Table 2. Comparison among AUCs derived from the analysis using all ROIs across all seven patients (Cambridge data set).
Penumbra thresholda |
AUC |
MTT threshold |
Comparison | Zb | ||
---|---|---|---|---|---|---|
AUC | CI | OT | 90%-T | |||
Classic | ||||||
MTT Abs | 0.81 | 0.76–0.85 | 7.8 seconds | 8.6 seconds | Abs versus delay | 0.56 |
MTT delay | 0.79 | 0.74–0.84 | 2.8 seconds | 4.3 seconds | Abs versus ratio | 0.52 |
MTT ratio | 0.79 | 0.75–0.83 | 151% | 214% | Delay versus ratio | 0.06 |
Normalized | ||||||
MTT delay | 0.89 | 0.84–0.94 | 4.3 seconds | 5.2 seconds | ||
MTT ratio | 0.89 | 0.85–0.94 | 186% | 223% | Delay versus ratio | 0.15 |
Stringent | ||||||
MTT Abs | 0.84 | 0.79–0.90 | 8.1 seconds | 9.5 seconds | Abs versus delay | 0.68 |
MTT delay | 0.82 | 0.77–0.86 | 2.8 seconds | 5.4 seconds | Abs versus ratio | 1.28 |
MTT ratio | 0.79 | 0.76–0.88 | 169% | 232% | Delay versus ratio | 0.55 |
Abs, absolute; AUC, area under the curve; CI, confidence interval; MTT Abs, absolute MTT value; OT, optimal threshold; 90%-T, 90% specificity threshold (see text); ROI, region of interest.
See the ‘Methods' section for details.
Z-scores <1.96 are not statistically significant.