Recognition of parasitic helminth eggs via a deep learning-based platform

Accurate and rapid diagnosis is crucial for the effective treatment of parasitosis. Traditional etiological methods, especially microscopic examination, are time-consuming, labor-intensive, and prone to false or missed detections. In response to these challenges, this study explores the use of artificial intelligence (AI) for the detection and classification of human parasite eggs through the YOLOv4 deep learning object detection algorithm.

Eggs from species such as Ascaris lumbricoides (A. lumbricoides), Trichuris trichiura (T. trichiura), Enterobius vermicularis (E. vermicularis), Ancylostoma duodenale (A. duodenale), Schistosoma japonicum (S. japonicum), Paragonimus westermani (P. westermani), Fasciolopsis buski (F. buski), Clonorchis sinensis (C. sinensis), and Taenia spp. (T. spp.) were collected and prepared as both single species and mixed egg smears. These samples were photographed under a light microscope and analyzed using the YOLO (You Only Look Once) v4 model.

The model demonstrated high recognition accuracy, achieving 100% for Clonorchis sinensis and Schistosoma japonicum, with slightly lower accuracies for other species such as E. vermicularis (89.31%), F. buski (88.00%), and T. trichiura (84.85%). For mixed helminth eggs, the recognition accuracy rates arrived at Group 1 (98.10, 95.61%), Group 2 (94.86, 93.28 and 91.43%), and Group 3 (93.34 and 75.00%), indicating the platform's robustness but also highlighting areas for improvement in complex diagnostic scenarios.

The results show that this AI-assisted platform significantly reduces reliance on professional expertise while maintaining real-time efficiency and high accuracy, offering a powerful tool for the diagnosis and treatment of parasitosis. With further optimization, such as expanding training datasets and refining recognition algorithms, this AI system could become a key resource in both clinical and public health efforts to combat parasitic infections.