Rapid automated detection and classification of Babesia spp., Theileria spp. and Toxoplasma gondii using artificial intelligence (YOLOv8) model

Authors: Hajir Muqdad Alhmdany, and Basima Abdulfatah Albadrani

Ger. J. Vet. Res 2026. vol. 6, Iss. 3 pp:93-106
Doi: https://doi.org/10.51585/gjvr.2026.3.0204

Download Options

×

• Download PDF
• Download XML
• Download Figures
• Supplementary Files

Close

Abstract:

Blood-borne parasites such as Babesia spp., Theileria spp., and Toxoplasma gondii are among the most important pathogens affecting cattle, causing considerable economic losses and adversely affecting animal health. While microscopic examination remains a fundamental diagnostic method for blood-borne parasites, its accuracy depends on examiner expertise and variability in microscopic image interpretation, highlighting the need for reliable artificial intelligence (AI)-based diagnostic solutions. The present study developed and evaluated the performance of a YOLO (You Only Look Once)-based deep learning model (YOLOv8) for automatic multiclass detection and classification of these parasites using images of Giemsa-stained and wet blood smears. Polymerase chain reaction (PCR) assays served as the reference standard to confirm parasite infections before image acquisition. For each parasite species (Babesia spp., Theileria spp., and Toxoplasma gondii), we selected 30 PCR-confirmed blood samples with single infections, resulting in a dataset of 2,799 images for training and evaluating the YOLOv8 model. The YOLOv8 model achieved precision, recall, F1-score, and mean average precision (mAP) values of 83.4%, 81.5%, 82.4%, and 82.5%, respectively. The model demonstrated robustness to variations in parasite morphology, staining intensity, and microscopic background, and accurately distinguished parasites from cellular artifacts that could lead to diagnostic errors. The primary contribution of this study is the development and validation of a clinically relevant multiclass YOLOv8 framework, using PCR-confirmed blood samples from naturally infected cattle, for the simultaneous detection of Babesia spp., Theileria spp., and Toxoplasma gondii. Evaluation with both Giemsa-stained and unstained wet blood smear images confirmed the robustness of the YOLOv8 framework under routine microscopic conditions, supporting its potential as a rapid decision-support tool in veterinary diagnostic laboratories.

Keywords:

Microscopy, PCR, Babesia, Theileria. Toxoplasma, YOLOv8

Statistics:

Article Views: 3004
PDF Download: 31