| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, Mayo Clinic, 13400 E Shea Blve., Scottsdale, AZ
85259.
2 Department of Radiology, Mayo Clinic, Rochester, MN.
3 Department of Physiology and Biomedical Engineering, Mayo Clinic, Rochester,
MN.
4 Department of Health Sciences Research, Section of Biostatistics, Mayo Clinic,
Rochester, MN.
Received June 8, 2007;
accepted after revision August 16, 2007.
Address correspondence to C. D. Johnson.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Patients with known or suspected colorectal neoplasms were potentially eligible for participation, regardless of the presence or absence of gastrointestinal symptoms. Subjects ingested 21.6 g of barium in nine divided doses. CT colonography was performed in the standard fashion. Data sets were randomly evaluated by two of three experienced radiologists, with subsequent reanalysis of each data set after electronic stool subtraction at least 6 weeks later. Optical colonoscopy was performed after purgation and served as the reference standard.
RESULTS. One hundred thirty-one adenomatous neoplasms were
identified among 114 subjects. On a per subject basis, the sensitivity for
detecting adenomas 6–9 or 
10 mm in diameter ranged from 53% to 88%
and 84% to 93% without stool subtraction, respectively. By including stool
subtraction, these sensitivity estimates improved to 68% to 92% and 93% to
94%, respectively. Specificity ranged from 71% to 91% and 88% to 100% for
lesions 6–9 and 10 mm in size, respectively. Double reading
resulted in detection of 27 (87%) of 31 and 65 (96%) of 68 patients with
6–9 and 10 mm adenomas, respectively. With double reading, the area
under the receiver operating characteristic curve for large adenomas was
0.97.
CONCLUSION. In this increased-risk referral population, CT colonography in the non–cathartic-tagged colon without dietary restrictions compared favorably with optical colonoscopy.
Keywords: colon cancer • colon polyps • CT colonography • electronic stool subtraction
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
1 cm in diameter range from 55% to 92%,
albeit primarily in high-risk patient populations
[3–5].
The required cathartic bowel preparation is an important disincentive for
patients to undergo recommended CRC screening because the preparation can be
uncomfortable and inconvenient
[6].
A large study evaluating patient perceptions and preferences at CTC and
colonoscopy found that 72% of patients preferred CTC over colonoscopy, and a
significant number of patients would have the examination more frequently if
bowel preparation was not required
[7]. Thus, if the cathartic
preparation was not required, it is likely that more patients would undergo
screening at recommended intervals
[6]. In the cathartic-prepared
colon, oral contrast material is used to tag residual liquid stool in the
colon. This is done to improve examination specificity and to allow a complete
3D fly-through of the colon if the labeled liquid is subtracted
electronically. Labeling stool in the non–cathartic-prepared colon with
an oral contrast agent is possible, and feasibility studies have shown that
colorectal neoplasms can be discriminated from stool
[8–13].
In fact, in small cohorts the detection of large ( 1 cm) neoplasms can be
performed with performance characteristics similar to the cathartic-prepared
colon [8,
9,
12].
Electronic stool subtraction is performed in the noncathartic colon to remove the distraction of tagged stool particles and increase the conspicuity of colorectal lesions. Ideally, electronic stool subtraction would permit both 2D and 3D primary review of the colon data set, which has been shown to improve the detection of colorectal neoplasms [4, 14]. Although electronic stool subtraction has been described by a number of authors [9, 10], its effectiveness has not been rigorously evaluated.
The purpose of this study is to determine the performance of noncathartic CTC (without dietary restriction) in a patient population with a high prevalence of disease and to estimate the added value of electronic stool subtraction over interpretations without stool subtraction.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Patients received the following oral preparation before the CTC examination: 21.6 g of barium sulfate ingested over 48 hours in divided oral doses at each meal, at bedtime, and on rising the morning of the CT examination (four capsules each containing 600 mg of barium sulfate for nine consecutive doses) plus 250 mL of a 2.1% weight/volume (w/v) barium suspension 1 hour before the CTC examination (Scan C, Lafayette Pharmaceuticals). No laxatives were administered. There were no dietary restrictions. To keep preparation easy and patient compliance high, we did not require a low-fiber diet.
