AI-Assisted Diarrheal Parasite Detection with Smartphone Microscopy

Diarrheal diseases are a major threat to children under five globally, especially in Nepal, caused by parasites like Giardia and Cryptosporidium from contaminated food and water. Traditional detection methods are expensive, require specialized expertise, and costly reagents, making them inaccessible in resource-constrained regions.

Our goal is to develop an AI-assisted parasite detection tool specifically designed for smartphone microscopes. Our approach involves building a comprehensive database of annotated microscopic images of Giardia and Cryptosporidium cysts from various sample types (water, vegetables, stools), with annotations from both experts and non-experts, enabling reliable testing of large numbers of samples quickly without requiring experienced users.

The project has created Nepal's largest annotated parasite dataset with over 437,000 images, including 23,000 expert-labeled samples. AI models trained on this dataset achieve accuracy comparable to conventional lab methods, providing a practical, low-cost solution for field diagnostics. The dataset and tools are publicly available, supporting further research and deployment in resource-limited settings.