Observer Performance Using Monitors With Different Phosphors

1
Observer Performance Using Monitors With
Different Phosphors

• Elizabeth A. Krupinski, PhD
• Hans Roehrig, PhD
• University of Arizona
• Tucson, AZ

2
P104 vs P45

• P104
• bigger grain size
• fewer grains per pixel
• two components
• more non-uniformity
• Lower SNR for P104

3
Examples
P45 Phosphor P104 Phosphor
4
Max SNR / CRT Pixel
120
P45 Phosphor
80
.
P 104 Phosphor
40
0
0
20
40
60
80
Luminance (cd/m2)
5
Display Function
200.00
160.00
120.00
P104
P45
80.00
40.00
0.00
0.00
100.00
200.00
300.00
Command Level (ADU)
6
Veiling Glare
0.020
0.016

0.012
Veiling glare ratio

P45
0.008
0.004
P104
0
0
200
400
600
800
1000
Disk Diameter in of Pixels
7
Pixel Emitted intensity
1.2
1.0
P45
0.8
P104
0.6
0.4
0.2
0
2
4
6
8
10
Pixels in Pixel String
8
MTF
P45 P104
1.2
1.2
113 ADU
113 ADU
0.8
0.8
-
Vertical MTF
Vertical
MTF
200 ADU

200 ADU
0.4
0.4
113 ADU
113 ADU
Horizontal
MTF
Horizontal MTF
200 ADU
200 ADU
0.0
0.0
Frequency(lp/mm)
9
Luminance Uniformity
10
JND Performance
F 28.55 df 23 p 11
Methods Materials

• 20 chest images
• 5 lesion free
• 15 solitary pulmonary nodule
• Image Systems P45 P104 monitors
• 1600 x 1200 resolution
• Pixel size 0.245 mm
• 6 radiologists as observers

12
Images
- Conventional radiographs 150 mA 125 kVp -
CT confirmation - Digitized with VIDAR VXR -
2100 x 2552 x 16-bits
13
Viewing Conditions

• Eye-position recorded
• ASL 4000 SU with head-tracker
• Ambient lights off
• No time restrictions
• 2 weeks between sessions

14
ROC Az Results

• Mean Az P45 0.9235
• P104 0.8851
• ANOVA
• Treatment F 0.4516, p 0.5097
• Cases F 0.4038, p 0.9864
• Readers F 0.1548, p 0.9781
• No significant interaction effects

15
Eye-Position Results
16
Eye-Position Results

• Mean Viewing Times Overall
• - P45 22.36 seconds
• - P104 27.41 seconds
• 75 of the data. Future will split out
  lesion vs lesion-free images residents vs
  radiologists

17
Eye-Position Results

• Median Dwell Times x Decision (msec)
• P45 P104
• TP 2043
• FN 1528 1411
• FP 1986
• TN 470

18
Eye-Position Results

• Mean Time to First Hit nodules (msec)
• P45 P104
• TP 312
• FN 563

19
Conclusions

• Phosphor may not affect diagnostic accuracy
• May affect search parameters
• Might influence purchase clinical use choices

20
Caveats

• Detection not classification
• Pulmonary nodules only
• No image processing (window / level) allowed
• Task stimuli may influence results
• Further studies should be done