Skip to main content
. 2019 Apr;9(4):671–680. doi: 10.21037/qims.2019.04.14

Table 2. ROC results of partial DWI and DKI metrics histogram parameters.

Diagnostic index AUC Cutoff value Se (%) Sp (%) PPV (%) NPV (%) Accuracy (%)
ADC
     50th 0.64 1.455 60.0 (24/40) 72.7 (24/33) 72.7 (24/33) 60.0 (24/60) 65.8 (48/73)
     Mean 0.634 1.445 55.0 (22/40) 66.7 (22/33) 66.7 (22/33) 55.0 (22/40) 60.3 (44/73)
     Kurtosis 0.671 2.72 84.8 (28/33) 50.0 (20/40) 58.3 (28/48) 80.0 (20/25) 54.8 (40/73)
Dapp 25th 0.657 1.755 60.0 (24/40) 72.7 (24/33) 72.7 (24/33) 60.0 (24/60) 65.8 (48/73)
Kapp
     10th 0.719 0.225 66.7 (22/33) 70.0 (28/40) 64.7 (22/34) 71.8 (22/39) 60.3 (44/73)
     25th 0.87 0.455 66.7 (22/33) 92.5 (37/40) 88.0 (22/25) 77.1 (37/48) 80.8 (59/73)
     50th 0.881 0.655 75.8 (25/33) 85.0 (34/40) 80.6 (25/31) 81.0 (34/42) 80.8 (59/73)
     75th 0.876 0.835 81.8 (27/33) 85.0 (34/40) 81.8 (27/33) 85.0 (34/40) 83.6 (61/73)
     90th 0.867 0.935 87.9 (29/33) 77.5 (31/40) 76.3 (29/38) 88.6 (31/35) 82.2 (60/73)
     Mean 0.889 0.625 87.9 (29/33) 80.0 (32/40) 78.4 (29/37) 88.9 (32/36) 83.6 (61/73)
     SD 0.771 0.345 81.8 (27/33) 65.0 (26/40) 65.9 (27/41) 81.2 (26/32) 72.6 (53/73)

DWI, diffusion-weighted imaging; DKI, diffusion kurtosis imaging; AUC, area under the curve; Se, sensitivity; Sp, specificity; PPV, highest positive predictive value; NPV, highest negative predictive value; ADC, apparent diffusion coefficient.