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) 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 HighWire Citing Articles via Google Scholar Google Scholar Articles by Bodily, K. D. Articles by Crownhart, B. S. Search for Related Content PubMed PubMed Citation Articles by Bodily, K. D. Articles by Crownhart, B. S. Related Collections Related Article

Crohn Disease: Mural Attenuation and Thickness at Contrast-enhanced CT Enterography—Correlation with Endoscopic and Histologic Findings of Inflammation¹

¹ From the Mayo Clinic College of Medicine, Rochester, Minn (K.D.B.), Department of Radiology (J.G.F., C.D.J., J.L.F., J.M.B., M.R.B., C.H.M.), Division of Gastroenterology, Department of Internal Medicine (C.A.S., W.J.S., E.V.L.), and Department of Health Sciences Research, Section of Biostatistics (W.S.H., B.S.C.), Mayo Clinic Rochester, 200 First Street SW, Rochester, MN 55905. Received July 9, 2004; revision requested September 14; revision received January 14, 2005; accepted February 16; final version accepted, March 16. Address correspondence to J.G.F. (e-mail: fletcher.joel{at}mayo.edu).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION References

 
Purpose: To determine retrospectively if quantitative measures of small-bowel mural attenuation and thickness at computed tomographic (CT) enterography correlate with endoscopic and histologic findings of small-bowel inflammation and to estimate the performance of these measures in predicting inflammatory Crohn disease.

Materials and Methods: The institutional review board approved this HIPAA-compliant retrospective study, which was conducted with patient informed consent. CT enterography data in 96 patients (31 male patients and 65 female patients) who underwent ileoscopy with or without biopsy were examined for CT signs of active Crohn disease. The most highly enhancing segment of terminal ileum and a normal-appearing ileal loop were identified. After it was confirmed that semiautomated software could accurately measure mural attenuation and thickness, the selected terminal ileal and normal-appearing (control) ileal loops were examined (20 automated measurements at each location) to quantify mural attenuation and wall thickness. Results were compared with endoscopy and histology reports by using logistic regression analysis and receiver operating characteristic curves.

Results: Quantitative measures of terminal ileal mural attenuation and wall thickness correlated significantly with active Crohn disease (P < .001). Small-bowel wall thickness was not a significant factor after attenuation was taken into account. A threshold attenuation value with a sensitivity of 90% (18 of 20) for definite Crohn disease (compared with a sensitivity of 80% [16 of 20] for radiologist assessment) was selected. In patients who underwent ileal biopsy, threshold attenuation had a sensitivity identical to that of ileoscopy (81% [26 of 32]; 95% confidence interval: 64%, 93%) in predicting histologic inflammation.

Conclusion: Quantitative measures of mural attenuation and wall thickness at CT enterography correlate highly with ileoscopic and histologic findings of inflammatory Crohn disease. Quantitative measures of mural attenuation are sensitive markers of small bowel inflammation.

© RSNA, 2006

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION References

 
Crohn disease is an inflammatory disease, primarily of the bowel, that affects more than a half million Americans annually and has been increasing in prevalence since the 1980s (1,2). The past decade has seen an explosion of therapies aimed at the altered immune response observed in patients with Crohn disease. Chief among these therapies are targeted monoclonal antibodies such as infliximab (directed against human tumor necrosis factor ) and natalizumab (directed against -4 integrin, a leukocyte adhesion molecule) (3–5). Given the expense and potential self-limited administration of these drugs (secondary to the development of autoantibodies or other complications) (6,7), as well as the ever-growing arsenal of other antiinflammatory drugs and probiotics, reproducible, objective measures of Crohn disease inflammatory activity are needed to justify their use and measure their effectiveness.

The clinical trials task force of the International Organization of Inflammatory Bowel Disease has recommended the use of the Crohn disease activity index, or CDAI, as an endpoint in therapeutic trials designed to assess predominantly inflammatory Crohn disease activity (8). The task force report summarizes the strengths and weaknesses of the CDAI and other Crohn activity indexes. Importantly, however, none of the reviewed indexes score inflammatory disease within the small bowel (as opposed to within the colon or the entire patient), despite the fact that the small bowel is the principal site of Crohn disease activity.

Mural stratification (ie, visualization of two or three layers of the bowel wall), mural hyperenhancement (ie, segmental regions of increased mural enhancement compared with enhancement in the adjacent bowel), and small-bowel wall thickening (ie, mural thickness greater than 3 mm) at contrast material–enhanced computed tomography (CT) have been described previously as well-known signs of Crohn disease (9–14). Mural stratification and mural hyperenhancement likely indicate that transmural fibrosis has not occurred and are thought to correlate with clinically active disease (10,13,14). However, few radiology-based studies have involved correlation of these specific CT findings with findings at endoscopic and histologic assessment, particularly on a large scale. Furthermore, correlation of quantitative mural attenuation values and bowel wall thickness with endoscopic and histologic reference standards may help establish objective criteria that identify small-bowel inflammation and develop methods that can reproducibly measure disease burden.

Magnetic resonance (MR) imaging and CT enterography are methods for imaging the small bowel that display distended small-bowel loops and permit exquisite visualization of the small-bowel wall (15–24). These methods generally consist of high-spatial-resolution contrast-enhanced CT or MR imaging performed after ingestion of a large volume of oral contrast material. The purpose of our study was to retrospectively determine if quantitative measures of small-bowel mural attenuation and thickness at CT enterography correlate with endoscopic and histologic findings of small-bowel inflammation and to estimate the performance of these measures in predicting inflammatory Crohn disease.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION References

 
Study Population and Design
Our institutional review board approved this retrospective study, which was conducted with data from institutional patient databases and archives and was performed in compliance with the Health Insurance Portability and Accountability Act and other institutional review board regulations. Patient inclusion criteria for this study were as follows: (a) Patients had given informed consent to allow the use of medical data for research purposes, (b) patients had undergone a clinically indicated outpatient contrast-enhanced CT enterographic examination performed between August 2001 and July 2003, and (c) patients had undergone colonoscopy with intubation of the terminal ileum (ileoscopy) within 30 days before or after the date of the CT enterographic examination. Patients were not required to have undergone a terminal ileal biopsy procedure along with this ileoscopic examination.

