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) Alert me when this article is cited 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 Johnson, K. T. Articles by Summers, R. L. Search for Related Content PubMed PubMed Citation Articles by Johnson, K. T. Articles by Summers, R. L. Social Bookmarking                      What's this?

CT Colonography Using 360° Virtual Dissection: A Feasibility Study

¹ Department of Radiology, Mayo Clinic Rochester, 200 First St. SW, Rochester, MN 55905.
² Department of Radiology, National Institutes of Health, Bethesda, MD 20892.

Received November 15, 2004; accepted after revision January 4, 2005.

 
R. L. Summers has patents pending and awarded in the subject area of this article.

The opinions and assertions contained herein are the private views of the authors and are not to be construed as official or as representing the views of the United States government.

Address correspondence to C. D. Johnson.

Abstract

Top Abstract Introduction Materials and Methods Results Discussion References

 
OBJECTIVE. Using a 3D rendering technique called "virtual dissection," we sought to evaluate polyp and fold distortion using a colon phantom, estimate the polyp detection performance in humans, and estimate the added benefit of double interpretation and computer-aided diagnosis.

MATERIALS AND METHODS. A colon phantom containing 144 polyps of varying sizes (5-12 mm) and shapes (flat, sessile, pedunculated) was scanned. Polyp shape and distortion at virtual dissection were categorized as flame, club, pea, or bizarre. Haustral fold distortion was graded. The CT colonography examinations in 20 consecutive patients (colonoscopically proven normal findings, n = 5; polyps 1 cm, n = 17 in 15 patients) were blindly reviewed by three radiologists using the virtual dissection technique. The added benefits of double interpretation and computer-aided diagnosis were tabulated.

RESULTS. Sessile polyps appeared flame (35/48 [73%]) or pea (11/48 [23%]) in shape. Flat polyps appeared flame-shaped (31/47 [66%]) or pea-shaped (16/47 [34%]). Pedunculated polyps were flame (15/45 [33%]), club (20/45 [44%]), or pea (6/45 [13%]) in shape. Axial distortion occurred along the longitudinal axis. The sensitivities of the three observers for polyps of 1 cm or more were 16/17 (94%), 14/17 (82%), and 15/17 (88%). The specificities were 5/5 (100%), 5/5 (100%), and 4/5 (80%). Sensitivities after double interpretation and computer-aided diagnosis improved but did not reach statistical significance.

CONCLUSION. Although distortion of colonic structures exists at virtual dissection, it does so in recognizable patterns, so that sensitivity for polyp detection is not compromised.

Keywords: colon polyps • CT colonography • virtual dissection

Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

 
CT colonography is a promising technique for the evaluation of the colorectum. Recent reports of its performance indicate its high potential to become an alternative to optical colonoscopy in screening for colorectal polyps and cancer [1]. Other reports, however, indicate the performance of CT colonography is clearly lower than that of colonoscopy [2, 3]. The cause for the discrepant findings is unclear, but perceptive and technical issues are likely at fault [3-5]. Perceptive causes for error have been attributed as the major cause for error among lesions of 1 cm or larger [3]. The large amount of data on CT may contribute to these perceptive errors: Assuming conventional CT colonography interpretation methods and a screening population, a radiologist would likely review more than 14,000 images before a single 1-cm polyp is discovered (700 images per data set, 5% incidence of 1-cm lesions). Reducing the number of images to review may reduce radiologist fatigue (or complacency) and result in improved polyp detection.

Virtual dissection is a new 3D rendering technique that draws a midline trace through the colon and displays the entire luminal surface of the colon as a flattened 2D image. The image produced is similar to a Mercator map and resembles the pathologic display of a resected colon specimen. Using this interpretation paradigm, the number of images required to view the luminal surface of the colon is dramatically reduced to a few images (depending on the preferred length of the colon displayed). Unlike conventional-perspective volume-rendered 3D endoluminal images, a tedious centimeter-by-centimeter fly-through is not required. Rather, the colonic lumen is viewed by sequentially looking at the colon segment by segment, similar to a barium enema. Furthermore, computer-aided diagnosis (CAD) can be applied to the same data set, potentially improving polyp detection. Double interpretation has also been proven to reduce interobserver variability associated with conventional CT colonography interpretation methods [6].

