Table 2.

Segmentation results on the (a) MPUH; (b) Radboud test sets.

(a)
Classes	UNet + FPN + ASPP — ResNeXt50	Ours	UNet++ — EfficientNetB4	DeepLabV3+
Benign	0.852	0.86	0.887	0.887
Grade 3	0.881	0.897	0.836	0.856
Grade 4	0.90	0.912	0.905	0.889
Grade 5	0.923	0.923	0.95	0.952

(b)
Classes	UNet + FPN + ASPP — ResNeXt50	Ours	UNet++ — EfficientNetB4	DeepLabV3+
Benign	0.721	0.745	0.673	0.759
Grade 3	0.849	0.830	0.712	0.726
Grade 4	0.802	0.795	0.584	0.787
Grade 5	0.920	0.925	0.676	0.767