Three hundred fifty patients underwent CT enterography during the catchment period. Ninety-six of these patients underwent terminal ileoscopy within 30 days of the CT examination (median, 1.0 day; mean, 2.9 days; range, 0–28 days). Of the 96 patients, 31 were male and 65 were female. There was no statistically significant difference in age distribution and birth sex between the male (mean age, 44 years; median age, 42 years; age range, 15–77 years) and the female (mean age, 43 years; median age, 41 years; age range, 17–72 years) patients.

After we enrolled a cohort of patients who had undergone CT enterography and correlative ileoscopy, we validated existing software (Analyze 5.0; AnalyzeDirect, Lenexa, Kan) for use in measuring small-bowel mural attenuation in a series of eight bowel wall phantoms. Thereafter, we validated the same software tools for measuring small-bowel wall thickness by comparing automated measurement of small-bowel wall thickness to manual measurement. These software tools were used in each CT enterography data set as described to obtain quantitative measurements of terminal ileal and control ileal mural attenuation and wall thickness. Control ileal measurements were obtained in a normal-appearing ileal loop selected by a gastrointestinal radiologist (J.G.F.). These quantitative measurements were compared with results of a reference standard assessment of Crohn disease activity.

Spiral CT Imaging
CT enterography was performed in each patient in a similar manner. Each patient was given 10 mg of oral metoclopramide (Reglan; Pharmaceutical Associates, Greenville, SC) 75 minutes before the CT examination to increase gastric emptying and small-bowel peristalsis. Patients then drank four 450-mL aliquots of water (1800 mL total). The first aliquot was given immediately after the metoclopramide, with subsequent aliquots given 25, 50, and 65 minutes after ingestion of the peristaltic agent. CT was performed 75 minutes after metoclopramide ingestion. Patients were given 1 mg of intravenous glucagon (Eli Lilly, Indianapolis, Ind) immediately before scanning. Contrast-enhanced CT imaging was performed by using 270 mA, 120 kVp, and 150 mL of intravenous contrast material (iopamidol, Isovue 300; Bracco Diagnostics, Princeton, NJ) 70 seconds after the injection of the intravenous contrast material, which was administered intravenously at a rate of 3 mL/sec. Section thickness was 2.5 mm, and images were reconstructed every 1.5 mm. Scanning was performed by using a multi–detector row CT scanner (LightSpeed Plus or LightSpeed Ultra; GE Healthcare, Milwaukee, Wis).

Selection of Bowel Segments for Quantitative Analysis
A gastrointestinal radiologist with 5 years of subspecialized clinical expertise (J.G.F.) reviewed all examinations and was blinded to all clinical endoscopic and pathologic information, as well as to results of other imaging examinations, with the exception of the extent (in centimeters) of the terminal ileum visualized at ileoscopy. This distance (mean, 12 cm) was reported in 83 of 96 ileoscopy reports. When no distance was reported for the visualized terminal ileum (in 13 of 96 patients), the distance of 6 cm (one-half of the mean distance) was used. For each patient, this radiologist noted the presence or absence of mural hyperenhancement (segmental increased attenuation of the bowel wall compared with the attenuation of normal-appearing ileum), mural thickening (>3 mm), mural stratification (visualization of two or three layers within the bowel wall), increased mesenteric fat attenuation, and the comb sign (prominent vasa recta) (25). He further defined two bowel segments to be used in the quantitative analysis for each patient: a normal-appearing control ileal loop and the most highly enhancing segment of the terminal ileum. The selection of this latter segment was limited to the region visualized at ileoscopy.

Validation of Attenuation Measurements
To validate semiautomated measurements of mural attenuation in the normal bowel wall (see Semiautomated Quantitative Image Evaluation), we constructed a set of phantoms composed of three materials: polystyrene (to simulate paraenteric fat with a CT number of approximately –50 HU), "solid liver" (Gammex-RMI, Middleton, Wis) (to simulate normal bowel wall attenuation of approximately 80 HU), and water (representing the gut lumen with a CT number of approximately 0 HU) (Fig 1). A second set of phantoms was constructed to simulate the hyperenhancing bowel wall. In this set of phantoms, acrylic (with a CT number of approximately 120 HU) was substituted for the material that mimicked solid liver. The phantom bowel walls of the solid liver material and acrylic were precisely cut into 1-, 2-, 3-, and 4-mm wafers, placed adjacent to a polystyrene cube, and submerged in a 30-cm-diameter water bath that mimicked the body cavity (Fig 1), thus resulting in one set of four phantoms that simulated normal bowel wall and a corresponding set that simulated hyperenhancing diseased bowel. This arrangement was constructed to mimic the interposition of the bowel wall between paraenteric fat and intraluminal fluid. These eight phantoms were scanned by using the CT parameters described earlier with a 50-cm field of view, a 2.5-mm section thickness, and a 1.25-mm reconstruction interval (Fig 1).

View larger version (139K): [in this window] [in a new window] [Download PPT slide]   Figure 1a: Small-bowel phantom simulating normally enhancing bowel that was used to validate automated attenuation measurements. (a) Transverse CT image shows 2-mm wafer of "solid liver" (white arrow) that simulates the attenuation of normally enhancing bowel. This wafer was placed between a polystyrene block with the attenuation of perienteric fat (black arrow) and surrounding water, which mimicked the attenuation of intraluminal fluid. (b) Enlarged transverse CT image shows how line tool (thick white bar) was placed perpendicularly across the solid liver wafer. The line tool acquired maximal attenuation measurements at 1-mm increments in the direction of the black arrow.

View larger version (139K): [in this window] [in a new window] [Download PPT slide]   Figure 1b: Small-bowel phantom simulating normally enhancing bowel that was used to validate automated attenuation measurements. (a) Transverse CT image shows 2-mm wafer of "solid liver" (white arrow) that simulates the attenuation of normally enhancing bowel. This wafer was placed between a polystyrene block with the attenuation of perienteric fat (black arrow) and surrounding water, which mimicked the attenuation of intraluminal fluid. (b) Enlarged transverse CT image shows how line tool (thick white bar) was placed perpendicularly across the solid liver wafer. The line tool acquired maximal attenuation measurements at 1-mm increments in the direction of the black arrow.