View larger version (137K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 1A —360° virtual dissection colonography of glass colon phantom. Single virtual dissection image shows entire luminal surface that contains 144 polyps. All 360° of colon phantom is displayed along its entire longitudinal axis.

View larger version (95K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 1B —360° virtual dissection colonography of glass colon phantom. CT topogram of colon phantom shows multiple polyps in air-filled lumen.

Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

 
Phantom Study
A phantom study was conducted to display polyps of known size and location in order to assess image distortion that normally occurs with virtual dissection image display techniques.

The CT data set of a previously studied colon phantom [6] was evaluated using virtual dissection software (Virtual Dissection, version 3.0.64q, GE Healthcare). The phantom is constructed of glass and contains 144 soft-tissue-attenuation polyps of different sizes (5, 7, 10, and 12 mm), shapes (flat, sessile, and pedunculated), and locations (on the fold, on the tip of the fold, on the base of the fold, and on the wall). The phantom was placed in a water bath simulating the body cavity and was scanned using a LightSpeed Ultra-8 CT scanner (GE Healthcare). Scanning parameters were 8 x 1.25 mm detector configuration, 2.5-mm slice thickness, 1.25-mm reconstruction interval, 120 kVp, 130 mA, 65 mAs, 13.5 mm/rotation table speed, and 0.5-sec tube rotation time. This technique is identical to our clinical CT colonography technique. The CT data were sent to a workstation (ADW 4.2, dual monitors, GE Healthcare) and viewed using the virtual dissection software. This requires a midline trace that is performed automatically with manual confirmation of its accuracy. If the midline trace is not accurately mapped, manual seed placement occurs.

Three-hundred-sixty degree virtual dissection images of the phantom (Figs. 1A and 1B) were assessed in an unblinded fashion for the presence and morphologic appearance of each polyp. On the basis of preliminary review, 3D polyp morphology was classified as having a flame, club, pea, or bizarre shape (Figs. 2A, 2B, 2C, and 2D). Haustral fold distortion was also noted as none, mild, moderate, or severe. The direction of distortion (axial, longitudinal, or both) in reference to the midline trace and the degree of distortion (none, mild, moderate, and severe) were recorded.

View larger version (56K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 2A —Virtual dissection CT colonography images depict four morphologic types of polyps and their distorted appearances. Four polyp shapes are flame (A), club (B), pea (C), and bizarre (D).

View larger version (92K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 2B —Virtual dissection CT colonography images depict four morphologic types of polyps and their distorted appearances. Four polyp shapes are flame (A), club (B), pea (C), and bizarre (D).

View larger version (62K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 2C —Virtual dissection CT colonography images depict four morphologic types of polyps and their distorted appearances. Four polyp shapes are flame (A), club (B), pea (C), and bizarre (D).

View larger version (120K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 2D —Virtual dissection CT colonography images depict four morphologic types of polyps and their distorted appearances. Four polyp shapes are flame (A), club (B), pea (C), and bizarre (D).

Three radiologists who had extensive experience with CT colonography (each having interpreted > 750 cases) reviewed the images in a blinded fashion using 360° virtual dissection software as the primary interpretation method; they noted the presence, size, and location of any lesions. Two-dimensional axial and multiplanar images and 3D endoluminal views were used for problem solving and to improve reviewer confidence. To evaluate the effectiveness of double interpretation, the individual interpretations of each radiologist were paired with those of each of the other two radiologists. Consensus interpretation was not used.

View larger version (14K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 3 —Bar graphs show morphology of polyps at virtual dissection. Sessile and flat polyps appear predominately as flame- and pea-shaped. Pedunculated polyps appear flame- and club-shaped.

Results

Top Abstract Introduction Materials and Methods Results Discussion References

 
Phantom Study
All 144 polyps in the colon phantom were found at nonblinded review using virtual dissection software. Distortion occurred in nearly all polyps. A summary of the three types of polyps and their distorted appearances is shown in Figure 3. Sessile and flat polyps were either flame-shaped (35/48 [73%] and 31/47 [66%]) or pea-shaped (11/48 [23%] and 16/47 [34%]), respectively. Pedunculated polyps had a more varied distorted shape. A few polyps did not readily fit into these shape categories and were classified as having a bizarre shape. Four of five polyps classified as bizarre were pedunculated.

