Close
  Indian J Med Microbiol
 

Figure 4: Visualization of parameter spaces for a segmentation pipeline using watershed transform (WS) - based cell splitting (a, b, c) or seeded watershed transform (SWS) - based cell splitting (d, e, f). For both combinations of methods, the parameters were automatically adjusted toward the ground truth, using genetic algorithms and a Jaccard similarity - based objective function. Based on this optimal parameterization, exploration of each module's parameter space was performed by varying the corresponding parameters. In the resulting plots, the gray values encoded the combined Jaccard similarity values for different combinations of parameters. Note that the range of the gray values is different for each subfigure

Figure 4: Visualization of parameter spaces for a segmentation pipeline using watershed transform (WS) - based cell splitting (a, b, c) or seeded watershed transform (SWS) - based cell splitting (d, e, f). For both combinations of methods, the parameters were automatically adjusted toward the ground truth, using genetic algorithms and a Jaccard similarity - based objective function. Based on this optimal parameterization, exploration of each module's parameter space was performed by varying the corresponding parameters. In the resulting plots, the gray values encoded the combined Jaccard similarity values for different combinations of parameters. Note that the range of the gray values is different for each subfigure