Related Experiment Video
Updated: Jun 13, 2025
08:23
Phenotypic Analysis of Rodent Malaria Parasite Asexual and Sexual Blood Stages and Mosquito Stages
Published on: May 30, 2019
11.5K
Application of Machine Learning in a Rodent Malaria Model for Rapid, Accurate, and Consistent Parasite Counts
Sean Yanik1,2, Hang Yu3, Nattawat Chaiyawong1,2
1Department of Molecular Microbiology and Immunology, Johns Hopkins School of Public Health, Baltimore, Maryland.
The American Journal of Tropical Medicine and Hygiene
|September 10, 2024
Summary
Automating malaria parasite counting in rodents using machine learning software speeds up research. Malaria Screener R standardizes infected red blood cell counts, improving antimalarial and vaccine testing.
Area of Science:
- Parasitology
- Machine Learning
- Biomedical Research
Background:
- Rodent malaria models are crucial for preclinical antimalarial and vaccine development.
- Manual counting of infected red blood cells (iRBCs) is time-consuming and prone to inconsistency.
- Existing methods lack standardization, hindering reliable evaluation of treatment outcomes.
Purpose of the Study:
- To develop and validate a machine learning-based software, Malaria Screener R, for automated counting of Plasmodium-infected red blood cells (iRBCs) in rodent models.
- To expedite and standardize parasitemia assessment in preclinical malaria research.
- To provide a reliable and accessible tool for evaluating novel antimalarials and vaccines.
Main Methods:
- Development of a machine learning model leveraging pretrained weights from a previous human malaria study.
- Retraining the model with new data specific to Plasmodium yoelii and Plasmodium berghei in mouse models.
- Utilizing Giemsa-stained blood smear images captured via standard microscopes.
- Implementing the model in a user-friendly desktop application for Windows and MacOS.
Main Results:
- The developed ML model achieved high accuracy in measuring P. yoelii and P. berghei parasitemia (R2 = 0.9916).
- The software demonstrated generalizability across different staining durations and microscope types.
- The model met WHO competency level 1 for parasite counting, indicating high reliability.
- Automated analysis significantly reduces time and improves consistency compared to manual counting.
Conclusions:
- Malaria Screener R offers a reliable, automated solution for quantifying parasitemia in rodent malaria models.
- The software facilitates rapid and consistent evaluation of antimalarials and vaccines.
- Standardized automated analysis enhances the efficiency and reproducibility of preclinical malaria research.
