Related Experiment Video
Updated: May 10, 2026
09:36
Outer-Boundary Assisted Segmentation and Quantification of Trabecular Bones by an Imagej Plugin
Published on: March 14, 2018
9.3K
Research Note: A deep learning method segments chicken keel bones from whole-body X-ray images
Moh Sallam1, Samuel Coulbourn Flores2, Dirk Jan de Koning1
1Department of Animal Biosciences, Swedish University of Agricultural Sciences, Box 7023, 750 07, Uppsala, Sweden.
Poultry Science
|August 27, 2024
Summary
Automating keel bone health monitoring in laying hens is now possible. A deep learning model accurately segments the keel bone from X-ray images, enabling future fracture risk assessment.
Area of Science:
- Veterinary Radiology
- Animal Science
- Biomedical Imaging
Background:
- Commercial laying hens frequently experience sternum (keel) bone damage, including deviations and fractures.
- Accurate assessment of keel bone condition is crucial for animal welfare and productivity.
Purpose of the Study:
- To develop and train a deep learning model for automatic segmentation of the keel bone from whole-body X-ray images of laying hens.
- To establish a foundation for automated measurements of keel bone geometry and density.
Main Methods:
- Acquisition of 1,051 full-body X-ray images from laying hens.
- Manual outlining of the keel bone on each image for annotation.
- Training a U-net deep learning model for keel bone segmentation.
- Evaluation using 5-fold cross-validation.
Main Results:
- The trained U-net model achieved high and repeatable segmentation accuracy, with Dice coefficients ranging from 0.88 to 0.90.
- The model demonstrated robust performance across multiple validation folds.
Conclusions:
- Automatic segmentation of the laying hen keel bone from X-ray images is feasible with high accuracy.
- Accurate segmentation is a prerequisite for automated measurements, facilitating the connection between bone characteristics and fracture susceptibility.
