January 1, 2006Open Access
Malaria Parasite Detection in Peripheral Blood Images
Authors
Discussion
Loading...
Member takes
Implication
Diagnostic algorithm study reveals 98% specificity for malaria parasite detection in Giemsa-stained blood samples, highlighting viable computerized screening.
Key Points
- To develop and evaluate an automated computer-vision framework for detecting Plasmodium malaria parasites in digitized light microscopy blood films.
- Standardized Giemsa-stained peripheral blood smear images to match reference color characteristics.
- Segmented stained pixels using a Bayesian classifier with non-parametric histogram density estimation, followed by feature extraction using Hu moments, histograms, shape metrics, and color auto-correlograms.
- Classified parasite versus non-parasite regions using a trained distance-weighted k-nearest neighbour algorithm.
- Parasite detection yielded a sensitivity of 74% and a specificity of 98%.
- The classification system achieved an 88% positive predictive value and a 95% negative predictive value.
Cite This Study
Tek et al. (2006) studied this question.