Body Composition Estimation in Breeding Ewes Using Live Weight and Body Parameters Utilizing Image Analysis

Area of Science:

• Animal Science
• Agricultural Engineering
• Veterinary Medicine

Background:

• Farmers need objective, non-invasive methods to assess ewe health.
• Live weight and body condition score are traditional but limited indicators.
• Understanding body composition (fat, muscle, bone) is crucial for strategic management.

Purpose of the Study:

• To establish relationships between body composition and easily measurable body parameters in ewes.
• To develop and validate predictive models for ewe body composition using image analysis.
• To provide farmers with advanced tools for monitoring ewe condition.

Main Methods:

• Utilized computerized tomography (CT) to determine actual body composition in 88 Coopworth ewes.
• Employed an image processing application to automatically capture body parameters.
• Applied multivariate linear regression (MLR), artificial neural network (ANN), and regression tree (RT) for model development.
• Validated predictive models using a subset of the data.

Main Results:

• Artificial neural networks (ANN) demonstrated strong predictive power: r²=0.90 for fat, r²=0.72 for muscle, and r²=0.50 for bone.
• Significant correlations were found between CT-derived composition and parameters derived from live weight and image analysis.
• The developed models accurately estimated body composition at key life stages (weaning, pre-mating).

Conclusions:

• Image processing combined with machine learning offers a reliable, non-invasive method for assessing ewe body composition.
• These findings enable farmers to refine nutritional and management practices for improved flock health and productivity.
• This technology represents a significant advancement over traditional methods for monitoring animal condition.