HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) AVI Movies Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Huang, A. Articles by Pickhardt, P. J. Search for Related Content PubMed PubMed Citation Articles by Huang, A. Articles by Pickhardt, P. J.

Teniae Coli–based Circumferential Localization System for CT Colonography: Feasibility Study¹

¹ From the Diagnostic Radiology Department, Clinical Center, National Institutes of Health, 10 Center Dr, MSC 1182, Bldg 10, Room 1C351, Bethesda, MD 20892-1182 (A.H., D.A.R., R.M.S., M.F.); NIBIB/CDRH, Joint Laboratory for the Assessment of Medical Imaging Systems, U.S. Food and Drug Administration, Rockville, Md (N.P.); Uniformed Services University of the Health Sciences, Bethesda, Md (J.R.C., P.J.P.); Walter Reed Army Medical Center, Washington, DC (J.R.C.); and National Naval Medical Center, Bethesda, Md (P.J.P.). Received February 24, 2006; revision requested April 25; revision received May 19; accepted June 8; final version accepted September 1. Supported by intramural programs of the Diagnostic Radiology Department of the Clinical Center, National Institutes of Health. D.A.R. supported by a fellowship through the Clinical Research Training Program, a public-private partnership supported jointly by the National Institutes of Health and a grant to the Foundation for the National Institutes of Health from Pfizer Pharmaceuticals Group. Address correspondence to R.M.S. (e-mail: rms{at}nih.gov).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

 
This HIPAA-compliant study, with institutional review board approval and informed patient consent, was conducted to retrospectively develop a teniae coli–based circumferential localization method for guiding virtual colon navigation and colonic polyp registration. Colonic surfaces (n = 72) were depicted at computed tomographic (CT) colonography performed in 36 patients (26 men, 10 women; age range, 47–72 years) in the supine and prone positions. For 70 (97%) colonic surfaces, the tenia omentalis (TO), the most visible of the three teniae coli on a well-distended colonic surface, was manually extracted from the cecum to the descending colon. By virtually dissecting and flattening the colon along the TO, the authors developed a localization system involving 12 grid lines to estimate the circumferential positions of polyps. A sessile polyp would most likely (at 95% confidence level) be found within ±1.2 grid lines (one grid line equals 1/12 the circumference) with use of the proposed method. By orienting and positioning the virtual cameras with use of the new localization system, synchronized prone and supine navigation was achieved. The teniae coli are extractable landmarks, and the teniae coli–based circumferential localization system helps guide virtual navigation and polyp registration at CT colonography.

Supplemental material: http://radiology.rsnajnls.org/cgi/content/full/243/2/551/DC1

© RSNA, 2007

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

 
Computed tomographic (CT) colonography is emerging as a tool for screening patients for colorectal polyps and cancer (1). Imaging the patient in both the prone and the supine positions increases sensitivity by improving the visualization of regions on one scan that may be obscured on the complementary scan owing to various factors (2–4). Specificity may be improved if the potential lesions that are visible on both scans are examined side by side (4).

Current reference systems for polyp registration at CT colonography generally can be grouped into two categories. In one system, the colon is divided into different anatomic sections (5); the other system relies on the endoluminal centerline distance along the colon (6,7). The sectional approach is not precise because the anatomic sections are relatively long and their boundaries are poorly defined. The centerline approach is more precise. However, the colon rotates between scan acquisitions such that no reliable orientation information is available to pinpoint the circumferential position of a lesion on the colonic wall. This drawback makes polyp registration between scans time consuming and subject to error (5).

The teniae coli are three approximately 8-mm-wide longitudinal bands of smooth muscle that extend from the cecum to the sigmoid colon (8). The widths of the teniae remain fairly constant along the length of the colon (9) until they broaden to occupy more of the circumference of the sigmoid colon in its distal portion and unite to form a complete longitudinal muscle covering for the rectum. The teniae are not conspicuous on two-dimensional CT images. In fact, the wall of the colon should be barely perceptible if the colon is well distended (8). However, at external inspection of three-dimensional (3D) air-distended colonic surfaces, the teniae coli are readily visible as continuous longitudinal flat bands (10,11) (Fig 1).

View larger version (47K): [in this window] [in a new window] [Download PPT slide]   Figure 1a: (a) Anterior, (b) posterior, and (c) posterior oblique views of a 3D CT colonographic surface show the teniae coli in a 61-year-old woman in the supine position. The tenia omentalis (TO, solid arrows) is located on the anterior surface of the transverse colon. (a, b) The tenia libera (TL, dashed arrow) appears on the inferior aspect of the transverse colon, whereas the tenia mesocolica (TM, arrowheads) is located on the posterior surface. In c, the TO is clockwise from the ileocecal valve (arrowhead).

