Skip to main content

Table 2 Diagnostic efficiency of different models in the training cohort and test cohort

From: Differentiation of acute and chronic vertebral compression fractures using conventional CT based on deep transfer learning features and hand-crafted radiomics features

Model

Training cohort

Test cohort

AUC (95% CI)

ACC (%)

SEN (%)

SPE (%)

F1-score

AUC (95% CI)

ACC (%)

SEN (%)

SPE (%)

F1-score

Radiomics

0.973 (0.955-0.990)

92.3

95.2

92.0

0.950

0.854 (0.773-0.934)

84.6

77.8

75.1

0.802

DLR

0.992 (0.983-0.999)

94.9

94.4

96.1

0.958

0.871 (0.805-0.938)

81.7

88.9

70.7

0.855

Features Fusion

0.997 (0.994-0.999)

97.1

96.0

98.1

0.973

0.915 (0.855-0.974)

88.5

93.6

82.9

0.914

DLRN

0.998 (0.996-0.999)

98.2

95.3

97.0

0.975

0.946 (0.906-0.987)

90.1

92.6

85.2

0.925

  1. ACC Accuracy, SEN Sensitivity, SPE Specificity
Back to article page

BMC Musculoskeletal Disorders

ISSN: 1471-2474

Contact us