The mean measured CT number of a block of "solid liver" was 82 HU ± 29 (standard deviation). The absolute errors in the semiautomated CT number measurements were +17.6 HU, +8.8 HU, +2.0 HU, and –34.2 HU for wafer thicknesses of 4, 3, 2, and 1 mm, respectively. The high absolute error for the 1-mm phantom bowel wall was not unanticipated because the in-plane resolution approached 1 mm, while the other absolute errors were well within one standard deviation of the known true mean value. The mean known attenuation CT number for the acrylic was 120 HU ± 31. The absolute errors in attenuation with the automated method were +13.9 HU, +8.9 HU, –8.6 HU, and –46.9 HU for wafer thicknesses of 4, 3, 2, and 1 mm, respectively, demonstrating the same pattern observed by using the solid liver phantom. On the basis of the mean wall thickness for the normal-appearing ileum in our study population (reported below), it was concluded that CT number measurements obtained by using our semiautomated technique would be accurate for the human small-bowel wall.

Validation of Wall Thickness Measurements
As a quality-control measure to validate our wall thickness measurement techniques, the caliper tool in Analyze 5.0 was used to manually measure the transmural thickness of the selected terminal ileal and control ileal segments (K.D.B.). These results were compared with automated measurements of bowel wall thickness by using the perpendicular line tool or the radial line tool from the Analyze 5.0 software package (see Semiautomated Quantitative Image Evaluation below), which automatically calculated bowel wall thickness by using a full width at half maximum technique and creating a histogram of the CT numbers across the bowel wall from the intraluminal fluid to the paraenteric fat (Fig 2). Instances in which the bowel wall could not be measured by using either method were counted. The manually measured bowel wall thicknesses were compared with those measured by using the semiautomated technique.

View larger version (112K): [in this window] [in a new window] [Download PPT slide]   Figure 2a: Semiautomated measurement of terminal ileal attenuation. (a) Transverse CT image of inflamed ileal loop shows that the gut lumen is transected parallel to the transverse cut plane. The green box indicates the portion of the image that is cropped to isolate the bowel segment to be analyzed. (b) Within the cropped image, the line tool (red line) was placed over the bowel wall to obtain measurements of mural attenuation and wall thickness every 1 mm along a 3.0-cm bowel segment (along the blue line, in the direction of the open arrow). (c) Graph corresponding to ileal measurements in b. The numbers on the y-axis are Hounsfield units; those on the x-axis represent distance in millimeters. The red curve plots CT attenuation versus distance, with values corresponding to the red line traversing the bowel wall in b. Graph shows maximum mural attenuation (arrow), the automated mural thickness measurement obtained by using a full width at half maximum technique (H-shaped measurement), and attenuation values corresponding to luminal fluid (large arrowhead) and perienteric fat (small arrowheads). (d) Transverse CT image demonstrates analysis of another ileal segment in which the gut lumen is transected perpendicular to the transverse cut plane, with the line tool (green and red line) placed in the center of the bowel lumen to obtain measurements of mural attenuation and wall thickness at 15° increments rotating around the central axis of the bowel. (e) Graph corresponding to ileum in d shows maximum mural attenuation (arrow), the automated mural thickness measurement obtained by using a full width at half maximum technique (H-shaped measurement), and attenuation values corresponding to luminal fluid (large arrowhead) and perienteric fat (small arrowheads).

View larger version (132K): [in this window] [in a new window] [Download PPT slide]   Figure 2b: Semiautomated measurement of terminal ileal attenuation. (a) Transverse CT image of inflamed ileal loop shows that the gut lumen is transected parallel to the transverse cut plane. The green box indicates the portion of the image that is cropped to isolate the bowel segment to be analyzed. (b) Within the cropped image, the line tool (red line) was placed over the bowel wall to obtain measurements of mural attenuation and wall thickness every 1 mm along a 3.0-cm bowel segment (along the blue line, in the direction of the open arrow). (c) Graph corresponding to ileal measurements in b. The numbers on the y-axis are Hounsfield units; those on the x-axis represent distance in millimeters. The red curve plots CT attenuation versus distance, with values corresponding to the red line traversing the bowel wall in b. Graph shows maximum mural attenuation (arrow), the automated mural thickness measurement obtained by using a full width at half maximum technique (H-shaped measurement), and attenuation values corresponding to luminal fluid (large arrowhead) and perienteric fat (small arrowheads). (d) Transverse CT image demonstrates analysis of another ileal segment in which the gut lumen is transected perpendicular to the transverse cut plane, with the line tool (green and red line) placed in the center of the bowel lumen to obtain measurements of mural attenuation and wall thickness at 15° increments rotating around the central axis of the bowel. (e) Graph corresponding to ileum in d shows maximum mural attenuation (arrow), the automated mural thickness measurement obtained by using a full width at half maximum technique (H-shaped measurement), and attenuation values corresponding to luminal fluid (large arrowhead) and perienteric fat (small arrowheads).

View larger version (19K): [in this window] [in a new window] [Download PPT slide]   Figure 2c: Semiautomated measurement of terminal ileal attenuation. (a) Transverse CT image of inflamed ileal loop shows that the gut lumen is transected parallel to the transverse cut plane. The green box indicates the portion of the image that is cropped to isolate the bowel segment to be analyzed. (b) Within the cropped image, the line tool (red line) was placed over the bowel wall to obtain measurements of mural attenuation and wall thickness every 1 mm along a 3.0-cm bowel segment (along the blue line, in the direction of the open arrow). (c) Graph corresponding to ileal measurements in b. The numbers on the y-axis are Hounsfield units; those on the x-axis represent distance in millimeters. The red curve plots CT attenuation versus distance, with values corresponding to the red line traversing the bowel wall in b. Graph shows maximum mural attenuation (arrow), the automated mural thickness measurement obtained by using a full width at half maximum technique (H-shaped measurement), and attenuation values corresponding to luminal fluid (large arrowhead) and perienteric fat (small arrowheads). (d) Transverse CT image demonstrates analysis of another ileal segment in which the gut lumen is transected perpendicular to the transverse cut plane, with the line tool (green and red line) placed in the center of the bowel lumen to obtain measurements of mural attenuation and wall thickness at 15° increments rotating around the central axis of the bowel. (e) Graph corresponding to ileum in d shows maximum mural attenuation (arrow), the automated mural thickness measurement obtained by using a full width at half maximum technique (H-shaped measurement), and attenuation values corresponding to luminal fluid (large arrowhead) and perienteric fat (small arrowheads).