Seventy-five patients received glucagon, 1 mg, subcutaneously 10 minutes before CT acquisition, unless contraindicated or refused by the patient (routinely used for all CTC examinations at our institution). Thirty-nine did not receive glucagon because of conflicting appointments (usually laboratory work requesting a serum glucose test) or patient refusal. Patients were placed in the left lateral decubitus position for enema tip insertion and slow manual insufflation of approximately 2 L of carbon dioxide was performed (until the patient verbally indicated air administration had reached maximum tolerance). Only manual insufflation was used because this trial began before the benefits of mechanical insufflation were proven.
To ensure complete colon insufflation, a CT scout image was acquired before acquiring data with the patient in the supine and prone positions. Additional carbon dioxide was added (as tolerated by the patient) as needed to fully distend the colon. All examinations were performed using an 8-MDCT scanner (LightSpeed Ultra, GE Healthcare). Images were acquired using 1.25-mm collimation, table speed of 13.5 mm/s (pitch of 1.35), 1.25-mm reconstruction intervals, matrix of 512 x 512, field of view to fit, 70 mAs, 120 kVp, standard reconstruction algorithm, and 28-second breath-holds.
The image data were networked to an offline workstation with colonography software, Voxtool 5.4.46 (AW 4.1, GE Healthcare). The software allowed simultaneous viewing of prone and supine images and adjustments of the field of view. The CTC images were randomly distributed to two of three experienced radiologists to be read independently using a primary 2D (magnified axial and multiplanar reformatted images) search method. A primary 3D search method was not possible because of the presence of abundant stool in nearly all patients (unsubtracted data sets) or artifacts from the stool subtraction process (subtracted data sets). All three reviewers had interpreted more than 1,000 colonoscopy-verified examinations before this study. Experience in interpreting the non–cathartic-prepared colon was limited to approximately 20 examinations before beginning this study.
Each CTC examination was interpreted twice, first without stool subtraction
and subsequently with stool subtraction (with ready access to the
nonsubtracted data sets for review and problem solving). Each interpretation
(when performed by the same reader using a different search method) was
separated by at least a 6-week interval to reduce recall bias. Lesion
location, size, and observer confidence were noted for each abnormality.
Observer confidence for each lesion was rated on a scale of 1 to 5 (doubtful
to definite). Each data set (supine, prone) was evaluated sequentially, and
suspicious regions were compared between the two position data sets. Reviewers
were instructed to record only lesions that were 5 mm in diameter. Lesion
size was determined from 2D images using the greatest lesion diameter.
Interpretations were performed during times allocated for research without the
rush of a busy clinical practice. Reader 1 performed all lesion matching using
the annotated slice number, recorded colon segment, and size recorded by the
reader; colonoscopy report; pathology report; and videotaped colonoscopy
examination when required. Results are reported for each individual reviewer
(by polyp size) and for double reading (the combined reports of the two
individual readers).
Stool subtraction was performed using the algorithm described by Carston and coworkers [11], which was based on thresholding, morphologic operations, partial volume estimation, and knowledge of the CT scanner point-spread function (PSF). The processing created masks of voxels deemed to be entirely stool, air, or tissue on the basis of intensity and gradient values. An iterative classifier then classified unknown voxels by examining their neighbors and applying various decision rules. The masks were dilated using a structuring element created from PSF voxels with values greater than 0.05. The resulting stool mask was then convolved with the PSF, assigning a fractional amount of stool to voxels near the stool mask. The intensity of each voxel was adjusted by subtracting away its stool fraction times the difference between the original intensity and air (–1,000 H). Finally, morphologic processing was performed to remove any objects in the lumen that were entirely surrounded by air, which can arise from areas of poorly tagged stool, by setting their intensities to –1,000 H. The processing was performed entirely in 3D.
Reference Standard
Colonoscopy was performed in the standard fashion to examine the entire
colorectum after laxative purgation, with a mean time between CTC and
colonoscopy examinations of 4.1 days; 102 (89%) of 114 follow-up colonoscopies
were performed within 7 days of the CTC examination, although the interval
ranged from 0 to 100 days. All of the colonoscopy examinations were directly
performed or supervised by staff gastroenterologists or colorectal surgeons
experienced with this technique. Colonoscopists were blinded to the CTC
results. Colorectal inspection was complete to the level of the cecum for 96%
of the study subjects (n = 108), with intubation to the ileocolonic
anastomosis (n = 1), hepatic flexure (n = 3), transverse
colon (n = 1), and splenic flexure (n = 1) in the remaining
six subjects. The findings in these patients were compared between CTC and
colonoscopy only in the segments of the colon that were visualized
endoscopically.