Table 1 summarizes the locations and amounts of haustral fold distortion. Mild haustral fold distortion usually occurred along the superior (top) and inferior (bottom) aspects of the virtual dissection image. Severe distortion (marked widening and contour irregularity) was most likely to occur in the middle of the haustral fold. Haustral distortion occurred more frequently and to a greater degree at a colonic bend or flexure. Nearly all folds (90%) had at least mild distortion along the longitudinal (midline trace) axis of the colon. No axial (perpendicular to the midline) distortion was observed.

Human Study
Sensitivities of the three radiologists for detecting the 17 proven colorectal polyps or cancers of 1 cm or larger were 16/17 (94%), 14/17 (82%), and 15/17 (88%), with specificities of 5/5 (100%), 5/5 (100%), and 4/5 (80%). For the 20 polyps of 5 mm in diameter or larger, the sensitivities were 17/20 (85%), 15/20 (75%), and 17/20 (85%) with specificities of 5/5 (100%), 5/5 (100%), and 4/5 (80%).

View larger version (67K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 4A —82-year-old man with ascending colon carcinoma. 360° virtual dissection image of right colon reveals lesion (arrow) that was considered to be normal ileocecal valve. A 2.5-cm adenoma was found at colonoscopy. Normal valve can be identified in more proximal colon (arrowhead).

View larger version (85K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 4B —82-year-old man with ascending colon carcinoma. Three-dimensional endoluminal view shows proven adenoma.

View larger version (24K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 5 —Bar graphs show added benefit of double interpretation (gray bars). All combinations of the three reviewers are listed. Sensitivities tended toward improvement in every case, but improvement did not reach statistical significance. Black bars indicate single interpretation.

View larger version (15K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 6 —Bar graphs show benefit of adding computer-aided diagnosis (gray bars) to single blinded interpretation (black bars).

View larger version (86K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 7A —87-year-old man with annular carcinoma in sigmoid colon. 360° virtual dissection image shows annular carcinoma in sigmoid colon. Abrupt margins of tumor are identified at arrows.

View larger version (77K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 7B —87-year-old man with annular carcinoma in sigmoid colon. Typical features of annular cancer (arrow) are visible on this conventional axial image at level of sigmoid colon.

View larger version (88K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 8A —Typical example of a sessile polyp in 77-year-old woman. 360° virtual dissection image shows 1-cm polyp (arrow) that appears elongated or flame-shaped.

View larger version (141K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 8B —Typical example of a sessile polyp in 77-year-old woman. Three-dimensional endoluminal image shows sessile polyp.

Sensitivities using CAD alone for polyps of 1 cm or greater and 5 mm or greater were 12/17 (71%) and 13/20 (65%), respectively. False-positives existed in all cases; therefore, the specificity was 0. There were 1.3 false-positive findings per patient. The added benefit of using CAD combined with each of the three reviewers improved the sensitivity for a single reviewer from 14/17 (82%) to 16/17 (94%) for polyps 1 cm or larger. Two of three reviewers benefited from CAD for detecting lesions 5 mm or greater, with sensitivities improving 5-15% to 18/20 (90%), 18/20 (90%), and 17/20 (85%). One reviewer found no additional polyps (Fig. 6). The improved performance was not statistically significant.

Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

 
CT colonography interpretation using virtual dissection 3D endoluminal rendering was evaluated. Much like a Mercator map of the earth on which the polar regions are artificially enlarged and distorted, the colon anatomy is altered as it is mapped into a rectangular 2D strip. Complex alterations in the colon shape (most notably at flexural regions) result in increased anatomic distortion. Despite distortion inherent in straightening and flattening a 3D structure into a 2D image, polyps and cancers can be recognized and detected with virtual dissection rendering (Figs. 7A, 7B, 8A, and 8B). The sensitivity for polyp detection was high among the three reviewers who evaluated the 20 patients in our study group. Polyp detection was not improved significantly with either double interpretation or CAD; however, a trend was seen that may have proven significant with a larger sample size.