View larger version (44K): [in this window] [in a new window] [Download PPT slide]   Figure 1b: (a) Anterior, (b) posterior, and (c) posterior oblique views of a 3D CT colonographic surface show the teniae coli in a 61-year-old woman in the supine position. The tenia omentalis (TO, solid arrows) is located on the anterior surface of the transverse colon. (a, b) The tenia libera (TL, dashed arrow) appears on the inferior aspect of the transverse colon, whereas the tenia mesocolica (TM, arrowheads) is located on the posterior surface. In c, the TO is clockwise from the ileocecal valve (arrowhead).

View larger version (71K): [in this window] [in a new window] [Download PPT slide]   Figure 1c: (a) Anterior, (b) posterior, and (c) posterior oblique views of a 3D CT colonographic surface show the teniae coli in a 61-year-old woman in the supine position. The tenia omentalis (TO, solid arrows) is located on the anterior surface of the transverse colon. (a, b) The tenia libera (TL, dashed arrow) appears on the inferior aspect of the transverse colon, whereas the tenia mesocolica (TM, arrowheads) is located on the posterior surface. In c, the TO is clockwise from the ileocecal valve (arrowhead).

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

 
This retrospective study was Health Insurance Portability and Accountability Act compliant. Informed patient consent was obtained at the time of the initial institutional review board–approved study (12) and included consent for future retrospective data use. Viatronix (Stony Brook, NY) supplied the virtual 3D colon navigation software used in the study free of charge. Four authors (R.M.S., A.H., D.A.R., M.F.) have patents pending and/or have been awarded patents for the navigation system described in this article and/or related subject matter and will receive royalty income for a patent license from iCAD (Nashua, NH). R.M.S. and M.F. have received royalty income for a temporary patent license from Viatronix and iCAD. P.J.P. and J.R.C. are on the medical advisory board of Viatronix.

Study Group
The study data set consisted of data for 36 patients selected from a population of 468 asymptomatic adults at one institution (San Diego Naval Medical Center, San Diego, Calif). These patients' data were from a larger multiinstitutional database of previously described CT colonographic data (12). All patients underwent both CT colonography and colonoscopy on the same day. The following three criteria were used in selecting the study data set: (a) There was at least one colonoscopy-proved polyp 0.6 cm or greater in diameter, (b) the polyp was located between the cecum and the descending colon (inclusive), and (c) the polyp was visible on both supine and prone images. Eighty-five of the 468 patients satisfied criterion 1, 50 satisfied criteria 1 and 2, and 36 satisfied all three criteria. The final study data set consisted of data for 26 men and 10 women ranging in age from 47 to 72 years (mean age, 59 years).

There were 47 polyps in the study data set: 13 (28%) were pedunculated and 0.7–1.7 cm in diameter, and 34 (72%) were nonpedunculated (33 sessile, one flat) and 0.6–4.2 cm in diameter. Optical colonoscopy revealed the sizes and shapes of all polyps except one 0.8-cm adenoma, whose shape information was not recorded. For this polyp, the CT description (sessile) was used instead. Of the 47 studied polyps, two were cancerous (1.0 and 4.2 cm), 38 were adenomatous (0.6–1.8 cm), five were hyperplastic (0.7–1.2 cm), and two were other benign polyps (1.1 and 1.5 cm). Of the 13 pedunculated polyps, three were located in the cecum, three were located in the ascending colon, three were located in the transverse colon, one was located in the splenic flexure, and three were located in the descending colon. Of the 34 nonpedunculated polyps, four were located in the cecum, nine were located in the ascending colon, three were located in the hepatic flexure, seven were located in the transverse colon, one was located in the splenic flexure, and 10 were located in the descending colon.

The supine and prone x, y, and z coordinates of the center of the head of 46 polyps depicted on the CT images were available in the database from a previous study (13). For one polyp, only the supine coordinates were available. Two authors (A.H., D.A.R.), who were trainees with less than 1 year of experience, reviewed the coordinates by using both in-house and commercially available (V3D Colon, version 1.3.0.0; Viatronix) image analysis software. If any discrepancy existed between the database and surface models, it was resolved by a senior radiologist (R.M.S., 8 years of experience), who examined both the two-dimensional CT and the 3D surface images.

CT Colonography Protocol
Patients underwent a standard 24-hour colonic preparation: oral administration of 90 mL of sodium phosphate (Fleet 1; Fleet Pharmaceuticals, Lynchburg, Va) and 10 mg of bisacodyl (Boehringer Ingelheim Consumer Healthcare, Ridgefield, Conn). As part of their clear-liquid diet, patients also drank 500 mL of barium (2.1% by weight, Scan C; Lafayette Pharmaceuticals, Lafayette, Ind) for solid-stool tagging and 120 mL of diatrizoate meglumine and diatrizoate sodium (Gastrografin; Bracco Diagnostics, Princeton, NJ) for opacification of the luminal fluid.

