Table 5 Diagnostic performance of ADC histogram parameters in differentiating between stage IA EC and BELs
Histogram features | Cut-off | AUC (95%CI) | Sensitivity (%) | Specificity (%) | Accuracy (%) |
|---|---|---|---|---|---|
ADC10th (× 10−3 mm2/s) | 0.973 | 0.920 (0.885–0.956) | 86.5 | 85.8 | 86.2 |
ADC90th (× 10−3 mm2/s) | 1.401 | 0.891 (0.849–0.933) | 88.9 | 75.5 | 82.8 |
ADCmin (× 10−3 mm2/s) | 0.829 | 0.876 (0.830–0.922) | 84.9 | 80.2 | 82.8 |
ADCmax (× 10−3 mm2/s) | 1.627 | 0.747 (0.684–0.810) | 68.3 | 68.9 | 68.5 |
ADCmean (× 10−3 mm2/s) | 1.147 | 0.926 (0.893–0.960) | 86.5 | 87.8 | 87.1 |
ADCmedian (× 10−3 mm2/s) | 1.161 | 0.928 (0.895–0.960) | 88.9 | 83 | 86.2 |
IQR (× 10−3 mm2/s) | 0.193 | 0.594 (0.520–0.668) | 52.4 | 64.2 | 57.8 |
MAD (× 10–3) | 0.111 | 0.595 (0.521–0.669) | 48.4 | 70.8 | 58.6 |
rMAD (× 10–3) | 0.08 | 0.603 (0.529–0.676) | 54 | 66 | 59.5 |
RMS (× 10–3) | 1.157 | 0.925 (0.890–0.959) | 85.7 | 86.8 | 86.2 |
Energy (× 10–3) | 0.366 | 0.673 (0.603–0.743) | 76.2 | 55.7 | 66.8 |
Total Energy (× 10–3) | 9.91 | 0.663 (0.592–0.734) | 74.6 | 57.5 | 66.8 |
Entropy (× 10–6) | 4.082 | 0.609 (0.534–0.683) | 55.6 | 65.1 | 59.9 |
Skewness (× 10–6) | 0.566 | 0.748 (0.685–0.811) | 63.5 | 80.2 | 71.1 |
Kurtosis (× 10–6) | 3.357 | 0.649 (0.579–0.720) | 52.4 | 77.4 | 63.8 |
Variance (× 10–6) | 0.063 | 0.583 (0.508–0.657) | 95.2 | 20.8 | 61.2 |
Uniformity (× 10–6) | 0.072 | 0.620 (0.547–0.694) | 52.4 | 70.8 | 60.8 |