Table 2 Evaluation metrics for four pancreas segmentation models.

From: Automated pancreas segmentation and volumetry using deep neural network on computed tomography

 

Precision

Recall

DSC

Trainable parameter

Basic U-net

0.861 \(\pm\) 0.468

0.816 \(\pm\) 0.173

0.822 \(\pm\) 0.143

11,003,073

Dense U-net

0.864 \(\pm\) 0.114

0.828 \(\pm\) 0.165

0.831 \(\pm\) 0.134

35,261,601

Residual U-net

0.843 \(\pm\) 0.127

0.810 \(\pm\) 0.178

0.808 \(\pm\) 0.146

2,350,857

Residual Dense U-net

0.869 \(\pm\) 0.110

0.842 \(\pm\) 0.156

0.842 \(\pm\) 0.128

47,074,657

  1. Results are indicated as mean ± standard deviation, and the best performances are indicated in bold. The results are highlighted in italics if the residual dense u-net performs significantly better than the corresponding method. We used a significance level of 0.05 and a paired t test for network comparison.
  2. DSC, dice similarity coefficient.
Back to article page