Maximum colonic distention was achieved by means of patient-controlled rectal insufflation of room air. Scanning was performed in the supine and prone positions while the patient held his or her breath. A four-channel CT scanner (GE LightSpeed Plus; GE Medical Systems, Waukesha, Wis) with a 4 x 2.5-mm detector configuration was used. The CT technique involved the use of 2.5-mm collimation, a table speed of 15 mm per second, a reconstruction interval of 1 mm, and scanner settings of 100 mAs and 120 kVp. Images were reconstructed on a 512 x 512 matrix, which resulted in an in-plane pixel diameter of 0.596–0.865 mm. With use of the in-house software (14), 3D CT colonographic colon surfaces were reconstructed with the residual stool tagged and the fluid electronically removed.

Data Processing
TO detection and colon flattening.—The TO is the most easily identifiable of the three teniae coli because it is not obscured at the hepatic and splenic flexures. It generally occupies an anterior position in the transverse colon and a lateral position in the ascending and descending colonic segments (Fig 1). The TM and TL appear, respectively, posteriorly and inferiorly in the transverse colon and posteromedially and anteromedially in the ascending and descending colons.

At visualization of the external appearance of CT colonography–depicted colonic surfaces, the TO path can be identified efficiently as follows: We first look for the TO in the cecum by locating the ileocecal valve, which has the appearance of an indentation on the external surface (Fig 1c). The TO is located at about one-third of the colon circumference, clockwise from the ileocecal valve as viewed from the cecum toward the ascending colon. Once the TO is found, sequential TO points are manually chosen every 2–10 cm, depending on the local surface tortuosity, from the cecum to the descending colon by tracing the continuous longitudinal TO path (Fig 1). One author (A.H.), who was blinded to the polyp location, extracted the TO by using software developed by us and written in Visual C++ language (Microsoft, Richmond, Wash) with Qt (Oslo, Norway) and Open Inventor libraries. The software was run on a Dell workstation (Dell, Austin, Tex) with dual 1.5-GHz Xeon processors (Intel, Santa Clara, Calif), 2 GB of memory, and a Wildcat II 5110 graphics card (3Dlabs, Milpitas, Calif). The TO points were selected from the cecum through the descending colon only, because the teniae coli become difficult to distinguish anatomically in the sigmoid colon. The manual extraction generally takes 10–15 minutes per colonic surface. The shortest path (graph geodesic) through the TO points was then calculated and drawn on the surface by the software (15).

Next, the surface was virtually dissected along the TO path and flattened onto a rectangular sheet by using a uniform parameterization method (16); both long edges of the sheet corresponded to the TO (Fig 2). Assuming that the teniae were equidistant from each other, our software inferred the positions of the TM and TL (green and yellow stripes, respectively, in Fig 2) and drew them longitudinally such that the sheet was divided into smaller rectangular thirds. Nine additional blue lines were drawn such that a group of three lines divided each rectangular third longitudinally into four smaller rectangles. All 12 lines on the rectangular sheet were then mapped back onto the original surface (Fig 3) and reviewed by a separate author (D.A.R.). For those surfaces where there was doubt regarding the path of the TO, agreement was reached by consensus between two authors (A.H., D.A.R.) by examining both the flattened (Fig 2) and the 3D colonic surfaces (Fig 3). The lines corresponding to the teniae coli, by dividing the surface into circumferential thirds, served as the major grid lines for the localization system. The remaining nine lines served as minor grid lines.

View larger version (87K): [in this window] [in a new window] [Download PPT slide]   Figure 2: Segment of the two-dimensional flattened sheet derived from the 3D CT colonographic surface shown in Figure 1. The top and bottom edges of the sheet are formed by the TO (solid blue arrows). The TL (yellow stripe) and TM (green stripe) are inferred, respectively, at one-third and two-thirds the height of the sheet and approximate the longitudinal flat bands (arrowheads and dashed arrows) between the haustral folds (white arrows). A 0.9-cm adenomatous polyp (circle) can be seen touching the TM.

View larger version (52K): [in this window] [in a new window] [Download PPT slide]   Figure 3a: Teniae coli and minor grid lines on (a) anterior and (b) posterior views of the 3D CT colonographic surface shown in Figure 1. The TO (solid arrows), TL (dashed arrow), and TM (arrowheads) were mapped from the two-dimensional sheet (Fig 2) back to the 3D surface.

View larger version (49K): [in this window] [in a new window] [Download PPT slide]   Figure 3b: Teniae coli and minor grid lines on (a) anterior and (b) posterior views of the 3D CT colonographic surface shown in Figure 1. The TO (solid arrows), TL (dashed arrow), and TM (arrowheads) were mapped from the two-dimensional sheet (Fig 2) back to the 3D surface.

