Eur Radiol (2008) 18: 429–437DOI 10.1007/s00330-007-0764-1 GASTROINTESTINAL

Andrik J. AschoffAndrea S. ErnstHans-Juergen BrambsMarkus S. Juchems

Received: 8 February 2007Revised: 25 July 2007Accepted: 24 August 2007Published online: 25 September 2007# European Society of Radiology 2007

CT colonography: an update

Abstract Computed tomographic(CT) colonography (CTC)—alsoknown as “virtual colonoscopy”—wasfirst described more than a decadeago. As advancements in scannertechnology and three-dimensional(3D) postprocessing helped developthis method to mature into a potentialoption in screening for colorectalcancer, the fundamentals of the ex-amination remained the same. It is aminimally invasive, CT-based proce-dure that simulates conventionalcolonoscopy using 2D and 3D com-

puterized reconstructions. The prima-ry aim of CTC is the detection ofcolorectal polyps and carcinomas.However, studies reveal a wide per-formance variety in regard to polypdetection, especially for smallerpolyps. This article reviews the avail-able literature, discusses establishedindications as well as open issues andhighlights potential future develop-ments of CTC.

Keywords CT colonography .Screening . Colon cancer

Introduction

“Virtual colonoscopy” was first described by David Viningmore than a decade ago [1]. Its potential was quicklyevident and it developed from a research tool to a potentialscreening method for colorectal cancer. The basic approachof acquiring a three-dimensional (3D) dataset after colondistension in combination with 2D/3D postprocessing hasnot changed, though. The preferred terminology for thismethod should be computed tomographic colonography(CTC) [2].

Epidemiology

Colorectal cancer is one of the most common causes forcancer deaths. With an estimated number of 106,680 newcases in 2006 and 55,170 estimated deaths in the same year,it is the third most frequent cancer found among men andwomen in the United States and the third most commoncause of death among cancers [3].

A variety of predisposing factors for developing colo-rectal cancer are known. These include: a first-degree

relative in whom colon cancer or a large adenomatouspolyp was diagnosed before the age of 60 years; inflam-matory bowel disease; a history of familial adenomatouspolyposis or hereditary nonpolyposis; colorectal cancersyndromes and prior adenomatous polyps or colon cancers.

In developing colorectal cancer, the colonic polyps—inparticular the adenomatous polyps—play a key role.Through a genetic mutation, normal colon cells maytransform into hyperproliverative epithelium, adenomasand finally into colorectal cancer [4]. Colorectal polyps canbe removed endoscopically (polypectomy), and colono-scopic polypectomy results in a lower than expectedincidence of colorectal cancer [5] (Fig. 1).

The symptoms of colorectal cancer are often unchar-acteristic (obstipation, diffuse pain, loss of weight) andmay delay the final diagnosis. Therefore tests are desired todetect colon cancer in its early stages or—even better—itsprecursors, the adenomatous polyps.

Various screening tests for colorectal cancer areavailable—fecal occult blood test (FOBT), sigmoidoscopyand colonoscopy are the most important ones. Each of thesemethods has its limitations. FOBT is a very specific but notsensitive test [6] and sigmoidoscopy does not assess the

A. J. Aschoff (*) . A. S. Ernst .H.-J. Brambs . M. S. JuchemsDiagnostic and Interventional Radiology,University Hospitals of Ulm,Steinhoevelstr. 9,89070 Ulm, Germanye-mail: andrik.aschoff@uni-ulm.deTel.: +40-731-50061003Fax: +40-731-50061005

whole colon. Nevertheless, even sigmoidoscopy leads to adecrease of mortality [7], although up to 52% of thecolorectal polyps are located proximal to the rectosigmoid[8] and significant polyps can be overseen if locatedproximal to the rectosigmoid [9].

Colonoscopy is considered the “gold standard” ofscreening for colorectal polyps and endoscopy incorporatesthe possibility to perform a biopsy or excision of a polyp[10, 11]. It offers high sensitivity and specificity, but somelimitations apply to this method as well, including theinvasiveness with a low but existent complication rate andthe inconvenience patients have to suffer [12–14]. The rateof serious complications during or immediately aftercolonoscopy has been reported to be below 0.1% [15].

CTC

Indications

Besides the potential of a screening option for colorectalpolyps and cancer, which is still controversial, CTC hasincreasingly established clinical indications, includingincomplete colonoscopy, evaluation of the colon proximalto a stenosing lesion, and investigation of patients that arenot in a suitable condition for undergoing a conventionalcolonoscopy (e.g., patients with bleeding disorders), or forpatients refusing conventional colonoscopy [16–19].

