Table 6.

Sensitivity and specificity at optimal thresholds for each software and magnet strength. Metric: Mean rCBV.

Class. criteria	Software	1.5 T		3 T		1.5 T		3 T		1.5 T		3 T
		IB Neuro				FuncTool				nordicICE
		Thresh: 1.09		Thresh: 1.10		Thresh: 1.06		Thresh: 1.33		Thresh: 1.16		Thresh: 1.39
		Sens	Spec	Sens	Spec	Sens	Spec	Sens	Spec	Sens	Spec	Sens	Spec
Survival	IB Neuro	0.44	0.67a	0.80	0.63a	0.44b	0.56	0.40b	0.88	0.31b	0.67	0.40b	0.88
	FuncTool	0.60	0.44	1.00	0.38	0.60	0.44	0.80	0.75	0.47	0.44	0.60	0.75
	nordicICE	0.63	0.22a	0.80	0.25a	0.69b	0.22	0.80b	0.75	0.56b	0.44	0.80b	0.88
		Thresh: 0.87		Thresh: 1.41		Thresh: 0.98		Thresh: 1.39		Thresh: 1.13		Thresh: 1.50
		Sens	Spec	Sens	Spec	Sens	Spec	Sens	Spec	Sens	Spec	Sens	Spec
Treatment change	IB Neuro	0.79	0.55	0.57	0.90	0.53a,b	0.55	0.57a,b	0.90	0.26c,d	0.64	0.43c,d	1.00
	FuncTool	0.78	0.36	0.86	0.80	0.72b	0.45	0.86b	0.80	0.61c	0.45	0.57c	1.00
	nordicICE	0.89	0.27	0.71	0.90	0.84a	0.27	0.71a	0.80	0.68d	0.36	0.71d	1.00
		Thresh: 1.09		Thresh: 1.10		Thresh: 1.44		Thresh: 1.45		Thresh: 1.31		Thresh: 1.52
		Sens	Spec	Sens	Spec	Sens	Spec	Sens	Spec	Sens	Spec	Sens	Spec
Combined	IB Neuro	0.58	0.50	1.00	0.71	0.17	0.83	0.33	1.00	0.25	0.50	0.33	1.00
	FuncTool	0.82	0.17	1.00	0.43	0.36	0.67	1.00	0.86	0.45	0.33	0.67	1.00
	nordicICE	0.75	0.17	1.00	0.29	0.25	0.83	1.00	1.00	0.42	0.50	1.00	1.00

Note: Operator 1’s data. Comparison of sensitivity and specificity performed by pooling the 1.5 and 3 T cases of disagreement before calculating the test statistic.

^aComparison $p$ $\text{value}=0.016$

^bComparison $p$ $\text{value}=0.031$

^cComparison $p$ $\text{value}=0.046$

^dComparison $p$ $\text{value}=0.0047$