When a proven phantom with optimal preparation (no stool, fluid, or motion artifacts and full distention) is used, polyps are detectable at virtual dissection. Image distortion occurs at virtual dissection because a 3D cylindric and curved structure is mapped as a 2D straight image. Polyps are distorted into recognizable patterns: flame, club, pea, or bizarre. Haustral folds are distorted least along the superior and inferior aspects of the virtual dissection image display. Haustral folds in the middle aspect of the image often appear enlarged and bulbous. Distortion occurs solely along the longitudinal axis of the colon. The phantom study indicates the high potential of this technique. Like all 3D endoluminal renderings, the usefulness of this technique likely depends on ideal preparation of the colon.

In patients, polyps 5 mm or larger in diameter were detected with a sensitivity of 75-85%. This detection rate is not significantly different from those in reports using conventional CT colonography interpretation methods [1, 9-11]. As an early feasibility study, these results are encouraging because a totally different paradigm for image review is required at virtual pathology. It is possible that with additional experience, diagnostic performance might improve.

The causes for error were investigated. All polyps were found in retrospect. This indicates a high potential for virtual dissection. An incomplete midline trace through the cecum accounted for one third of the false-negative examinations. Since the completion of our study the manufacturer has addressed this problem, and the midline trace nearly always extends the full distance of the colon automatically. In rare instances in which the trace is not extended to the base of the cecum, all the colonic mucosa should be manually reviewed.

Perceptive error accounted for two additional cecal errors. In one patient a cecal polyp was believed to be the ileocecal valve. This type of error occurred with conventional CT colonography interpretation methods early in the development of that technique. The normal valve will probably be readily recognized with more experience or by correlation with multiplanar reformatted 2D images. Another cecal polyp hidden behind a haustral fold was also overlooked. Careful inspection of each haustral fold must be performed to detect flat lesions and those that do not project markedly into the lumen. A diminutive polyp was overlooked in a partially collapsed cecum in a single patient. A single flat polyp in the sigmoid was difficult to identify.

Although interpretation times were not formally documented by all reviewers in this study, they were considered by our group to be highly variable. For those times recorded, interpretation times varied between 5 and 32 min. We believe that this was probably because of unfamiliarity with this technique, despite reviewer training. It is common that interpretation times can be long until confidence with the technique is gained.

Virtual dissection does require user input at two stages. The first stage is confirmation of the midline trace. If this is done automatically, it is quick and adds little extra time to the evaluation. However, if manual tracing is required and the colon is suboptimally distended, tracing can require considerable effort and time. For this reason, we elected to train registered technologists to trace the colons before interpretation. In this way, the radiologist can more rapidly assess the image for abnormalities.

Virtual dissection shows great promise in reducing the number of images to be interpreted. Cotton et al. [2] showed that 3D endoluminal images can increase the sensitivity for colorectal lesions at CT colonography by approximately 12%. Virtual dissection provides a parody for examining the colonic lumen, with only a few images that display entire colonic segments, dramatically reducing the time it takes to view 3D images of the colonic lumen. This markedly decreases the difficulty inherent in reviewing CT colonography examinations.

In summary, CT colonography using virtual dissection image display is a feasible and promising technique. Polyps and normal colonic anatomy are distorted but can be recognized with training. Virtual dissection addresses a fundamental problem of conventional CT colonography by reducing the number of images to review. Even at this early stage of its development, the estimated performance of virtual dissection is similar in sensitivity and specificity to in reports of conventional CT colonography. Polyp detection can be improved with the use of either double interpretation or CAD. Further study of this technique in larger in vivo studies should be encouraged.

References

Top Abstract Introduction Materials and Methods Results Discussion References

 

