Table 1 Segmentation performance comparison across methods in the internal modeling and external validation cohorts

From: Automated detection of radiolucent foreign body aspiration on chest CT using deep learning

Internal Modeling

DSC (%)

VOE (%)

RVD (%)

ASSD (mm)

Pre (%)

FNR (%)

FPR (%)

MIOU (%)

MedSeg

86.54

20.43

19.37

0.76

99.70

20.52

1.29

0.854

MedpSeg

87.48

18.23

18.75

0.71

99.89

20.15

1.22

0.897

AG-Unet

85.24

21.37

20.13

0.82

99.57

21.45

1.35

0.793

External Validation

DSC (%)

VOE (%)

RVD (%)

ASSD (mm)

Pre (%)

FNR (%)

FPR (%)

MIOU (%)

MedSeg

85.35

21.25

20.58

0.81

99.58

21.23

1.45

0.837

MedpSeg

86.58

18.75

19.28

0.75

99.81

21.05

1.37

0.878

AG-Unet

84.57

22.15

21.58

0.85

99.33

22.13

1.58

0.775

  1. DSC dice similarity coefficient, VOE volumetric overlap error, RVD relative volume difference, ASSD average symmetric surface distance, Pre precision, FNR false negative rate, FPR false positive rate, MIOU mean intersection over union.
Back to article page