CT equipment

CTC has benefited greatly from developments in computedtomography (CT) scanner technology and new methods of3D image processing. The introduction of spiral CT in theearly 1990s culminated in the development of multi-detector (MD) CT systems that allow scanning of greatervolumes with thinner slices in less time. The possibility toperform an examination of the whole abdomen within a

single breath hold without respiration artifacts is anadvantage that is particularly beneficial for CTC.

Patient preparation

Various studies have been published on how to preparepatients prior to the examination. A well-distended,virtually stool- and fluid-free colon is the traditional wayto perform a CTC [20, 21]. The cathartic cleansing keepsthe reader from misinterpreting residual feces and helps toinvestigate as much colonic surface as possible. Unfortu-nately, this is highly dependent on patient compliance.

Other publications emphasize that it is possible toperform a CTC without cathartic cleansing by using stool-and fluid-tagging [22, 23]. Fecal- and fluid-tagging isachieved by fractionated oral administration of contrastagent (barium, sodium/meglumine diatrizoate). This helpsto improve discrimination between stool residues andgenuine polyps, and thus potentially improves the speci-ficity of CTC [23, 24].

Different regimes and medications to perform a catharticbowel cleansing are in use. A lot of experience with thesemedications exists because most of them are also used priorto optical colonoscopy. In addition to home-made cleans-ing solutions that some favor, most institutions rely on oneof three commercially available groups of cleansingsolutions: polyethylene glycol [PEG; e.g., GoLytelypreparations (Delcoprep) and NuLytely preparations(Endofalk)]-based, phospho-soda-based solutions (e.g.,Fleet Phospho-soda) and magnesium citrate (e.g., LoSo-Prep). Although some studies show an advantage forphospho-soda, we favor a NuLytely PEG-based cleansingprotocol as we are constantly able to achieve sufficientcleansing with this preparation regime [25].

Macari et al. [20] compared phospho-soda-based (‘dry’)cathartic preparation with PEG-electrolyte-based solutions(‘wet’) for CTC and found significantly less residual fluidand a drier mucosal surface using phospho-soda. Althoughpreparations from both groups have been demonstratedsafe for use in healthy individuals, caution should be takenin selecting a bowel preparation for individuals withsignificant comorbid conditions [26]. Sodium phosphate iscontraindicated in patients with serum electrolyte imbal-ances, advanced hepatic dysfunction, acute and chronicrenal failure, recent myocardial infarction, unstable angina,congestive heart failure, ileus, malabsorption, and ascites[26].

As mentioned above, it is possible to perform CTCwithout cathartic cleansing (“dry” preparation) [22, 27].Bielen et al. [24] reported high patient compliance andacceptance using a “dry” bowel preparation. Especiallynoncompliant, elderly patients could benefit from a drypreparation. Combined with fecal tagging the use ofautomated cleansing techniques [28, 29] seems to be verypromising.

Fig. 1 a Normal colon as seen in optical colonoscopy and b cor-responding endoluminal CTC reconstruction

430

After colonic cleansing, good bowel distension duringscanning is required to further enhance visualization of thecolonic surface. The easiest and most inexpensive way ofachieving a pneumocolon is insufflating room air through arectal enema tube into the colon using a manual balloonpump. This can be performed physician or patientcontrolled. We routinely achieve satisfactory results usingthe later approach [25].

CO2 administration instead of room air can be doneautomatically and patient controlled as well [30]. Somefavor CO2 over room air as it is supposed to be bettertolerated by means of patient comfort and potentiallyprovides better colonic distension [31].

Several studies showed that it is imperative to performthe CT scan in the prone and supine patient positions [32–35]. This helps to reallocate fluid to reveal previouslyhidden colonic surface, and insufflated air is redistributedduring patient repositioning to further improve sensitivityand specificity in detecting colorectal polyps.

Another controversially discussed issue is the use ofspasmolytics (glucagon, N-butylscopolamin) prior to thescan to improve colonic distension. While some studiesclearly show beneficial results in regard to boweldistension and patient comfort when using spasmolytics[36, 37], others failed to do so [35, 38]. Nevertheless, weroutinely use spasmolytics.

CTC scanning protocols

Both collimation and pitch may influence the sensitivity forpolyp detection [39].

In the recent literature a clear trend for recommendinglower collimations can be recognized [18, 47, 48].According to the “Consensus statement onCTcolonography”of the European society of gastrointestinal and abdominalradiology ESGAR [40], a collimation of less than 3 mm iscurrently recommended.

