Skip to main content
. 2020 May 28;47(13):2968–2977. doi: 10.1007/s00259-020-04864-1

Table 2.

Area under the receiver operator characteristic curve (AUC) values of the prediction models for lymph node metastases. Four radiomic models using different input features were compared with conventional measures. The respective AUC values and 95% confidence intervals (95% CI) are depicted

Model Training cohort Testing cohort
n = 87 LN n = 66 LN
AUC (95% CI) AUC (95% CI)
LN short diameter 0.76 (0.65–0.86) 0.84 (0.74–0.94)
LN Volume 0.74 (0.62–0.84) 0.80 (0.67–0.91)
Expert rating 0.65 (0.59–0.70) 0.67 (0.61–0.74)
Radiomics-texture 0.78 (0.67–0.88) 0.78 (0.65–0.91)
Radiomics-shape 0.77 (0.66–0.86) 0.83 (0.69–0.93)
Radiomics-intensity 0.76 (0.65–0.86) 0.73 (0.57–0.88)
Radiomics-LBP 0.86 (0.77–0.94) 0.90 (0.78–1.00)
Radiomics-combined 0.89 (0.81–0.95) 0.95 (0.88–0.99)