The reference standard for our study was adenomatous polyps or cancer > 5 mm in size established by colonoscopy. Lesions containing both adenomatous and hyperplastic (mixed) pathologies were considered to be adenomatous polyps.
Lesions detected at CTC were matched manually with those found at
colonoscopy using both the colonoscopy and pathology reports, according to
established precedent [3].
Lesion size was determined by the pathology report unless the lesion was
removed in pieces. In these cases, the size estimate at colonoscopy was used.
If the lesion was within one colonic segment of the colonoscopic location and
within ± 50% in size, it was considered a match at CTC. The colon was
considered to have eight segments: cecum, ascending, hepatic, transverse,
splenic, descending, sigmoid, and rectum. Videotaped examinations were
reviewed if a convincing lesion ( 5 mm) was identified at CTC and not
reported on the colonoscopy report.
Statistical Analysis
Sensitivity and specificity were estimated for each reader, with 95% CIs,
for adenomatous polyps or cancers 6–9 and 10 mm in size. Double
reading included all detections from both independent interpretations
(conventional reading plus reading with stool subtraction). No discussion of
individual findings or consensus reporting occurred. Receiver operating
characteristic (ROC) curves were generated and area under the curves
calculated separately for each of the three reviewers and for the double
reading. In this analysis, the judgment of adenoma presence was graded by the
radiologists as not possible, doubtful, somewhat doubtful, conservatively
confident, moderately confident, and highly confident. For adenomas not
detected by a radiologist, an answer of "not possible" was
assigned. For the double reading, the higher of the two reviewer confidences
was used. Specificity was determined on a per patient basis and reported for
two groups of patients: those without a proven adenoma > 5 mm and those
without an adenoma 10 mm.
Stool labeling, effectiveness of stool subtraction, and stool subtraction artifacts were judged and recorded by one of the two radiologists randomly assigned to read each patient examination. Examination quality was recorded for each of four colon segments: ascending cecum, transverse cecum, descending sigmoid, and rectum. The amount of stool, stool labeling, percent subtracted, wall erosion, fold erosion, polyp distortion, islands of residual stool, and halos from subtraction were all rated using a 4-point scale for each of the four colonic regions.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
5 mm in diameter. Adenomatous lesions > 5 mm were
located in the cecum in 14 (11%) of 131, ascending in 17 (13%), hepatic
flexure in nine (7%), transverse in 10 (8%), splenic flexure in three (2%),
descending in nine (7%), sigmoid in 41 (31%), and rectum in 28 (21%) (Figs.
1A,
1B and
2A,
2B). There were no reported
adverse events.
|
|
|
|
10 mm in diameter,
respectively, without the use of stool subtraction
(Table 1). Combining
unsubtracted and subtracted displays, the sensitivity for adenomas 6–9
and 10 mm was 64–77% and 81–89%, respectively. Using double
reading (two readers using both unsubtracted and subtracted data sets) the
sensitivity was 80% and 87% for adenomas 6–9 and 10 mm in diameter,
respectively (Table 1).
|
On a per patient basis, the sensitivity for detecting a patient with any
adenoma 6–9 or 10 mm in diameter ranged from 53% to 88% and 84% to
93%, respectively, in the absence of stool subtraction. With the addition of
stool subtraction, the per-patient sensitivity estimates for adenomas
6–9 and 10 mm in diameter ranged from 68% to 92% and 93% to 94%,
respectively.
The number of additional lesions found using stool subtraction ranged from
1 to 8 and 0 to 6 for 6–9 and 10 mm adenomas, respectively,
resulting in sensitivities improving 3–28% for medium-sized polyps and
0–9% for larger polyps. Reader 2, but not reader 3, significantly
improved detection of 6–9 mm adenomatous polyps using the subtracted
images (p < 0.05). Reader 1 tended to detect more polyps 6–9
mm in size with stool subtraction (13/28 vs 18/28 polyps detected), but these
findings did not reach statistical significance. Double reading resulted in
detection of 27 (87%) of 31 and 65 (96%) of 68 of patients with 6–9 and
10 mm adenomas, respectively (Table
2).
|
The observed specificity for detecting adenomas 6–9 mm in diameter without stool subtraction ranged from 84% to 95%. Specificity for larger adenomas ranged from 94% to 100% without stool subtraction. Specificity tended to be slightly less using stool subtraction, but these changes were not statistically significant (p > 0.05). Double reading resulted in slight but insignificant reductions in specificity for medium-sized and large lesions as well (Table 2).