The downside of acquiring thinner slices may beincreased radiation dose. As radiation is of serious concernwhen performing CTC, exposure has to be kept as low aspossible. Consequently many studies investigated so-calledlow-dose protocols [41–44]. In a study that was publishedby Macari et al. in 2002 [44], 50 mAs tube current, 120 kVvoltage and a collimation of 4×1 mm were used. Largepolyps (>10 mm) were detected with a sensitivity of 93%,whereas the detection of intermediate size polyps (6–9 mm)was compromised totaling in a sensitivity of 70%. Evenmore encouraging results were published by Iannacone etal. in 2003 [43]. With a current of 10 mAs and a voltage of140 kVp, 100% sensitivity was reached for large polyps(>10 mm) as well as for intermediate sized ones (6–9 mm).

The “Consensus statement” mentioned above [44]currently recommends for supine and prone non-IV contrastenhanced acquisition with a tube current of 100 mAs or less(50 mAs or less), dependent on available CT technology.

2D and 3D data work-up

The most important 2D procedures are the evaluation of theaxial images and multiplanar reconstructions (MPR) thatcan be used to reconstruct any required scan orientation.

In early studies on CTC, the complementary use of 2Dand 3D images together was assumed to provide thehighest sensitivity [45, 46]. Although some of the newerstudies still favor a solitary 2D interpretation over 3Dviewing [47, 48], some studies yielded low sensitivities forlarge polyps, ranging from 55% to 64% [49–51] using aprimary 2D approach. On the other hand, some authorsprefer primary 3D work-up [52, 53]. The impressive results(sensitivity 93.8% for lesions >10 mm) achieved in thelatter study all go along with further improved CTCsoftware as well as in CT scanner hardware.

A major disadvantage of a primary 3D endoluminalapproach—similar to conventional colonoscopy—is thelack of ability to look behind haustral folds. Ante- andretrograde fly-throughs are therefore mandatory, resultingin increased interpretation time. New visualization meth-ods like an unfolded cube projection [54] and virtualdissection displays (e.g., Philips Filet view, Fig. 2, [47, 55,56]) have the potential to overcome this limitation.Significant decrease of evaluation time down to 10 minper case could be observed with these techniques.

Although it is our personal experience that lesiondetection is best performed using a 3D approach, it isnecessary to further evaluate detected lesions in a 2D orMPR work-up to avoid pitfalls caused by misinterpretingstool residuals, inverted diverticula, etc. as genuine colo-rectal polyps (Fig. 3).

Much emphasis has been put into CTC research, but therehas been much unsteadiness in the results obtained withregard to sensitivity and specificity, as outlined below. Itseems clear that reader performance increases with expe-rience [57, 58], and physicians who do CTC should receiveappropriate training to perform and evaluate CTC datasets.

A lot of research [59–63] has been performed onautomated polyp detection [computer-aided detection(CAD), Fig. 4], using different approaches: CAD as a “firstreader”, as a “concurrent reader,” or as a “second reader.”Asa “first reader”, CAD separates out negative cases beforephysicians read the cases. However, this paradigm bears therisk of missing colorectal lesions. CAD as a “concurrentreader” presents suspect lesions to the physician, who canthen further evaluate the presented lesion and decide whetherit is a genuine polyp or not. It is unclear whether this type ofevaluation biases the reader by drawing attention to CADfindings. Finally CAD as “second reader” re-reviews CTCcases. Thus, CAD findings can be used to alter a report iflesions have been missed in first place. Using CAD as a“second reader” in this specific setting may lead to anincrease in reading time, though [64].

Recently published data showed encouraging resultsusing CAD. In a study performed by Taylor et al. [62],

431

CAD reached a sensitivity of 92% for polyps >10 mm.Overall the CAD performance was comparable and insome parts even better than that of expert readers.

If a robust CAD algorithm can be found, a furtherdecrease in interpretation time may be on the horizon, andinexperienced readers might miss less polyps.

Although so-called “flat” adenomas are not considered tohave a markedly increased risk of developing into invasivecancer compared with sessile or pedunculated polyps bysome researchers [65], they still remain problematic to bedetected in CTC [66, 67]. Since up to 27% of all baselineadenomas may be flat [65], this remains a problem thatneeds researchers’ attention in the future. One possible

approach is using a “translucency view” (Fig. 5). Thispostprocessing algorithm assigns colors to the mucosabased on HU values. Besides ruling out false positives byhelping to differentiate between stool residuals and colo-rectal polyps, this might (especially if used in conjunctionwith IV contrast media) help to reveal flat adenomas.