View larger version (9K): [in this window] [in a new window] [Download PPT slide]   Figure 4: Local coordinate systems along the centerline (dotted line). The z direction is the tangent of the centerline, the y direction is the vector pointing to the TO (dashed line), and the x direction is the cross product of the y and z directions.

View larger version (111K): [in this window] [in a new window] [Download PPT slide]   Figure 5: Synchronized supine (left image) and prone (right image) navigation. A 0.6-cm adenomatous polyp in the transverse colon of a 65-year-old man, partially colored orange (arrows) on images created by using a computer-aided detection system, has circumference positions of 5.0 and 5.0 grid lines in the supine (arrowhead) and prone (dashed arrow) positions. The camera-controlling text boxes under the images orient the camera (from top to bottom) 49 cm away from the cecum, at a 0° rotation angle, at a 3° pitch up angle, and at a 15-unit (one unit = 0.1 radians) field of view (zoom factor); and orient the right camera 3 cm ahead in the centerline position. Analogous to clock positions, the TO (red-yellow-green stripe), TL (yellow stripe), and TM (green stripe) are labeled grid lines 12, 4, and 8, respectively.

Longitudinal position.—One author (A.H.) measured the longitudinal position of a given polyp by placing the camera view on the center of the polyp with use of the teniae coli–guided navigation system (Fig 5). The system measured the length from the polyp to the cecum along the centerline.

Circumferential position.—Since not all stalks of pedunculated polyps and not all points of stalk attachment to the colon were conspicuous, the center of the head of the polyp was used to measure the positions of both pedunculated and nonpedunculated polyps.

In the cecum-to-rectum direction, the TO, TL, and TM occur in this order in a clockwise orientation. Each grid line is numbered from 1 to 12 clockwise, with the TO occupying the top position and labeled "grid line 12" (Fig 5). The circumferential position of each polyp was estimated by one author (A.H. or D.A.R.), and image findings were recorded. The circumferential position of each polyp was determined by visually estimating where each polyp was located on or between the grid lines. For example, the polyps at the tips of the arrowheads and arrows in Figure 5 are at circumferential positions 5.0 and 5.0, respectively. For each polyp, a signed circumferential position difference between scans was computed by subtracting the supine scan position of the polyp from the prone scan position.

Statistical Analyses
We generated a predictive colonic surface patch (to estimate the error in polyp localization) from 95% prediction intervals of the circumferential position difference and the camera position offset difference by using a one-sample model. We predicted that on the basis of a given number of sampled polyps (n), the circumferential position and camera position offset of a subsequent n + 1th polyp would fall within the prediction bounds with a 95% confidence level. The prediction interval was calculated by using computer software (Microsoft Excel, version 11.0; Microsoft) and the following formula (17):

where t_/2 is the critical value in a two-tailed t test, is the sample mean, and s is the standard deviation. We assumed that the sample was drawn from a normally distributed population; normality tests were performed by using Matlab, version 7.04.352 (R14) Service Pack 2 (MathWorks, Natick, Mass). To assess for the effect of intrasubject correlations, the longitudinal and circumference position differences were analyzed by using generalized estimating equations (SAS, version 9.1 [TS1M3]; SAS Institute, Cary, NC).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

 
We were able to manually extract the TO from the cecum to the descending colon for 70 (97%) of the 72 colonic surfaces. For two (3%) colonic surfaces, we were able to extract the TO from the cecum to only the hepatic or splenic flexure because of collapsed colonic walls. For all patients, the TM and TL paths inferred from virtual dissection and flattening approximated our visually identified teniae coli paths on the flattened and 3D colonic surfaces.

Six (8%) of the 72 colonic surfaces had erroneous TL coordinates in a small section (1–15 cm long) of the inferred TL. The causes of these erroneous coordinates were leakages into the small bowel (three cases), holes on segmented surfaces (one case), and polyps touching different colonic sections (two cases). These errors were corrected manually. As a result, the synchronous navigation was possible from the cecum to the hepatic flexure for one (3%) of 36 patients, from the cecum to the splenic flexure for one (3%) patient, and from the cecum to the descending colon for 34 (94%) patients.

Forty-five (96%) (12 of 13 pedunculated polyps, 33 of 34 nonpedunculated polyps) of the 47 studied polyps were within the range of extractable TO on both scanned positions. The remaining two polyps were a 0.8-cm sessile polyp and a 1.2-cm pedunculated adenoma located in the poorly distended distal descending colon, where the TO became untraceable.

Circumferential Position
Mean values, standard deviations, and 95% prediction intervals for signed circumferential position differences (expressed in numbers of grid lines) are summarized in Table 1. The results indicate that the pedunculated polyps could move more than a third of the circumference (±4.8 grid lines) whereas the nonpedunculated polyps were relatively immobile (±1.2 grid lines).

