Figure 4.

Detailed test results of the proposed-augmented model trained on aggregated vessel constellation with subsequent segment washing, values reflect mean voxel-wise macro F1 scores from cross-validation, error bars correspond to standard deviation across folds.