Skip to main content

Table 7 The Dice metric and contour distance metrics between automated segmentation and manual segmentation for long-axis images, as well between segmentations by different human observers

From: Automated cardiovascular magnetic resonance image analysis with fully convolutional networks

(a) Dice metric

    
 

Auto vs Manual

O1 vs O2

O2 vs O3

O3 vs O1

 

(n = 600)

(n = 50)

(n = 50)

(n = 50)

LA cavity (2Ch)

0.93 (0.05)

0.92 (0.02)

0.90 (0.04)

0.90 (0.04)

LA cavity (4Ch)

0.95 (0.02)

0.95 (0.03)

0.94 (0.02)

0.94 (0.03)

RA cavity (4Ch)

0.96 (0.02)

0.95 (0.02)

0.95 (0.02)

0.95 (0.02)

(b) Mean contour distance (mm)

    

LA cavity (2Ch)

1.46 (1.06)

1.57 (0.39)

1.94 (0.68)

1.95 (0.57)

LA cavity (4Ch)

1.04 (0.38)

1.08 (0.40)

1.21 (0.33)

1.23 (0.35)

RA cavity (4Ch)

0.99 (0.43)

1.13 (0.35)

1.22 (0.37)

1.16 (0.37)

(c) Hausdorff distance (mm)

    

LA cavity (2Ch)

5.76 (5.85)

5.66 (1.97)

7.16 (3.12)

6.78 (2.53)

LA cavity (4Ch)

4.03 (2.26)

3.89 (1.85)

4.29 (1.97)

4.06 (1.44)

RA cavity (4Ch)

3.89 (2.39)

4.31 (2.20)

4.20 (2.16)

4.08 (2.06)

  1. The first column shows the difference between automated and manual segmentations on a test set of 600 subjects. The second to fourth columns show the inter-observer variability, which is evaluated on a randomly selected set of 50 subjects, each being analysed by three different human observers (O1, O2, O3) independently. The mean and standard deviation (in parenthesis) of the metrics are reported