Measurement | Method | Nonparametric Coefficient | Significance (P Value) | Post Hoc Cutoff | Sensitivity | Specificity | Accuracy |
---|---|---|---|---|---|---|---|
Mean Ktrans | 1-Nordic | 0.606 | .002 | ≥0.05 min−1b | 80% | 78% | 79% |
Max Ktrans | 1-Nordic | 0.542 | .006 | ≥0.2 min−1 | 80% | 78% | 79% |
Mean kep | 1-Nordic | 0.446 | .03 | ≥0.27 min−1 | 47% | 78% | 58% |
Mean vp | 1-Nordic | 0.555 | .005 | >2c | 71% | 89% | 79% |
Max vp | 1-Nordic | 0.513 | .01 | >9c | 73% | 67% | 71% |
Mean ve | 1-Nordic | 0.566 | .004 | >12c | 80% | 78% | 79% |
Short AUC | 2-Simple | 0.410 | .047 | >12%c | 93% | 67% | 84% |
Intermediate AUC | 2-Simple | 0.478 | .018 | >14%c | 93% | 67% | 84% |
Delayed short AUC | 2-Simple | 0.556 | .005 | >20%c,d | 93% | 78% | 88% |
Note:—Delayed short AUC indicates ratio of AUC from the lesion over the superior sagittal sinus vascular input integrated between the end of the initial vascular washout and early progressive leakage phases; max, maximum; kep, reflux rate constant; ve, extravascular, extracellular volume fraction.
↵a The methods given are for model-independent “simple” calculations of the signal with time (method 2) versus the pharmacokinetic model calculations using nordicICE (method 1). The pixel selection algorithm around the superior sagittal sinus was used for the latter. The nonparametric correlation to categoric ranking of pathology is given by a Spearman ρ correlation. Post hoc arbitrary cutoff values are given for the most accurate performance for determining tumor, with tumor (pathology grading of 1–3) regarded as a positive case for sensitivity and specificity.
↵b Best performing modeled variable for correlation with pathology.
↵c A relative unit.
↵d Best performing model independent variable.