Related Experiment Video
Updated: May 25, 2026
08:20
Superior Auto-Identification of Trypanosome Parasites by Using a Hybrid Deep-Learning Model
Published on: October 27, 2023
Automatic segmentation and classification of human intestinal parasites from microscopy images
Celso T N Suzuki1, Jancarlo F Gomes, Alexandre X Falcão
1Institute of Computing, University of Campinas, São Paulo 13084-971, Brazil. jgomes@ic.unicamp.br
IEEE Transactions on Bio-Medical Engineering
|February 14, 2012
Summary
This study presents a new computational method for diagnosing human intestinal parasites from microscopy images, even with fecal impurities. The approach accurately identifies 15 common species, paving the way for automated enteroparasitosis diagnosis.
Area of Science:
- Medical Parasitology
- Computational Pathology
- Image Analysis
Background:
- Human intestinal parasites are a significant health issue in tropical regions, leading to severe health consequences.
- Current diagnosis relies on microscopy, which is prone to moderate to high error rates and challenging with fecal impurities.
- Existing computational methods are limited to a few species and require parasite images free of impurities.
Purpose of the Study:
- To develop a robust computational method for the automated segmentation and classification of human intestinal parasites.
- To address the challenge of fecal impurities in routine diagnostic microscopy images.
- To identify the 15 most common species of protozoan cysts, helminth eggs, and larvae found in Brazil.
Main Methods:
- Utilized bright field microscopy images containing fecal impurities.
- Employed ellipse matching and image foresting transform for image segmentation.
- Applied multiple object descriptors optimized by genetic programming for feature representation.
- Implemented an optimum-path forest classifier for accurate object recognition.
Main Results:
- Successfully segmented and classified 15 common species of human intestinal parasites from impure microscopy images.
- Demonstrated the method's effectiveness in handling challenging fecal impurities.
- Achieved promising results for the automated diagnosis of enteroparasitosis.
Conclusions:
- The developed computational method offers a promising solution for the automated diagnosis of enteroparasitosis.
- This approach can overcome the limitations of manual microscopy and current computational techniques.
- Further development could lead to fully automated, accurate, and efficient parasite diagnosis in clinical settings.
