Skip to main content

Table 2 Model performance in the training and validation sets

From: An MR-based radiomics model for differentiation between hepatocellular carcinoma and focal nodular hyperplasia in non-cirrhotic liver

Model

Cutoff

AUC(95%CI)

Specificity

Sensitivity

Accuracy

P

Training set

 Clinical Model

0.684

0.937(0.887-0.970)

0.923

0.817

0.853

Ref

 Radiomics Model

0.695

0.960(0.916-0.985)

0.942

0.904

0.917

0.252

 Combined Model

0.607

0.984(0.949-0.997)

0.962

0.952

0.956

0.002

Validation set

 Clinical Model

0.625

0.903(0.807-0.962)

0.826

0.867

0.853

Ref

 Radiomics Model

0.658

0.931(0.843-0.978)

0.826

0.889

0.868

0.535

 Combined Model

0.859

0.972(0.900-0.997)

0.957

0.933

0.941

0.032

  1. Note: AUC area under the curve, CI confidence interval