CTC performance

Pickhardt and coworkers published in 2003 the first largeprospective study in an average-risk screening population,comparing more than 1,200 CT colonographies with

Fig. 3 Pitfall in CTC.a Polypoid lesion as seen inendoluminal and dissection 3Dreconstructions. b The corre-sponding MPR discriminatesresidual feces (arrow) contain-ing air from a genuine polyp

Fig. 2 Principle of a colondissection display. The colon,previously reconstructed asa tube, is cut at one side (a),unrolled (b) and presentedsimilar to a pathologicalspecimen (c)

432

optical colonography performed on the same day [53].Their sensitivity of CTC in the detection of adenomatouspolyps was 93.8% for polyps at least 10 mm in diameter,93.9% for polyps at least 8 mm in diameter, and 88.7% forpolyps at least 6 mm in diameter. The sensitivity of opticalcolonoscopy for adenomatous polyps was 87.5%, 91.5%,and 92.3% for the three sizes of polyps, respectively. Theyconcluded that CTC is an accurate screening method forthe detection of colorectal neoplasia in asymptomatic

average-risk adults and compares favorably with opticalcolonography (Fig. 6).

Not quite 2 years later, Rockey and coworkers publishedtheir prospective results of more than 600 CTCs incomparison with optical colonoscopy and air contrastbarium enema [51], using a similar study design to the oneused by Pickhardt et al. [53]. When analyzed on a per-patient basis, for lesions 10 mm or larger in size, thesensitivity of CTC was 59% (compared with 93.8%reported by Pickhardt and coworkers), and of colonoscopy98%. For lesions 6–9 mm in size, sensitivity was 51% forCTC and 99% for colonoscopy.

The exact reason for this striking difference in reportedperformance of CTC is yet unclear, although both paperswere extensively discussed and commented in the scientificcommunity.

Three recent major meta-analysis publications tried tosystematically approach the CTC performance in the lightof the reported variability [68–70] (Table 1).

In the largest meta-analysis published up to now,Mulhall et al. [69] systematically reviewed the testperformance of CTC, selecting prospective studies pub-lished before February 2005 from 33 studies (CTC afterfull bowel preparation and insufflation of the colon, withcolonoscopy or surgery as the gold standard, collimationsmaller than 5 mm, both 2D and 3D for scan interpretation)providing data on 6,393 patients. The sensitivity of CTC

Fig. 5 “Tranclucency view”(shown in the middle of theupper row). 1 untagged stool isrepresented in green; 2 pedun-culated polyp is shown in red

Fig. 4 CAD marked 8-mm polyp (blue) in endoluminal (a) anddissection (b) display mode

433

was heterogeneous but improved as polyp size increased[48% (95% CI, 25–70%) for detection of polyps <6 mm,70% (95% CI, 55–84%) for polyps 6–9 mm, and 85%(95% CI, 79–91%) for polyps >9 mm]. In contrast,specificity was homogenous [92% (95% CI, 89–96%)for detection of polyps <6 mm, 93% (95% CI, 91–95%)for polyps 6–9 mm, and 97% (95% CI, 96–97%) forpolyps >9 mm]. They came to the conclusion that, althoughCTC is highly specific, the range of reported sensitivities iswide, and patient or scanner characteristics do not fully

account for this variability. Collimation, type of scanner,and mode of imaging explain some of the discrepancy, butnot all of it.

Patients’ perception and tolerance of CTC comparedwith conventional colonoscopy

Many comparisons of patients’ acceptance and tolerancehave been published, and researchers were able to showthat either CTC is better tolerated than colonoscopy [13,14] or vice versa [71, 72]. The discussion is not very usefulbecause most colonoscopies are performed using sedatives,which makes a direct comparison impossible. Never-theless, bowel preparation seems to be the most uncomfort-able issue for both CTC and colonoscopy from the patients’point of view [14].

CTC in incomplete colonoscopy

CTC after incomplete colonoscopy may be especiallyhelpful for evaluation of the nonvisualized part of the colonafter incomplete colonoscopy, and it can increase thediagnostic yield of masses and clinically important polypsin this context [17, 19, 73]. The rate of incompletecolonoscopies varies between 4% [74] and 19% [75–77].This frequency serves as an important indication forperforming immediate CTC in order to avoid repeat bowelpreparation when possible.

CTC open issues

As discussed above and outlined in the meta-analysis byMulhall et al. [69], the wide range of reported sensitivitiesthat can not be completely explained must be resolvedbefore CTC can be advocated for generalized screeningfor colorectal cancer. CTC has not yet been officiallyaccepted as a screening procedure in any country, mainlybecause of the lack of reliable evidence resulting fromlarge trials.

