Ivan Miguel Leopoldo, Jerson Paul Leones
Soil-transmitted helminthiasis (STH) is a significant public health issue in the Philippines, with prevalence rates up to 94.4% for certain parasites. Current diagnostics rely on manual microscopy, often leading to misdiagnosis due to similarities between helminth eggs. This thesis proposes a deep learning approach for accurate classification of parasitic ova. A stacked transfer learning model is developed to identify three species: Ascaris lumbricoides, Trichuris trichiura, and hookworm. Microscope images undergo specialized preprocessing before being analyzed by a hybrid architecture combining YOLOv8 for detection and ResNet50 for classification, both adapted for parasite recognition. The model is trained and evaluated using a publicly available dataset from IEEE DataPort and compared with conventional convolutional neural networks and one-versus-all logistic regression classifiers. Performance is measured using precision, recall, specificity, and F1-score for each class. This work aims to improve diagnostic accuracy for STH infections, offering an automated tool that can enhance detection capabilities, especially in resource-limited tropical settings. © 2025 IEEE.
University of San Carlos, Information Sciences and Mathematics, Department of Computer, Cebu City, Philippines