1. 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]
2. Cotton PB, Durkalski VL, Pineau BC, et al. Computed tomographic colonography (virtual colonoscopy): a multicenter comparison with standard colonoscopy for detection of colorectal neoplasia. JAMA 2004; 291:1713 -1719[Abstract/Free Full Text]
3. Johnson CD, Harmsen WS, Wilson LA, et al. Prospective blinded evaluation of computed tomographic colonography for screen detection of colorectal polyps. Gastroenterology 2003;125 : 311-319[CrossRef][Medline]
4. Gluecker T, Fletcher JG, Welch TJ, et al. Characterization of lesions missed on interpretation of CT colonography using a 2D search method. AJR 2004; 182:881 -889[Abstract/Free Full Text]
5. Taylor SA, Halligan S, Bartram CI, et al. Multi-detector row CT colonography: effect of collimation, pitch, and orientation on polyp detection in a human colectomy specimen. Radiology2003; 229:109 -118[Abstract/Free Full Text]
6. Johnson KT, Johnson CD, Anderson SM, Bruesewitz MR, McCollough CH. CT colonography: determination of optimal CT technique using a novel colon phantom. Abdom Imaging 2004;29 : 173-176[Medline]
7. Summers R, Johnson C, Pusanik L, Malley J, Yousef A, Reed J. Automated polyp detection for CT colonography: feasibility assessment in a human population. Radiology 2001;219 : 51-59[Abstract/Free Full Text]
8. Summers R, Jerebko AK, Franaszek M, Malley JD, Johnson CD. Colonic polyps: complementary role of computer-aided detection in CT colonography. Radiology 2002;225 : 391-399[Abstract/Free Full Text]
9. Pineau BC, Paskett ED, Chen GJ, et al. Virtual colonoscopy using oral contrast compared with colonoscopy for the detection of patients with colorectal polyps. Gastroenterology 2003;125 : 304-310[CrossRef][Medline]
10. Fletcher JG, Johnson CD. Computed tomographic colonography: current and future status for colorectal cancer screening. Semin Roentgenol 2000; 35:385 -393[CrossRef][Medline]
11. Yee J, Akerkar G, Hung R, Steinauer-Gebauer A, Wall S, McQuaid K. Colorectal neoplasia: performance characteristics of CT colonography for detection in 300 patients. Radiology2001; 219:685 -692[Abstract/Free Full Text]

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				D. Hock, R. Ouhadi, R. Materne, A.-S. Aouchria, I. Mancini, T. Broussaud, P. Magotteaux, and A. Nchimi Virtual Dissection CT Colonography: Evaluation of Learning Curves and Reading Times with and without Computer-aided Detection Radiology, September 1, 2008; 248(3): 860 - 868. [Abstract] [Full Text] [PDF]

				C. D. Johnson, J. G. Fletcher, R. L. MacCarty, J. N. Mandrekar, W. S. Harmsen, P. J. Limburg, and L. A. Wilson Effect of Slice Thickness and Primary 2D Versus 3D Virtual Dissection on Colorectal Lesion Detection at CT Colonography in 452 Asymptomatic Adults Am. J. Roentgenol., September 1, 2007; 189(3): 672 - 680. [Abstract] [Full Text] [PDF]

				S. H. Kim, J. M. Lee, H. W. Eun, M. W. Lee, J. K. Han, J. Y. Lee, and B. I. Choi Two- versus Three-dimensional Colon Evaluation with Recently Developed Virtual Dissection Software for CT Colonography Radiology, September 1, 2007; 244(3): 852 - 864. [Abstract] [Full Text] [PDF]

				S. H. Park, E. K. Choi, S. S. Lee, J.-S. Byeon, J.-Y. Jo, Y. H. Kim, K. H. Lee, H. K. Ha, and J. K. Han Polyp Measurement Reliability, Accuracy, and Discrepancy: Optical Colonoscopy versus CT Colonography with Pig Colonic Specimens Radiology, July 1, 2007; 244(1): 157 - 164. [Abstract] [Full Text] [PDF]

				D. S. WERDERMAN CT Colonography Radiol. Technol., March 1, 2007; 78(4): 309CT - 324CT. [Abstract] [Full Text] [PDF]

				T. Mang, A. Maier, C. Plank, C. Mueller-Mang, C. Herold, and W. Schima Pitfalls in Multi-Detector Row CT Colonography: A Systematic Approach RadioGraphics, March 1, 2007; 27(2): 431 - 454. [Abstract] [Full Text] [PDF]

				A. C. Silva, C. V. Wellnitz, and A. K. Hara Three-dimensional Virtual Dissection at CT Colonography: Unraveling the Colon to Search for Lesions RadioGraphics, November 1, 2006; 26(6): 1669 - 1686. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) Alert me when this article is cited 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 Johnson, K. T. Articles by Summers, R. L. Search for Related Content PubMed PubMed Citation Articles by Johnson, K. T. Articles by Summers, R. L. Social Bookmarking                      What's this?