Automated Image Analysis for Detection of Coccidia in Poultry

Area of Science:

• Veterinary Parasitology
• Computational Biology
• Poultry Science

Background:

• Coccidiosis, caused by Eimeria protozoa, is a significant economic burden on the poultry industry.
• Current diagnostic methods for Eimeria species and infection severity rely on manual, labor-intensive techniques prone to error.
• Existing diagnostic tools cannot determine oocyst sporulation, a key indicator of infectivity.

Purpose of the Study:

• To develop an automated Artificial Intelligence (AI) model for rapid Eimeria species identification and oocyst enumeration.
• To enhance diagnostic accuracy and efficiency in poultry coccidiosis detection.
• To enable the determination of oocyst sporulation status for assessing infectivity.

Main Methods:

• Computer vision models based on the Mask R-CNN neural network architecture were trained and evaluated.
• The AI model was designed to detect and differentiate three Eimeria species and their sporulation status (six detection groups).
• Model performance was assessed by comparing AI counts against manual enumeration, calculating the mean relative percentage difference (RPD).

Main Results:

• The AI model achieved a mean RPD of 5.64% across all groups, indicating high agreement with manual counts.
• Individual group RPDs ranged from -33.37% to 52.72%, demonstrating variable but generally acceptable accuracy.
• The models successfully differentiated oocyst sporulation status, a capability lacking in current methods.
• The AI approach proved speedy and required minimal sample processing for field-quality samples.

Conclusions:

• AI-powered computer vision offers a rapid and accurate method for diagnosing Eimeria infections in poultry.
• The developed models can identify Eimeria species, enumerate oocysts, and crucially, determine sporulation status.
• This technology has significant potential for field applications, improving coccidiosis control and reducing economic losses in the poultry industry.