Table 1 Mean ± standard deviation values of Dice, precision and recall (sensitivity) for each individual model and our ensemble, as evaluated on the test set. Bolds indicate maximum values and asterisks indicate statistically significant differences with respect to the ensemble (p < 0.05) based on one-tail paired Wilcoxon sign-rank tests.

Dice

 U-Net

0.867 ± 0.075^*

0.908 ± 0.028^*

0.914 ± 0.026

0.908 ± 0.025^*

 All-Dropout

0.789 ± 0.175^*

0.885 ± 0.049^*

0.898 ± 0.032^*

0.894 ± 0.036^*

 BRU-Net

0.865 ± 0.076

0.905 ± 0.027^*

0.910 ± 0.029

0.908 ± 0.022^*

 U2-Net

0.874 ± 0.074

0.912 ± 0.026

0.917 ± 0.024

0.909 ± 0.024^*

 Ensemble

0.875 ± 0.075

0.912 ± 0.026

0.917 ± 0.024

0.911 ± 0.024

Precision

 U-Net

0.9160 ± 0.052

0.918 ± 0.043^*

0.919 ± 0.044^*

0.920 ± 0.039^*

 All-Dropout

0.930 ± 0.053

0.936 ± 0.046

0.937 ± 0.048

0.939 ± 0.029

 BRU-Net

0.889 ± 0.050^*

0.903 ± 0.050^*

0.904 ± 0.053^*

0.913 ± 0.036^*

 U2-Net

0.920 ± 0.047

0.923 ± 0.041

0.924 ± 0.043

0.925 ± 0.038

 Ensemble

0.925 ± 0.041

0.923 ± 0.043

0.923 ± 0.046

0.926 ± 0.037

Recall (Sensitivity)

 U-Net

0.832 ± 0.116

0.900 ± 0.042

0.910 ± 0.034

0.898 ± 0.045

 All-Dropout

0.708 ± 0.224^*

0.842 ± 0.068^*

0.864 ± 0.045^*

0.856 ± 0.068^*

 BRU-Net

0.854 ± 0.127

0.910 ± 0.035

0.918 ± 0.030

0.904 ± 0.042

 U2-Net

0.840 ± 0.117

0.902 ± 0.039

0.912 ± 0.033

0.895 ± 0.043

 Ensemble

0.840 ± 0.122

0.904 ± 0.038

0.913 ± 0.030

0.899 ± 0.044