View larger version (85K): [in this window] [in a new window] [Download PPT slide]   Figure 6a: A 1.6-cm sessile adenoma in the transverse colon of a 52-year-old woman with a –0.5 grid line difference. (a) On supine CT scan, the polyp (thick arrow) is in the dependent location. (b) On prone CT scan, the polyp (thick arrow) has moved to the lateral location. In a and b, a small amount of dependent contrast material–enhanced residual fluid (thin arrows) is present. Corresponding (c) supine and (d) prone measurements on 3D endoluminal images obtained by using our circumferential localization method show polyp (arrow) positions at 8.6 and 8.1 grid lines, respectively.

View larger version (87K): [in this window] [in a new window] [Download PPT slide]   Figure 6b: A 1.6-cm sessile adenoma in the transverse colon of a 52-year-old woman with a –0.5 grid line difference. (a) On supine CT scan, the polyp (thick arrow) is in the dependent location. (b) On prone CT scan, the polyp (thick arrow) has moved to the lateral location. In a and b, a small amount of dependent contrast material–enhanced residual fluid (thin arrows) is present. Corresponding (c) supine and (d) prone measurements on 3D endoluminal images obtained by using our circumferential localization method show polyp (arrow) positions at 8.6 and 8.1 grid lines, respectively.

View larger version (111K): [in this window] [in a new window] [Download PPT slide]   Figure 6c: A 1.6-cm sessile adenoma in the transverse colon of a 52-year-old woman with a –0.5 grid line difference. (a) On supine CT scan, the polyp (thick arrow) is in the dependent location. (b) On prone CT scan, the polyp (thick arrow) has moved to the lateral location. In a and b, a small amount of dependent contrast material–enhanced residual fluid (thin arrows) is present. Corresponding (c) supine and (d) prone measurements on 3D endoluminal images obtained by using our circumferential localization method show polyp (arrow) positions at 8.6 and 8.1 grid lines, respectively.

View larger version (104K): [in this window] [in a new window] [Download PPT slide]   Figure 6d: A 1.6-cm sessile adenoma in the transverse colon of a 52-year-old woman with a –0.5 grid line difference. (a) On supine CT scan, the polyp (thick arrow) is in the dependent location. (b) On prone CT scan, the polyp (thick arrow) has moved to the lateral location. In a and b, a small amount of dependent contrast material–enhanced residual fluid (thin arrows) is present. Corresponding (c) supine and (d) prone measurements on 3D endoluminal images obtained by using our circumferential localization method show polyp (arrow) positions at 8.6 and 8.1 grid lines, respectively.

View larger version (94K): [in this window] [in a new window] [Download PPT slide]   Figure 7a: A 1.6-cm pedunculated villous adenoma in the descending colon of a 63-year-old man with a –3.1 grid line difference. (a) On 3D endoluminal supine image, the polyp head (arrow) rests within a haustrum at 10.1 grid lines. (b) On 3D endoluminal prone image, the polyp head (arrow) touches a haustral fold at 7.0 grid lines. (c) On supine CT scan, the stalk of the polyp is visible. The polyp head (large arrow) and part of the stalk (small white arrow) are submerged in a pool of contrast-enhanced residual fluid. (d) On prone CT scan, the polyp head (large arrow) has moved owing to gravity and is touching the haustral fold. Small blue arrows in c and d indicate the approximate viewing directions for corresponding 3D endoluminal images in a and b and were generated by the colon navigation software.

View larger version (94K): [in this window] [in a new window] [Download PPT slide]   Figure 7b: A 1.6-cm pedunculated villous adenoma in the descending colon of a 63-year-old man with a –3.1 grid line difference. (a) On 3D endoluminal supine image, the polyp head (arrow) rests within a haustrum at 10.1 grid lines. (b) On 3D endoluminal prone image, the polyp head (arrow) touches a haustral fold at 7.0 grid lines. (c) On supine CT scan, the stalk of the polyp is visible. The polyp head (large arrow) and part of the stalk (small white arrow) are submerged in a pool of contrast-enhanced residual fluid. (d) On prone CT scan, the polyp head (large arrow) has moved owing to gravity and is touching the haustral fold. Small blue arrows in c and d indicate the approximate viewing directions for corresponding 3D endoluminal images in a and b and were generated by the colon navigation software.

View larger version (28K): [in this window] [in a new window] [Download PPT slide]   Figure 7c: A 1.6-cm pedunculated villous adenoma in the descending colon of a 63-year-old man with a –3.1 grid line difference. (a) On 3D endoluminal supine image, the polyp head (arrow) rests within a haustrum at 10.1 grid lines. (b) On 3D endoluminal prone image, the polyp head (arrow) touches a haustral fold at 7.0 grid lines. (c) On supine CT scan, the stalk of the polyp is visible. The polyp head (large arrow) and part of the stalk (small white arrow) are submerged in a pool of contrast-enhanced residual fluid. (d) On prone CT scan, the polyp head (large arrow) has moved owing to gravity and is touching the haustral fold. Small blue arrows in c and d indicate the approximate viewing directions for corresponding 3D endoluminal images in a and b and were generated by the colon navigation software.

