Table 2 Predictive performances of radiomics features from different sub-regions for predicting pCR in LA-ESCC

From: Sub-regional radiomics combining multichannel 2-dimensional or 3-dimensional deep learning for predicting neoadjuvant chemo-immunotherapy response in esophageal squamous cell carcinoma: a multicenter study

Variables

Sensitivity (95% CI)

Specificity (95% CI)

PPV (95% CI)

NPV (95% CI)

AUC (95% CI)

P-value (AUC)

Sub-region1

0.782 (0.614–0.948)

0.843 (0.753–0.881)

0.669 (0.472–0.801)

0.915 (0.843–0.964)

0.823 (0.719–0.908)

Reference

Sub-region2

0.738 (0.619–0.832)

0.691 (0.542–0.896)

0.646 (0.563–0.732)

0.899 (0.821–0.962)

0.757 (0.658–0.837)

0.291

Sub-region3

0.620 (0.457–0.784)

0.725 (0.596–0.813)

0.516 (0.355–0.652)

0.819 (0.737–0.884)

0.589 (0.481–0.675)

<0.001

Sub-region4

0.371 (0.213–0.509)

0.782 (0.627–0.873)

0.318 (0.159–0.421)

0.824 (0.716–0.914)

0.514 (0.405–0.604)

<0.001

Sub-region1 + 2

0.775 (0.614–0.908)

0.871 (0.749–0.954)

0.641 (0.502–0.756)

0.896 (0.807–0.975)

0.798 (0.682–0.872)

0.671

  1. AUC area under the receiver operating characteristic curve, CI confidence interval, CT computed tomography, LA-ESCC locally advanced esophageal squamous cell carcinoma, NPV negative predictive value, pCR pathological complete response, PPV positive predictive value.
Back to article page