Skip to main content

Table 4 Performance of radiomics and clinical models in training and external validation cohort

From: Preoperative DBT-based radiomics for predicting axillary lymph node metastasis in breast cancer: a multi-center study

DBT

AUC

Accuracy

SEN

SPE

PPV

NPV

P

Training set

      

-

Signaturetuomor+10 mm

0.806

0.723

0.763

0.697

0.626

0.815

<

Clinical model

0.580

0.544

0.763

0.397

0.458

0.715

<0.001

Nomogram

0.813

0.751

0.750

0.752

0.669

0.819

0.051

Radiologist

0.583

0.654

0.706

0.354

0.231

0.936

<0.001

External validation set

Signaturetuomor+10 mm

0.785

0.740

0.787

0.717

0.569

0.877

-

Clinical model

0.569

0.438

0.936

0.202

0.358

0.870

<0.001

Nomogram

0.792

0.726

0.851

0.667

0.548

0.904

0.184

Radiologist

0.582

0.705

0.611

0.719

0.234

0.929

<0.001

  1. AUC: Area Under Curve, SEN: sensitivity, SPE: specificity, ACC: Accuracy, PPV: PositivePredictive Value, NPV: Negative Predictive Value. p values represent the delong test of the efficacy of each model compared with the Signaturetuomor+10 mm model