Area under the ROC curve analysis for readers 1, 2, and 3 for detecting
6–9 mm adenomas without and with stool subtraction was 0.75, 0.75, and
0.88 and 0.79, 0.88, and 0.90, respectively. With double reading the area
under the curve was 0.89. For the detection of 10 mm adenomas, the area
under the curve for readers 1, 2, and 3 without and with stool subtraction was
0.94, 0.91, and 0.97 and 0.90, 0.90, and 0.95, respectively. The area under
the curve for double reading was 0.97
(Table 3).
|
Stool labeling was judged to be > 90% labeled in all colon segments in 47% of patients, 50–90% labeled in at least one segment in 30%, and less than 50% labeled in at least one segment in 23% of patients. Overall minimal (lowest score for the eight colon segments) stool subtraction was judged to be > 90% stool subtraction in 45%, 50–90% subtraction in 30%, and less than 50% subtraction in 25% of patients. The severity of stool subtraction artifacts is displayed in Tables 4 and 5.
|
|
Mean reading times for reviewers 1, 2, and 3 for conventional interpretations and those including stool subtraction were 12.5, 14.2, 12.4 and 11.7, 14.9, 13.0 minutes, respectively.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
10 mm) adenomas were detected with a sensitivity of
64–77% and 81–89%, respectively. Double reading and electronic
stool subtraction tend to improve the sensitivity for most readers, although
their benefit is most helpful for improving the detection of moderate-sized
polyps. Remarkably, we found that the sensitivity of noncathartic CTC for
detecting large adenomas was relatively high even without stool subtraction.
The sensitivity of one reader significantly improved with the addition of
stool subtraction for detecting moderate-sized polyps, whereas another reader
tended to detect more medium-sized polyps using electronic stool
subtraction.
Although the sensitivity of the examinations increased with the addition of
stool subtraction, the specificity of the examinations decreased (especially
for the detection of moderate-sized neoplasms). Overall, performance as
estimated by area under the ROC curve showed only modest improvement using
stool subtraction for the detection of both moderate-sized and large
neoplasms. The area under the curve was 0.89 and 0.97 for neoplasm groupings
of 6–9 and 10 mm, respectively, when double reading was used.
These data indicate that in a high-prevalence group, CTC in the non–cathartic-prepared colon is similar in performance to both conventional CTC and optical colonoscopy. Because the cathartic bowel preparation, which is viewed widely as inconvenient and uncomfortable, is a major disincentive for patients to undergo CRC screening [7], this represents a major advance in the tools available for direct colorectal imaging. The barium capsules that were administered over a 48-hour period were tolerated well by patients. Barium capsules were used because they were believed to be a more convenient and portable delivery method. Few complaints occurred as a result of the liquid barium that was given 1 hour before the procedure to label any residual untagged fluid in the colon. The lack of any requirements for dietary changes was also viewed positively by many patients. A low-fiber diet was not required in an effort to maintain preparation simplicity and compliance with preparation instructions.
At present, stool subtraction is technically difficult because stool can be heterogeneously tagged, colon transit is variable, the composition of stool changes as it progresses through the colon, and partial volume effects must be accounted for at stool–air and stool–tissue interfaces. It is imperative that polyps and normal colorectal mucosa are not altered by the subtraction process.
The method of stool subtraction used was described by Carston and coworkers [11]. This method of stool subtraction remains highly dependent on adequate stool labeling. Approximately half of all patients had ideal labeling, with 100% of colon segments containing well-tagged stool. About 25% had suboptimal stool labeling in at least one colon segment. This variability is likely due to differences in diet and bowel habits that were not controlled. Further research to improve overall stool labeling would be helpful. Stool subtraction artifacts considered to be least troubling included wall erosion (none or mild in 82%) and polyp distortion (none or mild in 83%). Fold erosion, islands of residual stool, and halos from subtracted stool were more prevalent artifacts and should be the focus of further subtraction algorithm development.
