Figure 2 | Scientific Reports

Figure 2

From: A vision transformer architecture for the automated segmentation of retinal lesions in spectral domain optical coherence tomography images

Figure 2

Model comparison results per class and lesion size. The boxplots show the value distribution of the entire dataset for the Dice scores. Lesion sizes are discretized into quartiles from small (q1) to large (q4). The boxes show the middle 50% of the values. Whiskers reach at most 1.5 times the interquartile range (= box height) on top of either end of the box. Points outside the whiskers are considered outliers and are not shown here. The horizontal bar within the box shows the median. The mean is depicted with a yellow diamond.

Back to article page