Skip to main content
. Author manuscript; available in PMC: 2020 Jul 1.
Published in final edited form as: Comput Med Imaging Graph. 2019 May 9;75:47–55. doi: 10.1016/j.compmedimag.2019.04.007

Table 2:

The quantitative results of FCN-2S-VGG16, shape-prior model, UNet and the proposed network in this paper on Dataset-1. The pairwise two-tailed t-test demonstrated the significantly better performance of proposed method in comparison to the mentioned three methods.

Model Jaccard Dice Sensitivity Specificity
FCN-2S-VGG16 (Lee et al., 2017) 86.10 ± 6.10* 92.40 ± 3.74* 88.46 ± 5.23* 99.70 ± 0.28*
Shape-prior model (Popuri et al., 2016) 90.00 ± 7.9* 94.53 ± 5.06* 95.44 ± 6.60* 99.36 ± 4.12*
UNet (Ronneberger et al., 2015) 94.58 ± 3.80* 97.17 ± 2.09* 96.53 ± 2.36* 99.78 ± 0.23*
Proposed network 96.34 ± 2.77 98.11 ± 1.47 98.15 ± 1.63 99.81 ± 0.19
*

significant at p < 0.0001