View larger version (30K): [in this window] [in a new window] [Download PPT slide]   Figure 7d: A 1.6-cm pedunculated villous adenoma in the descending colon of a 63-year-old man with a –3.1 grid line difference. (a) On 3D endoluminal supine image, the polyp head (arrow) rests within a haustrum at 10.1 grid lines. (b) On 3D endoluminal prone image, the polyp head (arrow) touches a haustral fold at 7.0 grid lines. (c) On supine CT scan, the stalk of the polyp is visible. The polyp head (large arrow) and part of the stalk (small white arrow) are submerged in a pool of contrast-enhanced residual fluid. (d) On prone CT scan, the polyp head (large arrow) has moved owing to gravity and is touching the haustral fold. Small blue arrows in c and d indicate the approximate viewing directions for corresponding 3D endoluminal images in a and b and were generated by the colon navigation software.

View larger version (109K): [in this window] [in a new window] [Download PPT slide]   Figure 8a: Initial polyp mislocalizations corrected by using proposed teniae coli–based navigation system. Synchronized 3D endoluminal (a) supine and (b) prone views of a 0.8-cm sessile adenoma (arrow) in the ascending colon of 65-year-old man. Synchronized 3D endoluminal (c) supine and (d) prone views of a polyp-like bump (arrow) 7 cm away from the polyp depicted in a and b. The polyp-like bump in d was initially mislabeled as a match to the polyp in a.

View larger version (110K): [in this window] [in a new window] [Download PPT slide]   Figure 8b: Initial polyp mislocalizations corrected by using proposed teniae coli–based navigation system. Synchronized 3D endoluminal (a) supine and (b) prone views of a 0.8-cm sessile adenoma (arrow) in the ascending colon of 65-year-old man. Synchronized 3D endoluminal (c) supine and (d) prone views of a polyp-like bump (arrow) 7 cm away from the polyp depicted in a and b. The polyp-like bump in d was initially mislabeled as a match to the polyp in a.

View larger version (113K): [in this window] [in a new window] [Download PPT slide]   Figure 8c: Initial polyp mislocalizations corrected by using proposed teniae coli–based navigation system. Synchronized 3D endoluminal (a) supine and (b) prone views of a 0.8-cm sessile adenoma (arrow) in the ascending colon of 65-year-old man. Synchronized 3D endoluminal (c) supine and (d) prone views of a polyp-like bump (arrow) 7 cm away from the polyp depicted in a and b. The polyp-like bump in d was initially mislabeled as a match to the polyp in a.

View larger version (109K): [in this window] [in a new window] [Download PPT slide]   Figure 8d: Initial polyp mislocalizations corrected by using proposed teniae coli–based navigation system. Synchronized 3D endoluminal (a) supine and (b) prone views of a 0.8-cm sessile adenoma (arrow) in the ascending colon of 65-year-old man. Synchronized 3D endoluminal (c) supine and (d) prone views of a polyp-like bump (arrow) 7 cm away from the polyp depicted in a and b. The polyp-like bump in d was initially mislabeled as a match to the polyp in a.

To eliminate dependencies due to within-patient correlations, we recomputed the data presented in Tables 1 and 2 by using only the largest polyp in each patient. The resultant 95% predictive intervals were similar (for circumferential and longitudinal camera offset differences: –5.42 to 5.07 grid lines and –6.89 to 6.64 cm, respectively, for pedunculated polyps [n = 8]; –0.88 to 1.20 grid lines and –5.56 to 6.25 cm, respectively, for nonpedunculated polyps [n = 26]; –2.12 to 2.28 grid lines and –5.45 to 5.92 cm, respectively, for all polyps [n = 34]). Normality tests of the polyp sample revealed that the assumption of normal distribution was reasonable. The mean values of the signed circumferential and longitudinal differences in polyp position between scans estimated by using generalized estimating equation analysis were nearly identical (–0.20 grid lines and 0.39 cm, respectively, for pedunculated polyps [n = 12]; 0.20 grid lines and 0.19 cm, respectively, for nonpedunculated polyps [n = 33]; 0.07 grid lines and 0.29 cm, respectively, for all polyps [n = 45]).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

 
The centerline approach (6,7) is one of the most precise localization systems currently available for CT colonography. Our study results indicate that a polyp found on one scan would most likely (with 95% confidence) be found within ±6.1 cm on the complementary scan with use of our navigation system. Although different software was used to measure the centerline distance in our study, our results are in agreement with those in previous studies (6,7). The searching range will be more precise if the circumferential position on the colonic wall is known. However, the colon rotates between the supine and prone scan acquisitions, and to our knowledge, no unambiguous circumferential localization schemes are currently available. A good indication of colon movement is the position change of a sessile polyp. In a study of 41 polyps performed by using a dependent location criterion, it was reported that six sessile polyps moved from a ventral location to a dorsal location relative to the colonic surface (5).