Another important open issue concerns the relevance ofsmall polyps. The overall prevalence of colorectal polypsindependent of size and histology in an average riskscreening population is estimated to be in the range of 25–

Fig. 6 Colonic polyp. a Dissection view, b endoluminal view,c endoscopic view. Note also diverticulae, especially apparent in a

Table 1 CTC performance as calculated in meta-analyses

Authors Year Numberof studiesincluded

Numberof patientsincluded

Per patient sensitivity;polyps >9 mm(95% CI)

Per patient sensitivity;polyps 6–9 mm(95% CI)

Per patient sensitivity;polyps <6 mm(95% CI)

Sosna et al. [68] 2003 14 1,324 88% (84–93%) 84% (80–89%) 65% (57–73%)

Mulhall et al. [69] 2005 33 6,393 85% (79–91%) 70% (55–84%) 48% (25–70%)

434

30% [78–80]. The current concept of polypectomy is toremove all polyps detected during optical colonoscopy(independent of size and histology). This results in asignificant decrease in mortality from colorectal cancer, asshown by the National Polyp Study Workgroup [5].Although some authors argue that polyps of less than5 mm [81] or 10 mm [82] in size may not need immediatepolypectomy, the majority of gastroenterologists disagree[83], and no prospective study has shown a decrease in themortality of colorectal cancer using any threshold size forpolypectomy. This has several implications for CTC.

Even if CTC would be capable of detecting all or at leastthe majority of the polyps smaller than 10 mm and no false-positive findings would occur, approximately every third orfourth patient would still have to undergo interventionaloptical colonoscopy with polypectomy after CTC. Sincethis seems to be an unacceptably high rate of follow-upendoscopies and the known lower sensitivity of CTC in thedetection of smaller polyps, a threshold size for “relevant”polyps would have to be defined before CTC could be usedin large screening settings.

The other major implication relates to the fact that CTCperforms poorer in the detection of smaller polyps andrelates to the clinical value of a “negative” CTC in terms ofneed for follow-up exams, and the definition of areasonable time interval to do so.

Addressing the significance of missed or found smallpolyps on CTC, Macari et al. [84] recommend a CTCfollow-up after 3 years for lesions smaller than 6 mm. TheESGAR consensus statement recommends that polyps lessthan 5 mm in size should not even be reported, providingthat agreement for such a reporting policy has been madewith those referring patients [40], but no generalrecommendation regarding follow-up intervals was made.

In a consensus proposal, the Working Group on VirtualColonoscopy discusses the so called “clinically important

polyp and the rationale for surveillance” and discriminatesbetween “diminuitive”, “intermediate”, “multiple interme-diate” and “lesions that are 1 cm in size or larger” [85].

In their opinion, “diminuitive” lesions (smaller than5 mm) should not be reported, similar to the approachsuggested by the ESGAR consensus [40]. Patients with oneor two “intermediate” lesions (6–9 mm) should berecommended interval surveillance after up to three yearsdepending on several factors, including patient age, sex,comorbidities, and preference and local practice. More thanthree lesions of 6–9 mm or one lesion of at least 10 mmrequire immediate colonoscopy.

Although aspects of this proposal may be discussedcontroversially, the proposed framework addresses thedefinite need to have a reference guide for interpretation ofCTC results.

Conclusion

CTC offers high sensitivity and specificity for the detectionof colon cancer and polyps larger than 10 mm in size, givensuitable equipment and experienced investigators. It is themethod of choice for incomplete colonoscopies and forpatients that to some reason can not undergo opticalcolonoscopy. Nevertheless, its use in routine screeningshould be subject to the introduction of quality controlswith prescribed examination standards. For screeningpurposes, the wide range of reported sensitivities that cannot be completely explained must be resolved and athreshold size for clinically relevant polyps has to bedefined, and its application backed by prospective studies.In particular, multicenter trials are necessary to achieve thisgoal. Computer-assisted evaluation has the potential toshorten the examination time and may further increasesensitivity in the near future.

References

1. Vining D, Shifrin R, Grishaw E (1994)Virtual colonoscopy. Radiology193:446

2. Johnson CD, Hara AK, Reed JE (1998)Virtual endoscopy: what’s in a name?AJRAm J Roentgenol 171(5):1201–1202

3. Jemal A et al (2006) Cancer statistics,2006. CA Cancer J Clin 56(2):106–130

4. Fearon ER, Vogelstein B (1990) Agenetic model for colorectal tumori-genesis. Cell 61(5):759–767

5. Winawer SJ et al (1993) Prevention ofcolorectal cancer by colonoscopicpolypectomy. The National PolypStudy Workgroup. N Engl J Med 329(27):1977–1981

6. Rockey DC et al (1998) Relativefrequency of upper gastrointestinal andcolonic lesions in patients with positivefecal occult-blood tests. N Engl J Med339(3):153–159

