Table 3.

A. ROC Measures (Initial Sample)			Poor vs (Good + Excellent)	Good vs Excellent
MAXVOX	AUC (+− 95% CI)		0.892 (0.866–0.919)	0.881 (0.862–0.901)^##
	Sensitivity/Specificity		0.848/0.830	0.782/0.827
	Youden Index		0.678	0.609
	Cutoff		7041	2282
	Classification Accuracy		84.60%	79.92%
MEANVOX	AUC (+− 95% CI)		0.922 (0.899–0.944)	0.839 (0.816–0.862)
	Sensitivity/Specificity		0.852/0.844	0.790/0.733
	Youden Index		0.696	0.523
	Cutoff		866.9	426.1
	Classification Accuracy		85.11%	76.78%
MOTION	AUC (+− 95% CI)		0.899 (0.871–0.927)	0.787 (0.761–0.814)
	Sensitivity/Specificity		0.881/0.769	0.720/0.748
	Youden Index		0.65	0.468
	Cutoff		0.824	0.411
	Classification Accuracy		86.66%	73.22%
TSNR	AUC (+− 95% CI)		0.93 (0.908–0.952)^**	0.711 (0.68–0.741)
	Sensitivity/Specificity		0.873/0.857	0.706/0.611
	Youden Index		0.73	0.317
	Cutoff		6.47	7.16
	Classification Accuracy		87.03%	66.86%
B. Classification Accuracy of QA Measures
		Poor	Good	Excellent
Initial Sample	Stage 1	126 (86%)	1055 (87%)
Initial Sample	Stage 2	-	385 (82%)	581 (78%)
Follow-up Sample	Stage 1	10 (83%)	348 (96%)
Follow-up Sample	Stage 2	-	31 (52%)	283 (94%)

^**

Significantly higher AUC than MAXVOX and MOTION

^##

Significantly higher AUC than TSNR, MEANVOX, and MOTION