Figure 4.

Stacked bar chart of the qualitative four-grade scoring (not clinically acceptable, clinically acceptable with major changes, clinically acceptable with minor changes, clinically acceptable) of the automatic multi-atlas (MA) and deep-learning (DL) based image segmentations. The numbers in the bars indicate the occurrence in each category by the medical experts. The sum of the occurrence is equal to 84 for all organs (14 patients, 3 scorers and left/right), except for the brainstem (N = 42).