We have developed a circumferential localization system of measuring the positions of a polyp by referring to the teniae coli. Teniae coli are recognizable intrinsic landmarks that can be extracted from CT colonographic surfaces by finding the TO path and virtually flattening the colon along the TO. Along the three teniae coli with nine additional grid lines drawn on the colonic wall, polyps can be located and correlated between scans within an accuracy of 1/12 the circumference. With use of our localization scheme, the mean signed circumferential position differences between scans were 0.17 grid line ± 0.49 and 0.02 grid line ± 2.07 for the nonpedunculated and pedunculated polyps, respectively. Of 33 nonpedunculated polyps within the range of extractable TO on both scanned positions, only one had a circumferential difference greater than one grid line. Of 12 pedunculated polyps within the range of extractable TO on both scanned positions, five (42%) differed by greater than one grid line. The higher circumferential position differences for these pedunculated polyps, except one polyp with which there was a surface segmentation issue, were caused mainly by their intrinsic mobility.

Results of our study indicate that a nonpedunculated or pedunculated polyp seen on one scan would most likely (with 95% confidence) be found within ±1.2 or ±4.8 grid lines, respectively, on the complementary scan. The low circumferential localization error in matching a nonpedunculated polyp on the complementary scan corresponds well with the fact that a sessile polyp is fixed on the colonic wall and has essentially no mobility. Since radiologists are often able to classify polyps as sessile or pedunculated, knowledge of these mobility characteristics helps to define the search area on the complementary scan. However, the high circumferential localization error in matching a pedunculated polyp serves as a reminder that the potential pitfalls associated with lesion mobility must be recognized to maintain an acceptable level of specificity.

Incorporating both the centerline and the circumference information, we built and tested a synchronous navigation system prototype that allowed the user to examine the same colonic region side by side in different scanned positions. The synchronized views were achieved by using two mechanisms: the TO-defined local coordinate system and an adjustable centerline offset. By using the navigation system to review a data set for 36 patients, we were able to find the complementary location of one polyp that was not found previously and to identify two mismatched polyps. Our navigation system, providing the capability to present the complementary images in the same unambiguous orientation, narrows the searching area from a full circumference of a colon section to two tenths the circumference (±1.2 grid lines) for matching a nonpedunculated polyp and to eight tenths the circumference (±4.8 grid lines) for matching a pedunculated polyp. As CT colonographic spatial resolution improves, it may become possible to identify the base of the stalk of pedunculated polyps to improve localization accuracy.

We believe that our teniae coli–based navigation method has, in addition to the three potential applications just described (more accurate localization of polyps, more accurate mobility assessment of suspicious colonic findings, and matching of lesions found on prone and supine scans), several other potential applications, including orienting and coordinating two-dimensional CT scans between acquisitions and synchronizing current and prior examinations. For all of these applications, a human observer will have an important role in verifying accuracy.

A limitation of our study was that we did not compare the accuracy of different human readers in the task of polyp matching on the supine and prone scans with and without the aid of our localization system. There are insufficient data in the literature regarding how well radiologists perform in such matching tasks. We can only report that the described teniae coli–guided system helped us find three mismatched polyps and one unmatched polyp.

Our teniae coli detection method has several limitations. It is not fully automatic: The user must manually select points along the TO. Identification of the TO is user dependent and thus introduces potential errors. Detection of the teniae coli does not include identification of the distal colon because the teniae coli become difficult to distinguish anatomically in the sigmoid colon as the muscle fibers begin to fan out into the rectum. Diverticulosis in the sigmoid colon can also impair tenia coli visibility. Therefore, polyps in the sigmoid colon and rectum cannot be localized on the colon circumference. Because our teniae coli detection method and the application of it in CT colonography are, to our knowledge, the first to be reported, the aforementioned limitations might eventually be minimized or eliminated as more advanced technologies are developed.

In conclusion, we found that teniae coli identification is possible and that the proposed teniae coli–based circumferential localization system complements the conventional centerline approach for virtual colon navigation and polyp registration in CT colonography.

	ADVANCES IN KNOWLEDGE

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

 

• The teniae coli are recognizable unambiguous anatomic landmarks that can be used to estimate the circumferential positions of potential lesions at CT colonography.
  
• The tenia omentalis (TO) extending from the cecum to the descending colon can be manually extracted from colonic surfaces depicted at CT colonography.
  
• The TO can be used to orient a camera during CT colonographic navigation and thus facilitate comparison of endoluminal views between scans.
  
