Table 2.

Dice score of each class calculated on the testset of the multi-class subset, $\mu$ being the mean over all classes

Every row represents one of the trials, fully supervised (FS), ensemble (EN), and implicit labelling (IL), trained on the binary and multi-class sets. The highest value per column and trainset is highlighted in bold. The classes Background, Abdominal Wall, Colon, Liver, Pancreas, Small Intestine, and Stomach are abbreviated by their first three letters. Cells are highlighted in green if the difference between IL and EN is strongly significant (p<.01) or yellow if significant (p<.05)