7. Selby JV et al (1992) A case-controlstudy of screening sigmoidoscopy andmortality from colorectal cancer.N Engl J Med 326(10):653–657

8. Bernstein MA et al (1985) Distributionof colonic polyps: increased incidenceof proximal lesions in older patients.Radiology 155(1):35–38

9. Imperiale TF et al (2000) Risk ofadvanced proximal neoplasms inasymptomatic adults according to thedistal colorectal findings. N Engl J Med343(3):169–174

10. Ransohoff DF, Sandler RS (2002)Clinical practice. Screening forcolorectal cancer. N Engl J Med 346(1):40–44

11. Rockey DC (2005) Colon imaging:computed tomographic colonography.Clin Gastroenterol Hepatol 3(7 Suppl1):S37–S41

12. Juchems MS, Ehmann J, Brambs H-J,Aschoff AJ (2005) A retrospectiveevaluation of patient acceptance of CTcolonography (“virtual colonoscopy”)in comparison to conventional colo-noscopy in an average risk screeningpopulation. Acta Radiol 46(7):664–670

435

13. Svensson MH et al (2002) Patientacceptance of CT colonography andconventional colonoscopy: prospectivecomparative study in patients with orsuspected of having colorectaldisease. Radiology 222(2):337–345

14. Thomeer M et al (2002) Patient accep-tance for CT colonography: what isthe real issue? Eur Radiol12(6):1410–1415

15. Rainis T et al (2007) Diagnostic yieldand safety of colonoscopy in Israelipatients in an open access referralsystem. J Clin Gastroenterol 41(4):394–399

16. Fenlon HM et al (1999) Occlusivecolon carcinoma: virtual colonoscopyin the preoperative evaluation of theproximal colon. Radiology210(2):423–428

17. Morrin MM et al (1999) EndoluminalCT colonography after an incompleteendoscopic colonoscopy. AJR Am JRoentgenol 172(4):913–918

18. Neri E et al (2002) Colorectal cancer:role of CT colonography in preopera-tive evaluation after incompletecolonoscopy. Radiology223(3):615–619

19. Gryspeerdt S et al (2005) CTcolonography with fecal tagging afterincomplete colonoscopy. Eur Radiol 15(6):1192–1202

20. Macari M et al (2001) Effect ofdifferent bowel preparations on residualfluid at CT colonography. Radiology218(1):274–277

21. Yee J et al (2001) Colorectal neoplasia:performance characteristics of CTcolonography for detection in 300patients. Radiology 219(3):685–692

22. Lefere PA et al (2002) Dietary fecaltagging as a cleansing method beforeCT colonography: initial results polypdetection and patient acceptance.Radiology 224(2):393–403

23. Callstrom MR et al (2001) CTcolonography without cathartic prepa-ration: feasibility study. Radiology 219(3):693–698

24. Bielen D et al (2003) Dry preparationfor virtual CT colonography with fecaltagging using water-soluble contrastmedium: initial results. Eur Radiol 13(3):453–458

25. Juchems MS et al (2006) Bowelpreparation for CT-colonography:comparison of two different cleansingprotocols. Eur J Radiol 60(3):460–464

26. Wexner SD et al (2006) A consensusdocument on bowel preparation beforecolonoscopy: prepared by a task forcefrom the American Society of Colonand Rectal Surgeons (ASCRS), theAmerican Society for GastrointestinalEndoscopy (ASGE), and the Society ofAmerican Gastrointestinal andEndoscopic Surgeons (SAGES). DisColon Rectum 49(6):792–809

27. Lefere P et al (2005) CT colonographyafter fecal tagging with a reducedcathartic cleansing and a reduced vol-ume of barium. AJR Am J Roentgenol184(6):1836–1842

28. Zalis ME, Hahn PF (2001) Digitalsubtraction bowel cleansing in CTcolonography. AJR Am J Roentgenol176(3):646–648

29. Zalis ME et al (2003) CTcolonography: digital subtraction bowelcleansing with mucosal reconstructioninitial observations. Radiology 226(3):911–917

30. Shinners TJ et al (2006) Patient-controlled room air insufflation versusautomated carbon dioxide delivery forCT colonography. AJR Am JRoentgenol 186(6):1491–1496

31. Burling D et al (2006) Automatedinsufflation of carbon dioxide forMDCT colonography: distension andpatient experience compared withmanual insufflation. AJR Am JRoentgenol 186(1):96–103

32. Chen SC et al (1999) CT colonography:value of scanning in both the supineand prone positions. AJR Am JRoentgenol 172(3):595–599

