Fig 1.
The processing flow of the image processing for parasite eggs.
The initial input is the original image of the eggs, captured at 40x magnification. Fourteen steps enhance contrast and filter out noise in order to obtain the final images that serve as inputs for the feature extraction process.
Fig 2.
Flowchart of proposed methodology to automatically recognize parasitic eggs from microscopic photographs.
Fig 3.
Flowchart of the feature extraction for Taenia sp.
Flowchart showing feature extraction for Taenia sp. eggs, resulting in 80 variables for statistical analysis.
Fig 4.
Flowchart of the feature extraction process for Fasciola hepatica, Diphyllobothrium latum, and Trichuris trichiura eggs.
This process results in 92 features for statistical analysis.
Fig 5.
Sensitivity and specificity of each regression model’s ability to recognize parasites in digital images of fecal smears given differing probability cutoff values.
Table 1.
Summary of the principle variables and the features they represent.
Table 2.
Summary of classification variables, regression model fit, and odds ratio of each regression’s ability to make an accurate classification.