View larger version (96K): [in this window] [in a new window] [Download PPT slide]   Figure 2d: Semiautomated measurement of terminal ileal attenuation. (a) Transverse CT image of inflamed ileal loop shows that the gut lumen is transected parallel to the transverse cut plane. The green box indicates the portion of the image that is cropped to isolate the bowel segment to be analyzed. (b) Within the cropped image, the line tool (red line) was placed over the bowel wall to obtain measurements of mural attenuation and wall thickness every 1 mm along a 3.0-cm bowel segment (along the blue line, in the direction of the open arrow). (c) Graph corresponding to ileal measurements in b. The numbers on the y-axis are Hounsfield units; those on the x-axis represent distance in millimeters. The red curve plots CT attenuation versus distance, with values corresponding to the red line traversing the bowel wall in b. Graph shows maximum mural attenuation (arrow), the automated mural thickness measurement obtained by using a full width at half maximum technique (H-shaped measurement), and attenuation values corresponding to luminal fluid (large arrowhead) and perienteric fat (small arrowheads). (d) Transverse CT image demonstrates analysis of another ileal segment in which the gut lumen is transected perpendicular to the transverse cut plane, with the line tool (green and red line) placed in the center of the bowel lumen to obtain measurements of mural attenuation and wall thickness at 15° increments rotating around the central axis of the bowel. (e) Graph corresponding to ileum in d shows maximum mural attenuation (arrow), the automated mural thickness measurement obtained by using a full width at half maximum technique (H-shaped measurement), and attenuation values corresponding to luminal fluid (large arrowhead) and perienteric fat (small arrowheads).

View larger version (19K): [in this window] [in a new window] [Download PPT slide]   Figure 2e: Semiautomated measurement of terminal ileal attenuation. (a) Transverse CT image of inflamed ileal loop shows that the gut lumen is transected parallel to the transverse cut plane. The green box indicates the portion of the image that is cropped to isolate the bowel segment to be analyzed. (b) Within the cropped image, the line tool (red line) was placed over the bowel wall to obtain measurements of mural attenuation and wall thickness every 1 mm along a 3.0-cm bowel segment (along the blue line, in the direction of the open arrow). (c) Graph corresponding to ileal measurements in b. The numbers on the y-axis are Hounsfield units; those on the x-axis represent distance in millimeters. The red curve plots CT attenuation versus distance, with values corresponding to the red line traversing the bowel wall in b. Graph shows maximum mural attenuation (arrow), the automated mural thickness measurement obtained by using a full width at half maximum technique (H-shaped measurement), and attenuation values corresponding to luminal fluid (large arrowhead) and perienteric fat (small arrowheads). (d) Transverse CT image demonstrates analysis of another ileal segment in which the gut lumen is transected perpendicular to the transverse cut plane, with the line tool (green and red line) placed in the center of the bowel lumen to obtain measurements of mural attenuation and wall thickness at 15° increments rotating around the central axis of the bowel. (e) Graph corresponding to ileum in d shows maximum mural attenuation (arrow), the automated mural thickness measurement obtained by using a full width at half maximum technique (H-shaped measurement), and attenuation values corresponding to luminal fluid (large arrowhead) and perienteric fat (small arrowheads).

Images for all selected bowel segments were digitally analyzed by using the line profile tool of the Analyze 5.0 software package to quantify average maximal bowel wall attenuation and average transmural thickness. Analysis of each image followed the same protocol, in which images were cropped to isolate the selected bowel loop or phantom to minimize file size and optimize zoom enlargement capability (Fig 2). Measurements were obtained in the transverse plane to maximize spatial resolution (K.D.B., J.G.F.).

For images that displayed bowel segments in which the bowel segment was transected parallel to the cut plane, the perpendicular line tool was placed over the bowel wall for a distance of 3.0 cm (Fig 2). If asymmetric enhancement predominantly affected one side of the bowel wall, the line tool was placed over the most highly enhancing part of the bowel wall. For bowel segments cut in cross section by the transverse plane, the radial line tool was employed (Fig 2).

For each image, the first 20 data points for maximal mural attenuation and thickness (as measured with the full width at half maximum technique) were saved; additional data points, if any, were discarded. In all cases, analysis of 30 segments spaced at 1-mm intervals or 24 segments spaced at 15° intervals was sufficient to collect 20 data points for mural attenuation (Fig 2). For each patient, mean values and standard deviations were calculated from the 20 data points that represented maximal transmural attenuation and wall thickness for both the terminal ileum and control segments.

Reference Standards for Small-Bowel Inflammation and Crohn Disease Activity
A gastroenterologist (C.A.S.) who was blinded to the results of CT enterography reviewed all ileocolonoscopy and histology reports for the study patients. The endoscopic findings in both the terminal ileum and anastomosis (if a patient had previously undergone surgery) were recorded. The anastomosis was defined as part of the terminal ileum for the purposes of comparison with CT enterography results.

Because some endoscopic and histologic findings are nonspecific for Crohn disease, and because these findings are evaluated together in conjunction with clinical presentation and results of patient examination in rendering a final diagnosis of Crohn disease, a panel of three gastroenterologists (C.A.S., E.V.L., and W.J.S.) who had 2, 9, and 11 years of subspecialty clinical experience in our inflammatory bowel disease clinic, respectively, was consulted. They constructed four reference standards against which CT enterography should be compared: (a) ileoscopy, (b) histologic examination, (c) a global assessment indicating probable active Crohn disease, and (d) a global assessment indicating definite active inflammatory Crohn disease. Ileoscopic findings were categorized as normal or abnormal. All abnormal lesions, such as erosions and/or ulcerations, stenosis, granularity, friability, erythema, and edema were recorded. In cases in which the terminal ileum was sampled for biopsy, histologic results were also classified as normal or abnormal. The following abnormal histologic categories were specifically recorded: acute ileitis, chronic mildly active ileitis, chronic moderate to severely active ileitis, and "other" lesions. No clinical history information, including the use of nonsteroidal antiinflammatory drugs, was available to the reviewer.

