

While the current study used post-mortem samples, the team aims to apply the method to images from live birds in future, which would further increase its practical value.

Improving welfare

Researchers found that the measurements taken by the AI tool closely matched the underlying genetics of bone strength, meaning the tool could be used to identify birds or family lines with stronger bones and guide breeding decisions.

The poultry breeding industry relies on assessing large numbers of birds to guide selection decisions, but practical limitations mean some important traits are difficult to measure routinely.

By providing a rapid, non-invasive and scalable way to assess bone strength, the tool could help address this gap and support efforts to reduce fractures in commercial flocks.

This research was published in Poultry Science, in collaboration with commercial partners Lohmann Breeders, Germany. The work was supported by the Roslin Foundation and the Foundation for Food and Agricultural Research Grant.

Related links

Research publication