Related Experiment Video
Updated: Jul 3, 2026
08:20
Superior Auto-Identification of Trypanosome Parasites by Using a Hybrid Deep-Learning Model
Published on: October 27, 2023
Artificial Intelligence with Convolutional Neural Networks for Microfilariae Detection in the Brazilian Amazon
João Carlos Silva de Oliveira1, Patrícia Moura Sousa1, Uziel Ferreira Suwa1
1Instituto Leônidas e Maria Deane, Fiocruz Amazônia. R. Teresina 476, Adrianópolis. 69057-070 Manaus AM Brasil. jcjunior182@gmail.com.
Ciencia & Saude Coletiva
|July 1, 2026
Summary
Artificial intelligence (AI) significantly improves filarial disease diagnosis in the Amazon. An AI model accurately detected microfilariae in dog blood samples, reducing analysis time by over 95%.
Area of Science:
- Veterinary Parasitology
- Medical Imaging
- Artificial Intelligence
Background:
- Filarial disease diagnosis in the Amazon is hindered by structural limitations.
- Microfilariae detection is crucial for diagnosing filarial diseases.
Purpose of the Study:
- To develop and evaluate an AI model for classifying microscopic images for microfilariae presence.
- To assess the efficiency of AI in parasitological screening.
Main Methods:
- Collected blood samples from 43 dogs in rural Manaus.
- Digitized stained blood slides to create 500 images.
- Trained an EfficientNetV2-B0 model using data augmentation and a patch-based approach.
- Validated AI results with expert morphological assessment and PCR confirmation.
Main Results:
- The AI model achieved 93.6% accuracy, 91.8% precision, 92.4% sensitivity, and 92.1% F1-score.
- AI analysis time per slide was 104 seconds, compared to 2,065 seconds for human analysis.
- Demonstrated a significant gain in diagnostic efficiency.
Conclusions:
- AI, specifically convolutional neural networks, shows high potential for accurate microfilariae detection.
- This AI approach offers a substantial improvement in efficiency for parasitological screening and epidemiological surveillance in resource-limited settings like the Amazon.
