Skip to main content

Table 2 Liver segmentation performance stratified by AAST grade using fivefold cross-validation on the training set

From: Automated quantitative assessment of pediatric blunt hepatic trauma by deep learning-based CT volumetry

AAST grade

Dice (%)

Precision (%)

Recall (%)

RVD (%)

I (n = 6)

94.70 ± 0.64

95.44 ± 1.37

94.02 ± 2.12

−1.454 ± 3.460

II (n = 42)

95.21 ± 0.97

95.15 ± 1.69

95.31 ± 1.23

0.201 ± 2.403

III (n = 39)

94.57 ± 1.97

95.10 ± 1.66

94.14 ± 3.57

−0.969 ± 4.310

IV (n = 27)

93.86 ± 1.76

93.91 ± 2.56

93.92 ± 3.15

0.115 ± 4.934

V (n = 4)

95.02 ± 0.97

94.55 ± 1.12

95.52 ± 1.55

1.031 ± 2.023

Overall (n = 118)

94.66 ± 1.61

94.84 ± 1.93

94.55 ± 2.74

−0.261 ± 3.807