Skip to main content
. Author manuscript; available in PMC: 2021 Jun 11.
Published before final editing as: IEEE Trans Image Process. 2019 Jun 19:10.1109/TIP.2019.2919937. doi: 10.1109/TIP.2019.2919937

TABLE VII.

Result Comparison With the State-of-the-Art Network Structures on the Nuclei Segmentation Dataset. The Boldface Results Indicate no Significant Difference From the Best Result (p-Value < 0.05 in Student’s t-Test). The Parameter Number (Param) of Each Network Is Also Reported

Networks F1-Score (%) Object Dice (%) H-Distance (mm) Param (M)
U-Net 87.9±13.4 86.8±11.1 6.93±8.73 24.16
DCAN 89.0±12.3 87.3±10.2 6.49±8.83 21.06
ResNet-34 88.5±12.6 87.2±10.5 6.50±7.59 28.03
DenseNet-121 89.9±11.3 88.2±9.7 5.74±7.02 74.90
ResNet-101 90.3±10.3 88.6±8.8 5.53±6.07 96.92
ResNet34+HR-L2 91.2±9.7 89.0±8.6 5.30±7.11 34.96
ResNet34+HR-L3 91.1±10.2 89.1±8.8 5.07±5.88 42.19
ResNet34+HR-L4 90.3±10.9 88.5±9.6 5.42±6.37 67.62