|| Login  
Krupinski, Lubin, Roehrig et al. (2006)

Reader and case sample combinations needed for 0.80 power based on the data of:

Krupinski EA, Lubin J, Roehrig H, Johnson J, Nafziger J. Using a human visual system model to optimize soft-copy mammography display: Influence of veiling glare. Academic Radiology 2006;13:289-295.

Readers: 6 board certified family physicians

Cases: 160 mammographic images, half containing a single mass

Conditions: images were processed to simulate four levels of veiling glare: (1) none, (2) comparable to a medical grade monochrome curved-screen cathode ray tube (CRT) display, (3) double that of the CRT and (4) quadruple that of the CRT. The images were shown on a liquid crystal display (LCD) that had essentially no veiling glare.

Results: Observer performance declined with increasing veiling glare, with quadruple veiling glare yielding significantly lower performance than the lower veiling glare levels.

 

ROC Model = PropROC, Parameter = AUC,

AUCs = 0.882, 0.872, 0.867, 0.798, respectively.

Readers = 6, Cases = 160,

MS( Treatment*Reader ) = 0.20555930,

MS( Treatment*Case ) = 0.09767099,

MS( Treatment*Reader *Case ) = 0.09186388.

 

AUC Difference = 0.03

Readers

Readers

and Cases

Random

Cases

Random

Readers

Random

3

.

553

.

4

.

420

.

5

.

340

.

6

.

286

.

7

.

248

.

8

.

220

.

9

.

197

.

10

.

180

.

11

.

165

.

12

.

153

.

13

.

143

.

14

.

134

.

15

.

126

.

Note: Cells containing a period indicate that more than 2000 cases would be required.

 

 

AUC Difference = 0.05

Readers

Readers

and Cases

Random

Cases

Random

Readers

Random

3

.

201

.

4

.

153

.

5

.

124

.

6

.

105

.

7

1518

91

1500

8

420

81

409

9

247

73

238

10

177

66

168

11

138

61

130

12

114

57

106

13

97

53

89

14

85

50

77

15

76

47

68

Note: Cells containing a period indicate that more than 2000 cases would be required.

 

AUC Difference = 0.10

Readers

Readers

and Cases

Random

Cases

Random

Readers

Random

3

.

52

.

4

224

40

234

5

86

33

87

6

54

28

54

7

40

25

39

8

32

22

31

9

27

20

25

10

23

20

22

11

20

20

20

12

20

20

20

13

20

20

20

14

20

20

20

15

20

20

20

Note: Cells containing a period indicate that more than 2000 cases would be required.

 

 


   Terms of Use   ||   Privacy Statement   ||   Copyright 2006 Medical Image Perception Laboratory