Table 1 Area under curve (AUC) for ROC analysis to discriminate between responders and non-responders. CI is 95% confidence interval.

From: Spatial Characterization of Tumor Perfusion Properties from 3D DCE-US Perfusion Maps are Early Predictors of Cancer Treatment Response

 

Data Set A/B (Train Data)

Data Set D (Test Data)

Patients (Test Data)

GLMNET

—

0.95 (CI: 0.83, 1.00)

0.97 (CI: 0.82, 1.00)

PCA-LDA1

0.97 (CI: 0.93, 1.00)

1.00 (CI: 1,1)

0.94 (CI: 0.73, 1.00)

PCA-LDA2

0.96 (CI: 0.91, 1.00)

0.88 (CI: 0.77, 1.00)

0.99 (CI: 0.9, 1.00)

ROI-LDA (Conventional)

0.78 (CI: 0.68, 0.87)

0.66 (CI: 0.40, 0.92)

0.39 (CI: -0.01, 0.79)

PE-LDA

0.74 (CI: 0.64, 0.83)

0.76 (CI: 0.53, 0.99)

0.61 (CI: 0.19, 1.00)

AUC-LDA

0.43 (CI: 0.33, 0.52)

0.51 (CI: 0.23, 0.79)

0.28 (CI: -0.08, 0.64)

MTT-LDA

0.62 (CI: 0.51, 0.72)

0.65 (CI: 0.39, 0.91)

0.17 (CI: -0.11, 0.45)

TP-LDA

0.43 (CI: 0.33, 0.52)

0.38 (CI: 0.11, 0.64)

0 (CI: 0, 0)

Back to article page