Table 3 Comparison between the segmentation accuracy of the FML method versus U-Net and BCDU-Net models.

From: Kidney segmentation from DCE-MRI converging level set methods, fuzzy clustering and Markov random field modeling

Method

All images

Low-contrast images

DSC

HD95

DSC

HD95

U-Net14

0.940 ± 0.041

10.30 ± 23.8

0.884 ± 0.071

19.9 ± 28.8

BCDU-Net34

0.942 ± 0.038

4.62 ± 12.35

0.90 ± 0.057

7.89 ± 12.27

FML (ours)

0.961 ± 0.017

0.68 ± 1.19

0.935 ± 0.037

2.23 ± 3.6

  1. Significant values are in [bold].
Back to article page