A provisional global assessment of Crohn disease (no Crohn disease, probable Crohn disease, definite Crohn disease) was created for each patient on the basis of the following criteria created by the gastroenterologist panel and the ileoscopy and biopsy findings. If biopsy was not performed, the diagnosis was based on the endoscopic findings alone. The presence of Crohn disease was considered probable if (a) ileoscopy revealed stenosis, ulcerations, granularity, or friability (but not erythema or mucosal edema alone) combined with normal or no biopsy findings, (b) ileoscopic results were normal but acute or chronic ileitis was observed at biopsy, or (c) ileoscopy showed erythema, granularity, friability, or edema and biopsy revealed acute ileitis. The presence of Crohn disease was considered definite if ileoscopy revealed erosions and/or ulcerations or stenosis or if abnormal ileoscopic results were combined with the finding of chronic ileitis at biopsy.

Statistical Analysis
Radiologist interpretations of specific CT enterography findings (mural hyperenhancement, mural thickening, mural stratification, increased attenuation of mesenteric fat, and the comb sign) were compared with the reference standards of Crohn disease activity. These reference standards, against which visual interpretation and quantitative measurements were compared, were (a) ileoscopy, (b) histologic examination (in patients in whom biopsy was performed), (c) a global assessment of definite active Crohn disease, and (d) a global assessment of probable or definite active Crohn disease. Sensitivity and specificity estimates for each of the CT enterography findings in correctly identifying each of the four reference-standard determinations were calculated, along with 95% exact binomial confidence intervals (CIs). Statistical software (SAS, version 8.2; SAS, Cary, NC) was employed for all statistical analyses.

Univariate logistic regression models, in which the binary dependent variable was a reference-standard determination of disease, were used to assess the ability of each quantitative CT enterography parameter (terminal ileal attenuation, terminal ileal loop attenuation compared with control ileal loop attenuation [in terms of both their ratio and additive effect], terminal ileal thickness alone, and terminal ileal thickness compared with control ileal loop thickness [in terms of both their ratio and additive effect]) to discriminate between patients with disease and patients without disease. Odds ratios (and 95% CIs) based on the logistic regression model estimates were computed. Corresponding multiple logistic regression models in which thickness and attenuation measurements of the terminal ileum and control ileal loop were considered were also evaluated. Receiver operating characteristic (ROC) curves were constructed to summarize the discriminatory ability of the logistic regression models fitted, and area under the ROC curve (A_z) estimates are reported.

Differences in measured attenuation and thickness and the three levels of Crohn disease diagnosis (normal, probable, and definite) for both the terminal ileum and the normal bowel were assessed by using a one-way analysis of variance. When there was a significant difference among the three diagnostic levels, pairwise comparisons were performed by using two-sample t tests. P values for these pairwise tests were adjusted by using the Bonferroni adjustment for multiple comparisons. Similarly, thickness was assessed among the three levels of Crohn disease diagnosis. The level for statistical significance was set at .05.

A terminal ileal attenuation cutpoint for categorizing a patient as having disease at CT enterography was chosen by examining the attenuation in the previously selected terminal ileal segment. With this criterion, sensitivity and specificity were again estimated.

In patients in whom biopsy was performed, we considered the histologic result to be the reference standard for active disease and estimated the sensitivity and specificity of ileoscopy for correctly identifying patients with disease.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION References

 
Of the 96 patients, all underwent terminal ileoscopy; biopsy was performed in 69 patients. According to reference-standard results, Crohn disease was definitely active in 20 patients, probably active in 15 patients, and absent in 61 patients.

Before proceeding with semiautomated analysis, we compared manual and semiautomated techniques for measuring small-bowel wall thickness, as described. In some cases, the Analyze 5.0 full width at half maximum algorithm failed to identify the bowel wall thickness owing to bowel wall collapse or artifacts from adjacent tissue structures (which modified the histogram profile), resulting in a missed segment that produced no data. The automated full width at half maximum technique failed to measure bowel wall thickness in the terminal ileum in 16 (17%) of 96 patients, but there were only two (2%) patients in whom we could not manually measure terminal ileal thickness. Such measurement failures in the normal-appearing ileum occurred in two patients and one patient, respectively. The average absolute difference between the manual and the automated methods for measuring bowel wall thickness in each patient was only 0.6 mm in the terminal ileum and only 0.4 mm in the control ileum—both of which differences were smaller than one pixel (approximately 0.6–0.9 mm, depending on the field of view). The mean wall thickness of the normal ileum was 2.0 mm (range, 1.4–3.1 mm) according to manual measurements and 2.3 mm (range, 1.1–3.5 mm) according to the automated full width at half maximum technique. We subsequently concluded that the semiautomated bowel wall measurement technique described enabled accurate measurements of bowel wall thickness.

Visual Prediction of Disease
Regarding the performance of specific CT findings in predicting the presence of probable or active Crohn disease as compared with reference-standard assessment, mural hyperenhancement had the highest sensitivity for predicting the presence of active inflammatory disease, with a sensitivity of 80% (16 of 20 patients; 95% CI: 56%, 94%) for patients with definite active Crohn disease and a sensitivity of 69% (24 of 35 patients; 95% CI: 51%, 83%) for patients with probable or definite active Crohn disease (Table 1).

Multivariate models of terminal ileal and control ileal mural attenuation and wall thickness were constructed by examining significant univariate findings (Table 2). Importantly, after terminal ileal attenuation was taken into account, all multivariate models failed to demonstrate that terminal ileal thickness was a significant independent variable.

ROC Analysis
The A_z values obtained with ROC analysis of continuous ileal attenuation and wall thickness data showed that absolute terminal ileal attenuation predicted the presence of disease at reference-standard assessments with A_z values that ranged from 0.79 to 0.81 and compared favorably with other measures of terminal ileal attenuation that took the control loop into account (Table 4). A_z values reflecting terminal ileal thickness performed similarly in identifying patients with active disease but less favorably in identifying patients with abnormal biopsy results.

View larger version (21K): [in this window] [in a new window] [Download PPT slide]   Figure 3: ROC curves for the performance of absolute terminal ileal attenuation (dotted line) and the ratio of terminal ileal attenuation to control ileal attenuation (dashed line) in predicting abnormal ileal biopsy results. Note the similar performance of both quantitative measures.

View larger version (28K): [in this window] [in a new window] [Download PPT slide]   Figure 4: Scatterplot of terminal ileal enhancement versus normal ileal enhancement (in Hounsfield units) in 96 patients who underwent CT enterography. Although there was a wide range of attenuation values in the terminal ileum, all but two patients had attenuation values of less than 109 HU in the normal-appearing control ileal loop.