• The circumferential and centerline positions of a polyp on one scan can be used to create an area of high probability on the complementary scan where the polyp is most likely to be located.
  

	ACKNOWLEDGMENTS

 
We thank William Schindler, DO, for providing the CT colonography data and Andrew Dwyer, MD, for critical review of the manuscript.

	FOOTNOTES

 

Abbreviations: TL = tenia libera • TM = tenia mesocolica • TO = tenia omentalis • 3D = three-dimensional

² Current address: Department of Radiology, University of Wisconsin Medical School, Madison, Wis.

No Food and Drug Administration endorsement of any product or company mentioned in this article should be inferred.

See Materials and Methods for pertinent disclosures.

Author contributions: Guarantors of integrity of entire study, A.H., R.M.S.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, A.H., D.A.R., R.M.S.; clinical studies, J.R.C., P.J.P.; statistical analysis, A.H., N.P.; and manuscript editing, A.H., R.M.S., N.P., J.R.C., P.J.P.

	References

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

 

1. Johnson CD, Dachman AH. CT colonography: the next colon screening examination? Radiology 2000;216:331–341.[Abstract/Free Full Text]
2. Yee J, Kumar NN, Hung RK, Akerkar GA, Kumar PR, Wall SD. Comparison of supine and prone scanning separately and in combination at CT colonography. Radiology 2003;226:653–661.[Abstract/Free Full Text]
3. Chen SC, Lu DS, Hecht JR, Kadell BM. CT colonography: value of scanning in both the supine and prone positions. AJR Am J Roentgenol 1999;172:595–599.[Abstract/Free Full Text]
4. Fletcher JG, Johnson CD, Welch TJ, et al. Optimization of CT colonography technique: prospective trial in 180 patients. Radiology 2000;216:704–711.[Abstract/Free Full Text]
5. Laks S, Macari M, Bini EJ. Positional change in colon polyps at CT colonography. Radiology 2004;231:761–766.[Abstract/Free Full Text]
6. Iordanescu G, Summers RM. Automated centerline for computed tomography colonography. Acad Radiol 2003;10:1291–1301.[CrossRef][Medline]
7. Li P, Napel S, Acar B, Paik DS, Jeffrey RB Jr, Beaulieu CF. Registration of central paths and colonic polyps between supine and prone scans in computed tomography: pilot study. Med Phys 2004;31(10):2912–2923.[CrossRef][Medline]
8. Horton KM, Corl FM, Fishman EK. CT evaluation of the colon: inflammatory disease. RadioGraphics 2000;20:399–418.[Abstract/Free Full Text]
9. Standring S, ed. Gray's anatomy: the anatomical basis of clinical practice. 39th ed. Edinburgh, NY: Elsevier Churchill Livingstone, 2005; 1178.
10. Benedict KT. Anatomy and physiology of the colon. In: Dreyfuss JR, Janower ML, eds. Radiology of the colon. Baltimore, Md: Williams & Wilkins, 1980.
11. Rubesin SE, Levine MS, Laufer I, Herlinger H. Double-contrast barium enema examination technique. Radiology 2000;215:642–650.[Abstract/Free Full Text]
12. Pickhardt PJ, Choi JR, Hwang I, et al. Computed tomographic virtual colonoscopy to screen for colorectal neoplasia in asymptomatic adults. N Engl J Med 2003;349:2191–2200.[Abstract/Free Full Text]
13. Summers RM, Yao J, Pickhardt PJ, et al. Computed tomographic virtual colonoscopy computer-aided polyp detection in a screening population. Gastroenterology 2005;129:1832–1844.[CrossRef][Medline]
14. Summers RM, Franaszek M, Miller MT, Pickhardt PJ, Choi JR, Schindler WR. Computer-aided detection of polyps on oral contrast-enhanced CT colonography. AJR Am J Roentgenol 2005;184:105–108.[Free Full Text]
15. Huang A, Summers RM, Hara AK. Surface curvature estimation for automatic colonic polyp detection. In: Amini AA, Manduca A, eds. Proceedings of SPIE: medical imaging 2005—physiology, function and structure from medical images. Vol 5746. Bellingham, Wash: SPIE–The International Society for Optical Engineering, 2005; 393–402.
16. Huang A, Roy D, Franaszek M, Summers RM. Teniae coli guided navigation and registration for virtual colonoscopy. Proc IEEE Vis 2005;279–285.
17. Preston S. Teaching prediction intervals. Journal of Statistics Education 2000; 8(3). http://www.amstat.org/publications/jse/secure/v8n3/preston.cfm. Accessed May 10, 2006.

This Article Abstract Figures Only Full Text (PDF) AVI Movies Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Huang, A. Articles by Pickhardt, P. J. Search for Related Content PubMed PubMed Citation Articles by Huang, A. Articles by Pickhardt, P. J.