Predictive precision: AI histology for preventing breast muscle myopathy and food loss

The global demand for high-yield and quality chicken meat emphasizes the need for innovative tools in poultry research. Moreover, the increasing incidence of breast muscle abnormalities, such as White Striping (WS), Wooden Breast (WB), and Spaghetti Meat (SM), amplifies the urgency for tools capable of rapid and precise diagnosis of breast muscle development and morphology in broiler chickens. In response to this critical need, we present a novel deep learning-based automated image analysis workflow, combining Fiji (ImageJ) with Cellpose and MorphoLibJ plugins to provide automated diameter and cross-sectional area quantification for broiler breast muscle. Previous research in our lab has demonstrated the efficiency and accuracy of this tool. Comparing the AI image analysis tool with manual analysis demonstrated an impressive accuracy rate of up to 99.91%, coupled with significantly enhanced speed and productivity. The automated workflow processed 70 times more data sets in 38-fold less time. The implementation of this method is intended to detect breast muscle myopathies (BMM) in broiler flocks at two weeks of age. We introduce a new concept of relative myofiber size, providing a unified metric and predictive threshold, showcasing a clear separation between moderate and severe myopathy broilers. This approach will enable early adjustment of poultry management, implementing preventive actions such as modifications of feeding programs and/or lighting regimen, to reduce growth rates, a factor known to be associated with increased myopathy prevalence.