It is likely that the reduced specificity encountered in this study when stool subtraction was used was due to stool that had been altered in attenuation or form to resemble a soft-tissue polyp. On the other hand, the added sensitivity for polyp detection using stool subtraction is likely due to less distraction for the radiologist when mental subtraction is required for every intraluminal filling defect. It is interesting to note that some readers appear to benefit from stool subtraction more than others. In our study, reader 2 identified eight additional neoplasms (five patients), with only a specificity reduction of two patients (Tables 1 and 2). Reader 3, however, detected only a single additional neoplasm, but the specificity was reduced by seven patients. Similarly, reader 1 identified five additional neoplasms in two patients at the cost of lowered specificity of six patients. It may be that stool subtraction is more helpful to some readers than others.
Thorough stool subtraction is needed if a primary 3D approach to interpretation is used. Residual stool and artifacts would create filling defects that must be individually characterized. Large numbers of filling defects create an insurmountable task at 3D review. In this study, only primary 2D interpretation was used. This approach allows the reader to visualize the internal attenuation of filling defects and quickly determine if they are soft tissue or contain internal air or barium (findings typical of stool).
Double reading in several studies has been helpful to reduce interobserver
variability and improve overall performance of the examination
[15–17].
In this study, interobserver variability was only modest; however, two of
three readers benefited from double reading for the detection of neoplasms
6–9 mm in diameter. Double reading added little for the detection of
large neoplasms ( 10 mm diameter) despite a doubling of the work effort.
Computer-aided diagnosis is another tool that can be used as a double reading
substitute. No reports exist of software for this application in the
non–cathartic-prepared colon; however, it is likely that stool
subtraction artifacts must be reduced substantially before it would be
effective.
Extrapolation of these findings from a population with a high prevalence of disease to a screening population would not be justified. Because most patients had a colorectal lesion, radiologists approached these studies differently than if the pretest probability for a large lesion had been low. Nonetheless, data from the present study are encouraging and warrant further testing in a true screening population with a lower prevalence of colorectal neoplasms.
Other studies have been performed evaluating the non–cathartic-prepared colon. Stool tagging using widely available barium was first reported by Callstrom and coworkers [8]. Others have reported successful tagging using iodinated contrast material [9, 10]. Efforts to reduce fecal volumes have used either a low-residue diet [12, 13] or a partial preparation in conjunction with fecal tagging [10]. Dachman et al. [18], reporting on patients prepared with a low-fiber diet, found that the major differences in the cathartic unprepared versus prepared colon were the amount of mucosal surface covered by stool, size of retained stool, and number of segments containing stool. Iannaccone et al. [9] reported on CTC without cathartic preparation or stool subtraction in 203 patients after the administration of 200 mL of oral iodinated contrast material (diatrizoate meglumine and diatrizoate sodium with an iodine concentration of 370 mg/mL). This agent often results in diarrhea when this amount of contrast material is ingested. Using this technique, the average sensitivity for the detection of polyps 8 mm or larger was 95.5%. To our knowledge, we report the largest experience to date in a population without dietary restriction and without agents that would induce catharsis. Our results are similar to those of Iannaccone et al. except that our specificity estimate was slightly lower, likely due to untagged stool or artifacts from the stool subtraction process. We believe that the lack of catharsis and dietary change is an important factor in increasing patient compliance with CRC screening recommendations.
In summary, in a patient population with a high prevalence of disease, CTC in the non–cathartic-prepared colon without dietary restrictions compares favorably with both cathartic-prepared CTC and optical colonoscopy for the detection of colorectal neoplasms 6 mm or larger. Additional studies in a screening population are needed to determine the performance characteristics of this technique in a lower-prevalence setting. Electronic stool subtraction does improve sensitivity of the examination, but often at the expense of specificity. Further development of stool tagging and subtraction techniques is needed to allow 3D image review and application of computer-assisted diagnosis.
Acknowledgments
We acknowledge Lynn Wilson for her assistance in patient recruitment and
data entry, Robert Wentz for his role in assisting with electronic stool
subtraction, Paul Limburg for manuscript review, and Debora Shreve and Taki
Ernst for manuscript preparation.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  S. Mahgerefteh, S. Fraifeld, A. Blachar, and J. Sosna CT Colonography With Decreased Purgation: Balancing Preparation, Performance, and Patient Acceptance Am. J. Roentgenol., December 1, 2009; 193(6): 1531 - 1539. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. D. Johnson CT Colonography: Coming of Age Am. J. Roentgenol., November 1, 2009; 193(5): 1239 - 1242. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E J HALL and D J BRENNER Cancer risks from diagnostic radiology Br. J. Radiol., May 1, 2008; 81(965): 362 - 378. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