View larger version (156K): [in this window] [in a new window] [Download PPT slide]   Figure 5a: Transverse CT enterographic images in 34-year-old man with abnormal ileal enhancement and no endoscopic or histologic evidence of ileal inflammation in whom abnormal bowel enhancement possibly arose from abnormalities in mesenteric venous circulation show (a) an enhancing terminal ileum (arrow) with attenuation of 118 HU, (b) a chronic portal vein clot (arrows) with cavernous transformation of the portal vein, (c) intraluminal esophageal varices (arrow) that indicated portal hypertension, and (d) evidence of previous small bowel resection, with mural hyperenhancement (arrows) and wall thickening consistent with active recurrent Crohn disease near the anastomosis. The mild dilatation of proximal small bowel (arrowhead) in d indicates partial small-bowel obstruction.

View larger version (152K): [in this window] [in a new window] [Download PPT slide]   Figure 5b: Transverse CT enterographic images in 34-year-old man with abnormal ileal enhancement and no endoscopic or histologic evidence of ileal inflammation in whom abnormal bowel enhancement possibly arose from abnormalities in mesenteric venous circulation show (a) an enhancing terminal ileum (arrow) with attenuation of 118 HU, (b) a chronic portal vein clot (arrows) with cavernous transformation of the portal vein, (c) intraluminal esophageal varices (arrow) that indicated portal hypertension, and (d) evidence of previous small bowel resection, with mural hyperenhancement (arrows) and wall thickening consistent with active recurrent Crohn disease near the anastomosis. The mild dilatation of proximal small bowel (arrowhead) in d indicates partial small-bowel obstruction.

View larger version (77K): [in this window] [in a new window] [Download PPT slide]   Figure 5c: Transverse CT enterographic images in 34-year-old man with abnormal ileal enhancement and no endoscopic or histologic evidence of ileal inflammation in whom abnormal bowel enhancement possibly arose from abnormalities in mesenteric venous circulation show (a) an enhancing terminal ileum (arrow) with attenuation of 118 HU, (b) a chronic portal vein clot (arrows) with cavernous transformation of the portal vein, (c) intraluminal esophageal varices (arrow) that indicated portal hypertension, and (d) evidence of previous small bowel resection, with mural hyperenhancement (arrows) and wall thickening consistent with active recurrent Crohn disease near the anastomosis. The mild dilatation of proximal small bowel (arrowhead) in d indicates partial small-bowel obstruction.

View larger version (107K): [in this window] [in a new window] [Download PPT slide]   Figure 5d: Transverse CT enterographic images in 34-year-old man with abnormal ileal enhancement and no endoscopic or histologic evidence of ileal inflammation in whom abnormal bowel enhancement possibly arose from abnormalities in mesenteric venous circulation show (a) an enhancing terminal ileum (arrow) with attenuation of 118 HU, (b) a chronic portal vein clot (arrows) with cavernous transformation of the portal vein, (c) intraluminal esophageal varices (arrow) that indicated portal hypertension, and (d) evidence of previous small bowel resection, with mural hyperenhancement (arrows) and wall thickening consistent with active recurrent Crohn disease near the anastomosis. The mild dilatation of proximal small bowel (arrowhead) in d indicates partial small-bowel obstruction.

View larger version (96K): [in this window] [in a new window] [Download PPT slide]   Figure 6: Transverse CT images in 28-year-old woman with endoscopically normal-appearing terminal ileum (for a distance of 15 cm) and normal ileal mucosa at random ileal biopsy show terminal ileal mural hyperenhancement and stratification (large arrow) and an ileocolic fistula (small arrows).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION References

Importantly, in multivariate models of enhancement, after accounting for mural attenuation, wall thickness was not significant. Similarly, results of visual assessment of the enterography data sets by a gastrointestinal radiologist also revealed that mural hyperenhancement was the most sensitive CT finding for inflammatory disease. These findings may have arisen from collapse mimicking disease in healthy patients or spasm precluding accurate thickness measurements in patients with disease. Although our software could always measure maximal ileal attenuation, attempts at automated measurement of terminal ileal wall thickness failed in 17% of patients. However, it should be noted that mural thickness performed comparably to mural attenuation at both visual and quantitative assessment and is a highly sensitive and specific finding for active Crohn disease. Our observations are similar, however, to those noted with MR imaging (21)—observations that indicate that Crohn disease severity is more accurately determined with assessment of mural enhancement after gadolinium chelate administration than with assessment of bowel wall thickness and signal intensity.

After we identified mural attenuation as the critical variable in predicting active disease, ROC analyses were performed to assess the performance of quantitative enhancement values. When we compared absolute terminal ileal enhancement and the ratio of terminal ileal enhancement to control ileal enhancement, the A_z values in these analyses were similar across the four reference standard assessments and ranged from 0.78 to 0.84. On the basis of this finding, we estimated the potential performance of terminal ileal enhancement at CT enterography in identifying patients with Crohn disease. We selected a threshold terminal ileal enhancement value for active disease on the basis of the observation that nearly all control ileal loops had enhancement values that were less than this value. With use of this threshold for absolute terminal ileal enhancement, CT enterography had a sensitivity of 90% for identifying patients with definite active Crohn disease and 80% for identifying patients with probable or definite Crohn disease (a similar value to that for ileoscopy). This performance slightly exceeded (by about 10%) that of an experienced gastrointestinal radiologist in predicting active Crohn disease on the basis of mural hyperenhancement. In this regard, quantitative measures of mural enhancement may help reduce interobserver variability by objectively defining disease presence.

The exact threshold value for defining abnormal mural attenuation will vary depending on the specifics of the contrast material injection protocol (ie, iodine concentration and amount, speed of injection, and delay before scanning), the degree of bowel distention, and the method of measurement (19,26). Furthermore, idiosyncratic factors such as cardiac output likely affect the absolute enhancement of bowel in any patient, even if they do not affect relative attenuation compared with the attenuation of normal-appearing bowel loops. In the future, because of the considerations described, software that uses quantitative measures to identify or quantify active Crohn disease burden will probably identify segmental regions of increased small-bowel attenuation relative to normal-appearing bowel as a control rather than rely on an absolute threshold value, which may be prone to variability.