33. Yee J et al (2003) Comparison ofsupine and prone scanning separatelyand in combination at CTcolonography. Radiology226(3):653–661

34. Fletcher JG et al (2000) Optimizationof CT colonography technique:prospective trial in 180 patients.Radiology 216(3):704–711

35. Morrin MM et al (2002) CTcolonography: colonic distention im-proved by dual positioning but notintravenous glucagon. Eur Radiol 12(3):525–530

36. Taylor SA et al (2003) Optimizingcolonic distention for multi-detectorrow CT colonography: effect of hyo-scine butylbromide and rectal ballooncatheter. Radiology 229(1):99–108

37. Rogalla P et al (2005) Spasmolysis atCT colonography: butyl scopolamineversus glucagon. Radiology 236(1):184–188

38. Bruzzi JF et al (2003) Efficacy of IVBuscopan as a muscle relaxant in CTcolonography. Eur Radiol 13(10):2264–2270

39. Taylor SA et al (2003) Multi-detectorrow CT colonography: effect of colli-mation, pitch, and orientation on polypdetection in a human colectomyspecimen. Radiology 229(1):109–118

40. Taylor SA et al (2007) Europeansociety of gastrointestinal and abdom-inal radiology (ESGAR): Consensusstatement on CT colonography. EurRadiol 17(2):575–579

41. Cohnen M et al (2003) Effective dosesin standard protocols for multi-slice CTscanning. Eur Radiol 13(5):1148–1153

42. Cohnen M et al (2004) Feasibility ofMDCT Colonography in ultra-low-dosetechnique in the detection of colorectallesions: comparison with high-resolution video colonoscopy. AJR AmJ Roentgenol 183(5):1355–1359

43. Iannaccone R et al (2003) Feasibility ofultra-low-dose multislice CTcolonography for the detection of co-lorectal lesions: preliminary experience.Eur Radiol 13(6):1297–1302

44. Macari M et al (2002) Colorectalneoplasms: prospective comparison ofthin-section low-dose multi-detectorrow CT colonography and conventionalcolonoscopy for detection. Radiology224(2):383–392

45. Hara AK et al (2001) Colorectal polypdetection with CT colography: two-versus three-dimensional techniques.Work in progress. Radiology 200(1):49–54

46. Royster AP et al (1997) CT colonos-copy of colorectal neoplasms:two-dimensional and three-dimensionalvirtual-reality techniques with colono-scopic correlation. AJR Am JRoentgenol 169(5):1237–1242

47. Hoppe H et al (2004) Virtual colondissection with CT colonography com-pared with axial interpretation andconventional colonoscopy: preliminaryresults. AJR Am J Roentgenol 182(5):1151–1158

48. Bruzzi JF et al (2004) Colonic surveil-lance by CT colonography using axialimages only. Eur Radiol 14(5):763–767

49. Cotton PB et al (2004) Computedtomographic colonography (virtualcolonoscopy): a multicenter compari-son with standard colonoscopy fordetection of colorectal neoplasia. Jama291(14):1713–1719

50. Johnson CD et al (2003) Prospectiveblinded evaluation of computed tomo-graphic colonography for screen detec-tion of colorectal polyps.Gastroenterology 125(2):311–319

436

51. Rockey DC et al (2005) Analysis of aircontrast barium enema, computedtomographic colonography, andcolonoscopy: prospective comparison.Lancet 365(9456):305–311

52. Beaulieu CF et al (1999) Displaymodes for CT colonography. Part II.Blinded comparison of axial CT andvirtual endoscopic and panoramic en-doscopic volume-rendered studies. Ra-diology 212(1):203–212

53. Pickhardt PJ et al (2003) Computedtomographic virtual colonoscopy toscreen for colorectal neoplasia inasymptomatic adults. N Engl J Med349(23):2191–2200

54. Vos FM et al (2003) Three-dimensionaldisplay modes for CT colonography:conventional 3D virtual colonoscopyversus unfolded cube projection. Radi-ology 228(3):878–885

55. Johnson KT et al (2006) CTcolonography using 360-degree virtualdissection: a feasibility study. AJR AmJ Roentgenol 186(1):90–95

56. Juchems MS et al (2006) CTcolonography: comparison of a colondissection display versus 3Dendoluminal view for the detection ofpolyps. Eur Radiol 16(1):68–72

57. Soto JA, Barish MA, Yee J (2005)Reader training in CT colonography:how much is enough? Radiology 237(1):26–27

58. Taylor SA et al (2004) CTcolonography: effect of experience andtraining on reader performance. EurRadiol 14(6):1025–1033

