Display Modes for CT Colonography1
Part I. Synthesis and Insertion of Polyps into Patient CT Data
Chandu Karadi, PhD,
Christopher F. Beaulieu, MD, PhD,
R. Brooke Jeffrey, Jr, MD,
David S. Paik, MS and
Sandy Napel, PhD
1 From the Departments of Medicine (C.K., D.S.P.) and Radiology (C.F.B., R.B.J., S.N.), Stanford University School of Medicine, Lucas MRS Center P-268, Stanford, CA 94305-5488. Received July 20, 1998; revision requested September 24; revision received November 18; accepted January 11, 1999. Supported in part by National Institutes of Health grants 1R01 CA72023, 1P41 RR09784-01, and LM 07033, the Packard Foundation (Los Altos, Calif), the Lucas Foundation (Menlo Park, Calif), and the Phil N. Allen Trust (Menlo Park, Calif). C.F.B. is a 1997 RSNA Scholar. C.K. is a 1998 GENDEX/RSNA Medical Student/Scholar Assistant. Address reprint requests to S.N. (e-mail: snapel@stanford.edu).
View larger version (28K):
[in a new window]
|
Figure 2. Three volume-rendered images illustrate creation of phantom data. (a) A synthetic free-standing polyp is inserted, or composited (+), into (b) polyp-free patient colon data, which results (=) in (c) a new colon data set with a polyp (arrow). These images are 2D renderings of the component parts a and b and the result c.
|
|
View larger version (48K):
[in a new window]
|
Figure 3. Top: Axial CT sections separated by 1-mm increments through a 10-mm synthetic spherical polyp (arrow). Bottom: Coronal CT sections separated by 0.7-mm increments through the same 10-mm synthetic polyp (arrow).
|
|
View larger version (137K):
[in a new window]
|
Figure 4. A 3D volume-rendered view of four 10-mm polyps (arrows), including the polyp (large arrow) shown in Figure 3, placed in patient colon data.
|
|
View larger version (67K):
[in a new window]
|
Figure 5. Paired 3D volume-rendered images (top) and 2D axial CT sections (bottom). (a, c) Real polyp (arrow). (b, d) Synthetic polyp (arrow).
|
|
View larger version (11K):
[in a new window]
|
Figure 6a. Histograms show distribution of scores, summed over all reviewers, for real (white bars) and synthetic (black bars) polyps on (a) 3D volume-rendered images and (b) 2D axial CT sections. Scores could range from 1, for a "definitely synthetic" polyp, to 5, for a "definitely real" polyp. If readers could definitively distinguish synthetic from real lesions, responses should cluster around scores of 1 and 5, respectively. Instead, responses for both real and synthetic lesions were clustered around a score of 3, for "unsure" whether the polyp was real or synthetic.
|
|
View larger version (12K):
[in a new window]
|
Figure 6b. Histograms show distribution of scores, summed over all reviewers, for real (white bars) and synthetic (black bars) polyps on (a) 3D volume-rendered images and (b) 2D axial CT sections. Scores could range from 1, for a "definitely synthetic" polyp, to 5, for a "definitely real" polyp. If readers could definitively distinguish synthetic from real lesions, responses should cluster around scores of 1 and 5, respectively. Instead, responses for both real and synthetic lesions were clustered around a score of 3, for "unsure" whether the polyp was real or synthetic.
|
|
View larger version (14K):
[in a new window]
|
Figure 7. ROC curves for the three radiologist observers (pooled results). The observers attempted to distinguish real from synthetic lesions on 3D volume-rendered images (•) and 2D axial CT sections ( ). If the readers had been able to distinguish real from synthetic lesions, the ROC curves for 3D and/or 2D images would be positioned above and to the left of the curve that represents random guessing (dashed line). Instead, the observed ROC curves were not significantly different from random guessing. FPF = false-positive fraction, TPF = true-positive fraction.
|
|
Copyright © 1999 by the Radiological Society of North America.
) and the edge of patient's colon (
) as a function of pixel location (X in a, Z in b). As one moves from the lumen (air attenuation, -1,000 HU) to the polyp or colonic wall (soft-tissue attenuation, approximately 50 HU), the slope of the line describing the change in attenuation is determined by either in-plane or through-plane partial volume averaging. By matching the input parameters of the CT simulator with those of the base data, closely matched profiles can be obtained.