Quantitative measures of ileal attenuation were less specific than visual estimates for identifying active Crohn disease (69% vs 82%, respectively). Radiologist visual evaluation of enterography data sets to exclude a variety of potential causes of increased mural attenuation that do not represent active Crohn disease will likely always be required (Fig 5). These causes may include collapse, superior mesenteric vein thrombosis, short gut, or backwash ileitis but remain to be entirely elucidated.

The ability of quantitative data from three-dimensional CT enterography data sets to accurately predict and measure small-bowel inflammation has several important implications. First, reproducible and accurate descriptions of Crohn disease presence and burden within the small bowel are needed (12). The de facto index describing inflammatory activity in Crohn disease is now the Crohn disease activity index, but variability exists in physician observers' assessments of case histories and patients' reports of subjective well being and abdominal pain (8). The use of CT enterography enables localization of disease activity to the small bowel. Although capsule endoscopy is very sensitive for detection of early Crohn disease (27,28), capsule endoscopy first requires the performance of another (probably radiologic) test to exclude obstruction before capsule ingestion because preliminary results obtained by using a nonradiologic patency capsule have been disappointing (29). Capsule endoscopy also does not depict extraenteric complications of Crohn disease such as fistula, phlegmon, and abscess and cannot be used to estimate the length of diseased bowel segments. Preliminary evidence (30) indicates that, because of the relative strengths of each test, visualization of the bowel wall at CT enterography likely permits identification of small-bowel Crohn disease that is on a par with the identification enabled by capsule endoscopy.

Quantitative interrogation of three-dimensional CT enterography data sets will likely allow reproducible estimates of disease length and severity on the basis of abnormal mural enhancement and other radiologic findings. Although CT enterography already appears to have distinct advantages over other tests (ie, it is more sensitive than small-bowel follow-through examination and less invasive than double-contrast enteroclysis and is able to depict active inflammation proximal to the reach of the endoscope), one unrealized potential advantage of CT enterography is that the three-dimensional nature of the data sets enables quantitative interrogation, which could be used to measure disease burden or severity.

Although the algorithms for determining disease presence on the basis of enhancement values in this report will necessarily change owing to differences in scanning protocols, particularly as such differences relate to the injection of intravenous contrast material, we believe that our findings support the idea that semiautomated interrogation of mural hyperenhancement may lead to highly reproducible algorithms for defining the presence and amount of inflammatory small-bowel activity. Similar techniques could likely be extended to MR enterography, where the signal intensity differences indicating small-bowel inflammation may be even greater.

Our study had several limitations. We relied solely on endoscopic (and often histologic) assessment in defining the presence of active Crohn disease. This method created bias against enterography: When the ileum appeared normal endoscopically, biopsy was either not performed or was not directed toward focal Crohn findings. Any radiologic studies that demonstrated mural hyperenhancement in these cases, no matter how convincing radiologically, were rated as false-positive studies, lowering the reported specificity. Furthermore, our study design relied on the accurate correlation of ileoscopic and radiologic findings. False-positive studies could arise from mural hyperenhancement that occurs in an ileal segment thought to have been viewed at correlative ileoscopy but actually located proximal to the point of ileoscopic assessment. Our algorithm for measuring maximum mean mural attenuation depended on a radiologist selecting the worst-appearing segment of the terminal ileum, potentially making our method prone to interobserver variability. In the future, it may be useful for a radiologist to interactively identify "normal enhancement" within a patient by using software to highlight areas of potential inflammatory activity that should be closely examined.

Finally, we performed CT enterography in the hepatic phase and did not vary the dose of intravenous contrast material between patients of different sizes. Although a late arterial phase or "enteric phase" may be the temporal window that displays maximum enteric enhancement (26), it remains unclear whether the enteric phase is the best temporal window in which to maximize the conspicuity of (or attenuation difference between) normal and inflamed bowel. In this study, we chose to scan in the hepatic phase on the basis of our earlier observation that there was no clear benefit to scanning in the enteric phase, but this decision was based on observations in a small number of patients (24). We do believe that the detection of mural hyperenhancement is compromised in very large patients and attempted to overcome such idiosyncratic variables affecting enhancement by taking the enhancement of the normal ileal loop into account. We now vary the dose of intravenous contrast material, as well as the radiation dose for CT enterography, according to patient size. Additional studies will be needed to determine the interobserver reproducibility of specific CT findings and quantitative measurements, as well as the interrelationship between visual and quantitative measurements.

In summary, mural attenuation and wall thickness as identified at contrast-enhanced CT enterography are sensitive markers for Crohn disease inflammatory activity. Semiautomated quantitative measurement of these variables is possible with existing software, and small-bowel mural attenuation is highly correlated with disease activity. Although CT enterography will likely continue to evolve as a small-bowel imaging technique, quantitative measurement of mural attenuation and wall thickness may enable objective measures of disease presence and severity to be standardized and defined in a highly reproducible fashion.

	FOOTNOTES

 

Abbreviations: A_z = area under ROC curve • CI = confidence interval • ROC = receiver operating characteristic

Authors stated no financial relationship to disclose.

See also Science to Practice in this issue.

Author contributions: Guarantors of integrity of entire study, K.D.B., J.G.F.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, K.D.B., J.G.F., C.D.J., J.L.F.; clinical studies, K.D.B., J.G.F., C.A.S., C.D.J., W.J.S., E.V.L.; experimental studies, K.D.B., J.G.F., M.R.B., C.H.M.; statistical analysis, K.D.B., J.G.F., W.S.H., B.S.C.; and manuscript editing, K.D.B., J.G.F., C.A.S., C.D.J., J.L.F., J.M.B., C.H.M. W.J.S., W.S.H.

