Quality Assessment for CT Colonography: Validation of Automated Measurement of Colonic Distention and Residual Fluid
Keshav K. Deshpande1,
Ronald M. Summers1,
Robert L. Van Uitert1,
Marek Franaszek1,
Linda Brown1,
Andrew J. Dwyer1,
Joel G. Fletcher2,
J. Richard Choi3 and
Perry J. Pickhardt4
1 Diagnostic Radiology Department, Clinical Center, National Institutes of
Health, Bldg. 10, Room 1C368X, Bethesda, MD 20892-1182.
2 Department of Radiology, Mayo Clinic, Rochester, MN.
3 Department of Radiology, Walter Reed Army Medical Center and Uniformed
Services University of the Health Sciences, Washington, DC.
4 Department of Radiology, University of Wisconsin Medical School, Madison,
WI.
View larger version (50K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1 —49-year-old man who underwent supine CT colonography.
Three-dimensional surface-rendered image shows colon separated into five
segments: ascending colon (proximal to orange divider), transverse
colon (between orange and purple dividers), descending colon
(between purple and red dividers), sigmoid colon (between
red and green dividers), and rectum (between green
and yellow dividers).
|
|
View larger version (60K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2 —53-year-old man who underwent prone CT colonography.
Three-dimensional surface-rendered image shows well-distended descending
colon. Automated quality assessment score was 98. Mean score from manual
assessment by radiologists using distention method 1 was 95.
|
|
View larger version (49K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3 —51-year-old man who underwent supine CT colonography.
Three-dimensional surface-rendered image shows poor distention of descending
colon. Automated quality assessment score was 29. Mean score from manual
assessment by radiologists using distention method 1 was 37.
|
|
View larger version (54K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4 —57-year-old man who underwent prone CT colonography.
Three-dimensional surface-rendered image shows ascending colon with low amount
of surface area obscured by residual fluid. Arrows indicate air–fluid
boundary. Automated quality assessment fluid score was 3. Mean score of
radiologists was 5.
|
|
View larger version (44K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 5 —51-year-old man who underwent supine CT colonography.
Three-dimensional surface-rendered image shows ascending colon with high
amount of surface area obscured by fluid. Arrows indicate air–fluid
boundary. Automated quality assessment fluid score was 83. Mean score of
radiologists was 75.
|
|
View larger version (8K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7A —Comparisons of automated and radiologist scoring with
distention method 1. Distention is expressed as percentage of colonic segment
of interest with diameter greater than 2 cm. Bland-Altman plot shows mean of
scores of radiologists and quality assessment software (x-axis)
compared with difference of mean scores of radiologists and quality assessment
software score (y-axis). Dashed line indicates mean difference
(0.1%); solid lines, lower limit (–25.6%) and upper limit (25.9%), where
limits are defined as 2 SD.
|
|
View larger version (8K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7B —Comparisons of automated and radiologist scoring with
distention method 1. Distention is expressed as percentage of colonic segment
of interest with diameter greater than 2 cm. Plot shows distention scores for
automated quality assessment versus mean from radiologists.
|
|
View larger version (7K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8A —Comparisons of automated and radiologist scoring of residual
colonic fluid. Amount of residual fluid is expressed as percentage of surface
area of colonic segment of interest covered by opacified fluid. Bland-Altman
plot shows mean of scores of radiologists and quality assessment software
(x-axis) compared with difference of mean scores of radiologists and
quality assessment software score (y-axis). Dashed line indicates
mean difference (–0.6%); solid lines, lower limit (–8.2%) and
upper limit (7.1%), where limits are defined as 2 SD.
|
|
View larger version (8K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8B —Comparisons of automated and radiologist scoring of residual
colonic fluid. Amount of residual fluid is expressed as percentage of surface
area of colonic segment of interest covered by opacified fluid. Plot shows
fluid scores for automated quality assessment versus mean scores of
radiologists.
|
|
View larger version (7K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 9 —Comparison of mean distention of colonic segment of interest
scored by three radiologists using distention methods 1 and 2 (n =
30) (r2 = 0.7752). Distention method 1 scores are fraction
of colonic segment of interest with diameter 2 cm or greater. Distention
method 2 scores are clinical scores in which 0 indicates complete collapse and
3 indicates optimal distention.
|
|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
Copyright © 2007 by the American Roentgen Ray Society.
