Figure 6
From: Unsupervised CT Lung Image Segmentation of a Mycobacterium Tuberculosis Infection Model

Dice similarity coefficient (DSC), Haussdorff distance (HD) and Haussdorff distance averaged (HDA) plots along the slices sorted in ascending order based on the DSC of the semi-automatic segmentation with respect to the surrogate ground truth. Data have been filtered with the locally weighted scatterplot smoothing (LOESS) model. The 95% confidence interval is drawn as a shadow of the same color as the corresponding line.