Table 5.

Combinations of cases and readers having .80 power to detect a .05 AUC difference between spin-echo and cine MRI based on the Van Dyke et al [21] data. The pilot-study estimate of ${\widehat{\sigma }}_{TR}^2$ was negative. The left-hand-side results are for a planned study that has the same normal-to-abnormal ratio as the pilot study; the right-hand-side results are for a planned study that has equal numbers of normal and abnormal cases.

	normal/abnormal ratio = 69/45 = 1.53				normal/abnormal ratio = 1
	${\widehat{\sigma }}_{TR}^2=0$		${\widehat{\sigma }}_{TR}^2=.0001$		${\widehat{\sigma }}_{TR}^2=0$		${\widehat{\sigma }}_{TR}^2=.0001$
Readers	Cases	Power	Cases	Power	Cases	Power	Cases	Power
3	559	0.800	1898	0.800	374	0.800	1282	0.800
4	343	0.800	491	0.800	229	0.800	328	0.800
5	266	0.801	330	0.801	177	0.801	220	0.801
6	225	0.800	263	0.800	150	0.801	176	0.802
7	200	0.800	227	0.801	133	0.800	151	0.801
8	183	0.800	203	0.801	122	0.801	135	0.801
9	171	0.801	187	0.802	114	0.802	124	0.801
10	162	0.802	174	0.800	108	0.803	116	0.802
11	154	0.800	165	0.801	103	0.803	110	0.803
12	148	0.800	158	0.802	99	0.803	105	0.803
13	143	0.800	151	0.800	95	0.801	101	0.803
14	139	0.801	146	0.800	93	0.804	97	0.801
15	136	0.802	142	0.801	90	0.801	94	0.800