Détection automatisée de Sarcocystis spp. chez les bovins par l’intelligence artificielle à partir de coupes histologiques

Abstract

Bovine sarcocystosis is an indirect lifecycle parasitosis caused by protozoa derived from the genus Sarcocystis, which belong to the Apicomplexa phylum, and carries a major economic and sanitary impact. Faced with the limitations of conventional microscopic diagnosis, this study proposes a modern approach combining histopathology and artificial intelligence. Investigations conducted at the Sétif slaughterhouse revealed a complete absence of macroscopic cysts, but confirmed the presence of the parasite through histological analysis, with an overall prevalence of 74%. The infection proved to be more frequent in animals older than 5 years and in females. Among the target organs, the cardiac muscle emerged as the primary site of parasite localization. The expert-validated histological sections enabled the creation of an indexed digital image bank. These data were used to develop an autonomous, embedded prototype capable of capturing, processing, and analysing muscle sections in real time. Thanks to the integration of an AI model optimized for local and automatic cyst detection, this device stands out as a reliable decision-support tool for pathologists.