Table 1 The nuclei segmentation results of the BayesNuSeg and the baseline models. The BayesNuSeg model with uncertainty estimation outperforms all the baseline systems. N.A.: Not available

FCN8

0.842 ± 0.008

0.732 ± 0.049

N.A.

U-Net

0.824 ± 0.009

0.791 ± 0.048

N.A.

SegNet

0.845 ± 0.018

0.803 ± 0.055

N.A.

Hover-net

0.851 ± 0.010

0.829 ± 0.032

N.A.

BayesNuSeg

0.848 ± 0.013

0.835 ± 0.003

N.A.

FCN8 + MC dropout

0.848 ± 0.009

0.764 ± 0.004

0.699 ± 0.050

U-Net + MC dropout

0.840 ± 0.009

0.804 ± 0.037

0.738 ± 0.034

SegNet + MC dropout

0.847 ± 0.006

0.828 ± 0.045

0.763 ± 0.046

Hover-net + MC dropout

0.871 ± 0.010

0.840 ± 0.031

0.789 ± 0.032

BayesNuSeg + MC dropout

0.893 \(\varvec{\pm }\) 0.008

0.868 \(\varvec{\pm }\) 0.003

0.796 \(\varvec{\pm }\) 0.004