59. Jerebko AK et al (2003) Computer-assisted detection of colonic polypswith CT colonography using neuralnetworks and binary classificationtrees. Med Phys 30(1):52–60

60. Luboldt W et al (2005) Automatedmass detection in contrast-enhanced CTcolonography: an approach based oncontrast and volume. Eur Radiol 15(2):247–253

61. Summers RM et al (2002) Colonicpolyps: complementary role of compu-ter-aided detection in CT colonography.Radiology 225(2):391–399

62. Taylor SA et al (2006) Computer-assisted reader software versus expertreviewers for polyp detection on CTcolonography. AJR Am J Roentgenol186(3):696–702

63. Kiss G et al (2002) Computer-aideddiagnosis in virtual colonography viacombination of surface normal andsphere fitting methods. Eur Radiol 12(1):77–81

64. Mang T et al (2007) Effect ofcomputer-aided detection as a secondreader in multidetector-row CT colo-nography. Eur Radiol DOI 10.1007/s00330-007-0608-z

65. O’Brien MJ et al (2004) Flat adenomasin the National Polyp Study: is thereincreased risk for high-grade dysplasiainitially or during surveillance? ClinGastroenterol Hepatol 2(10):905–911

66. Rex DK, Vining D, Kopecky KK(1999) An initial experience withscreening for colon polyps using spiralCT with and without CT colography(virtual colonoscopy). Gastrointest En-dosc 50(3):309–313

67. Van Gelder RE et al (2004) Computedtomographic colonography comparedwith colonoscopy in patients at in-creased risk for colorectal cancer. Gas-troenterology 127(1):41–48

68. Sosna J et al (2003) CT colonographyof colorectal polyps: a metaanalysis.AJR Am J Roentgenol 181(6):1593–1598

69. Mulhall BP, Veerappan GR, Jackson JL(2005) Meta-analysis: computed tomo-graphic colonography. Ann Intern Med142(8):635–650

70. Halligan S et al (2005) CT colonogra-phy in the detection of colorectalpolyps and cancer: systematic review,meta-analysis, and proposed minimumdata set for study level reporting.Radiology 237(3):893–904

71. Akerkar GA et al (2001) Patient expe-rience and preferences toward coloncancer screening: a comparison of vir-tual colonoscopy and conventional co-lonoscopy. Gastrointest Endosc 54(3):310–315

72. Forbes GM, Mendelson RM (2000)Patient acceptance of virtual colonos-copy. Endoscopy 32(3):274–275

73. Copel L et al (2007) CT colonographyin 546 patients with incomplete colo-noscopy. Radiology 244(2):471–478

74. Marshall JB, Barthel JS (1993) Thefrequency of total colonoscopy andterminal ileal intubation in the 1990s.Gastrointest Endosc 39(4):518–520

75. Shah HA et al (2007) Factors asso-ciated with incomplete colonoscopy: apopulation-based study. Gastroenterol-ogy 132(7):2297–2303

76. Dafnis G et al (2005) Patient factorsinfluencing the completion rate incolonoscopy. Dig Liver Dis37(2):113–118

77. Cirocco WC, Rusin LC (1995) Factorsthat predict incomplete colonoscopy.Dis Colon Rectum 38(9):964–968

78. Correa P (1978) Epidemiology ofpolyps and cancer. Major Probl Pathol10:126–152

79. Williams AR, Balasooriya BA, DayDW (1982) Polyps and cancer of thelarge bowel: a necropsy study inLiverpool. Gut 23(10):835–842

80. Winawer SJ et al (1992) The NationalPolyp Study. Design, methods, andcharacteristics of patients with newlydiagnosed polyps. The National PolypStudy Workgroup. Cancer 70(5Suppl):1236–1245

81. Aldridge AJ, Simson JN (2001) Histo-logical assessment of colorectal adeno-mas by size. Are polyps less than10 mm in size clinically important? EurJ Surg 167(10):777–781

82. Ransohoff DF (2005) CON: Immediatecolonoscopy is not necessary in pa-tients who have polyps smaller than1 cm on computed tomographic colo-nography. Am J Gastroenterol 100(9):1905–1907; discussion 1907–8

83. Rex DK (2005) PRO: Patients withpolyps smaller than 1 cm on computedtomographic colonography should beoffered colonoscopy and polypectomy.Am J Gastroenterol 100(9):1903–1905;discussion 1907–8

84. Macari M et al (2004) Significance ofmissed polyps at CT colonography.AJR Am J Roentgenol 183(1):127–134

85. Zalis ME et al (2005) CT colonographyreporting and data system: a consensusproposal. Radiology 236(1):3–9

437