	References

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION References

 

1. Loftus EV Jr, Silverstein MD, Sandborn WJ, Tremaine WJ, Harmsen WS, Zinsmeister AR. Crohn's disease in Olmsted County, Minnesota, 1940–1993: incidence, prevalence, and survival. Gastroenterology 1998;114:1161–1168. [Published correction appears in Gastroenterology 1999;116:1507.]
2. Loftus EV Jr. Clinical epidemiology of inflammatory bowel disease: incidence prevalence, and environmental in influences. Gastroenterology 2004;126:1504–1517.[CrossRef][Medline]
3. Present DH, Rutgeerts P, Targan S, et al. Infliximab for the treatment of fistulas in patients with Crohn's disease. N Engl J Med 1999;340:1398–1405.[Abstract/Free Full Text]
4. Targan SR, Hanauer SB, van Deventer SJ, et al. A short-term study of chimeric monoclonal antibody cA2 to tumor necrosis factor alpha for Crohn's disease. N Engl J Med 1997;337:1029–1035.[Abstract/Free Full Text]
5. Ghosh S, Goldin E, Gordon FH, et al. Natalizumab for active Crohn's disease. N Engl J Med 2003;348:24–32.[Abstract/Free Full Text]
6. Colombel JF, Loftus EV Jr, Tremaine WJ, et al. The safety profile of infliximab in patients with Crohn's disease: the Mayo Clinic experience in 500 patients. Gastroenterology 2004;126:19–31.[CrossRef][Medline]
7. Sandborn WJ, Loftus EV. Balancing the risks and benefits of infliximab in the treatment of inflammatory bowel disease. Gut 2004;53:780–782.[Free Full Text]
8. Sandborn WJ, Feagan BG, Hanauer S, et al. A review of activity indices and efficacy endpoints for clinical trials of medical therapy in adults with Crohn's disease. Gastroenterology 2002;122:512–530.[Medline]
9. Gore RM, Marn CS, Kirby DF, Vogelzang RL, Neiman HL. CT findings in ulcerative, granulomatous and indeterminate colitis. AJR Am J Roentgenol 1984;143:279–284.[Abstract/Free Full Text]
10. Gore RM, Balthazar EJ, Ghahremani GG, Miller FH. CT features of ulcerative colitis and Crohn's disease. AJR Am J Roentgenol 1996;167:3–15.[Free Full Text]
11. Wills JS, Lobis IF, Denstman FJ. Crohn disease: state of the art. Radiology 1997;202:597–610.[Free Full Text]
12. Maglinte DD, Gourtsoyiannis N, Rex D, Howard TJ, Kelvin FM. Classification of small bowel Crohn's subtypes based on multimodality imaging. Radiol Clin North Am 2003;41:285–303.[CrossRef][Medline]
13. Del Campo L, Arribas I, Valbuena M, Mate J, Moreno-Otero R. Spiral CT findings in active and remission phases in patients with Crohn's disease. J Comput Assist Tomogr 2001;25:792–797.[CrossRef][Medline]
14. Choi D, Lee S, Cho Y, et al. Bowel wall thickening in patients with Crohn's disease: CT patterns and correlation with inflammatory activity. Clin Radiol 2003;58:68–74.[CrossRef][Medline]
15. Gourtsoyiannis N, Papanikolaou N, Grammatikakis J, Maris T, Prassopoulos P. MR enteroclysis protocol optimization: comparison between 3D FLASH with fat saturation after intravenous gadolinium injection and true FISP sequences. Eur Radiol 2001;11:908–913.[CrossRef][Medline]
16. Mazzeo S, Caramella D, Battolla L, et al. Crohn disease of the small bowel: spiral CT evaluation after oral hyperhydration with isotonic solution. J Comput Assist Tomogr 2001;25:612–616.[CrossRef][Medline]
17. Raptopoulos V, Schwartz RK, McNicholas MM, Movson J, Pearlman J, Joffe N. Multiplanar helical CT enterography in patients with Crohn's disease. AJR Am J Roentgenol 1997;169:1545–1550.[Abstract/Free Full Text]
18. Maccioni F, Viscido A, Broglia L, et al. Evaluation of Crohn disease activity with magnetic resonance imaging. Abdom Imaging 2000;25:219–228.[CrossRef][Medline]
19. Rollandi GA, Curone PF, Biscaldi E, et al. Spiral CT of the abdomen after distention of small bowel loops with transparent enema in patients with Crohn's disease. Abdom Imaging 1999;24:544–549.[CrossRef][Medline]
20. Mako EK, Mester AR, Tarjan A, Karlinger K, Toth G. Enteroclysis and spiral CT examination in diagnosis and evaluation of small bowel Crohn disease. Eur J Radiol 2000;35:168–175.[CrossRef][Medline]
21. Low RN, Sebrechts CP, Politoske DA, et al. Crohn disease with endoscopic correlation: single-shot fast spin-echo and gadolinium-enhanced fast-suppressed spoiled gradient-echo MR imaging. Radiology 2002;222:652–660.[Abstract/Free Full Text]
22. Bender GN, Maglinte DD, Kloppel VR, Timmons JH. CT enteroclysis: a superfluous diagnostic procedure or valuable when investigating small bowel disease? AJR Am J Roentgenol 1999;172:373–378.[Free Full Text]
23. Bender GN, Timmons JH, Williard WC, Carter J. Computed tomographic enteroclysis: one methodology. Invest Radiol 1996;31:43–49.[CrossRef][Medline]
24. Wold PB, Fletcher JG, Johnson CD, Sandborn WJ. Assessment of small bowel Crohn disease: noninvasive peroral enterography compared with other imaging methods and endoscopy—feasibility study. Radiology 2003;229:275–281.[Abstract/Free Full Text]
25. Madureira AJ. The comb sign. Radiology 2004;230:783–784.[Free Full Text]
26. Horton KM, Eng J, Fishman EK. Normal enhancement of the small bowel: evaluation with spiral CT. J Comput Assist Tomogr 2000;24:67–71.[CrossRef][Medline]
27. Costamagna G, Shah SK, Riccioni ME, et al. A prospective trial comparing small bowel radiographs and video capsule endoscopy for suspected small bowel disease. Gastroenterology 2002;123:999–1005.[CrossRef][Medline]
28. Hara AK, Leighton JA, Sharma VK, Fleischer DE. Small bowel: preliminary comparison of capsule endoscopy with barium study and CT. Radiology 2004;230:260–265.[Abstract/Free Full Text]
29. Boivin ML, Volderholzer VW, Lochs H. The M2A patency capsule: the Berlin experience [abstr]. Gastroenterology 2004;124(suppl 1):A-459.
30. Heigh RI, Leighton JA, Hara A, et al. Diagnosing small bowel Crohn's disease: video caspule endoscopy and CT enterography lead the way—small bowel follow through less helpful [abstr]. Gastroenterology 2004;124(suppl 1):A-37.

This article has been cited by other articles: