| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Experimental Studies |
1 From the Departments of Intestinal Imaging (S.A.T., S.H., C.I.B., N.F.), Pathology (I.C.T.), Surgery (K.K.), and Endoscopy (B.P.S.) and the Cancer Research UK Colorectal Cancer Unit (S.H., B.P.S., W.A.), St Mark’s Hospital, Northwick Park, Watford Rd, Harrow, Middlesex HA1 3UJ, England; and General Electric Medical Systems, Slough, England (P.R.M.). Received May 10, 2002; revision requested June 14; final revision received February 5, 2003; accepted March 3. Address correspondence to S.H. (e-mail: s.halligan@ic.ac.uk).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: A colectomy specimen containing 117 polyps of different sizes was insufflated and imaged with a multi–detector row CT scanner at various collimation (1.25 and 2.5 mm), pitch (3 and 6), and tube current (50, 100, and 150 mA) settings. Two-dimensional multiplanar reformatted images and three-dimensional endoluminal surface renderings from the 12 resultant data sets were examined by one observer for the presence and conspicuity of polyps. The results were analyzed with Poisson regression and logistic regression to determine the effects of scanning parameters and of specimen orientation on polyp detection.
RESULTS: The percentage of polyps that were detected significantly increased when collimation (P = .008) and table feed (P = .03) were decreased. Increased tube current resulted in improved detection only of polyps with a diameter of less than 5 mm. Polyps of less than 5 mm were optimally depicted with a collimation of 1.25 mm, a pitch of 3, and a tube current setting of 150 mA; polyps with a diameter greater than 5 mm were adequately depicted with 1.25-mm collimation and with either pitch setting and any of the three tube current settings. Small polyps in the transverse segment (positioned at a 90° angle to the z axis of scanning) were significantly less visible than those in parallel or oblique orientations (P < .001). The effective radiation dose, calculated with a Monte Carlo simulation, was 1.4–10.0 mSv.
CONCLUSION: Detection of small polyps (<5 mm) with multi–detector row CT is highly dependent on collimation, pitch, and, to a lesser extent, tube current. Collimation of 1.25 mm, combined with pitch of 6 and tube current of 50 mA, provides for reliable detection of polyps 5 mm or larger while limiting the effective radiation dose. Polyps smaller than 5 mm, however, may be poorly depicted with use of these settings in the transverse colon.
© RSNA, 2003
Index terms: Colon, CT, 75.1211 • Colon neoplasms, 75.311 • Computed tomography (CT), technology
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Whereas the sensitivity of CT colonography for the detection of polyps larger than 1 cm exceeds 90% (2,4), polyps of 5 mm and smaller are less frequently detected, with reported sensitivities ranging from only 8% to 60% (5–9). Most investigators in these previously reported studies used single–detector row scanners, generally with a collimation of either 3 or 5 mm, although some attempted to simulate thin-section multi–detector row CT colonography (10). With the introduction of multi–detector row scanners, enhanced detection of smaller lesions has become possible, but clinical studies thus far have focused on the detection of large polyps (11,12). The purpose of this study was to investigate the effects of orientation, collimation, pitch, and tube current on the detection of adenomatous polyps at multi–detector row CT colonography and to determine the optimal scanning parameters for colon screening.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
At surgery, a standard subtotal colectomy was performed with ileorectal anastomosis, leaving a rectal stump. The colectomy specimen was immediately cleaned in the operating room with 0.9% saline solution, and as much residual luminal fluid as possible was drained. The distal sigmoid lumen was then securely closed with 2/0 prolene sutures to ensure an airtight seal. An 18-F Foley urinary catheter was inserted through the terminal ileal remnant, and the retention balloon was fully inflated with water. The catheter was withdrawn to ensure that the inflated balloon was positioned snugly against the ileocecal valve, and then the catheter was sutured in place. The colon was measured and divided into four segments (sigmoid, descending, transverse, and ascending) of approximately equal length. The junction between these segments was identified by 2/0 prolene sutures stitched through the outer colonic wall. Each suture line was marked with three superficial surgical skin staples (Proximate PX35; Ethicon Endo-Surgery, Cincinnati, Ohio) at intervals around the circumference of the colon; the staples provided radiopaque markers for segment identification during subsequent CT analysis. An insufflation bulb was attached to the Foley catheter and the specimen was distended gently to ensure that no leakage of air would occur during transfer to the CT suite. These procedures were performed by one of the surgeons who had performed the colectomy (K.K.).
In the CT suite, the specimen was placed in a plastic container containing 20 L of 0.9% saline solution to which had been added 60 mL diatrizoate meglumine containing 370 mg of iodine per milliliter (Gastrografin; Schering Health Care, Burgess Hill, West Sussex, England), providing an average attenuation of 36 HU (SD, ±6 HU), similar to that of abdominal tissue (14). Next, the colon specimen was insufflated with room air until all segments were dilated to at least 5 cm external diameter. The specimen then was arranged in the container to mimic the geometric relationships between each colonic segment in vivo (Fig 1) and was submerged in the saline bath by means of a cardboard sheet weighted with 1-L bags of normal saline solution.
|
Pathologic examination.—After CT scanning, the resected specimen was examined by an experienced gastrointestinal histopathologist. Each of the four colonic segments was identified by means of the external sutures and staples. The average length of the segments was 12.9 cm (range, 11.5–14.6 cm). The specimen was opened and examined under a handheld magnifying lens. Polyps identified were measured in situ with fine calipers; polyp size was given as the diameter of the polyp at its largest point. A total of 117 polyps ranging in size from 1 to 15 mm were counted in the four segments examined (Table 1). Polyps located on the suture line were included in the count.
|
|
|
Statistical Analysis
Statistical analysis was performed by using statistical software (Stata version 7.0; Stata, College Station, Tex). The number and conspicuity of polyps found in each segment at CT colonography with each combination of scanning parameters were compared with the findings at histologic evaluation of the specimen as follows: The effect of each of the three scanning parameters (collimation, pitch, and tube current) on polyp detection was assessed, and any interactions between them were investigated by using Poisson regression with robust standard errors to account for the effect of repeated observations. Data then were re-analyzed for two size groups (polyps with diameters <5 mm and polyps with diameters 
5 mm) to determine the effect of polyp size. To assess the effect of colonic segment orientation with regard to the z axis at scanning, we grouped data for polyps in the ascending colon with those for polyps in the descending colon (ie, polyps in both segments longitudinal to the z axis, ie, at an angle of 0°) and compared the percentage of polyps detected in these two segments with the percentages detected in the transverse colonic segment (at a 90° angle to the z axis) and the sigmoid colonic segment (at a 45° angle, oblique to the z axis). Because the numbers and sizes of polyps present in the specimen differed among orientations, the percentage of polyps detected was calculated separately for each orientation. For logistic regression analysis, the percentage of polyps detected was expressed as a binary variable (ie, <50% or 50%). The dividing line of 50% was chosen because it created two groups of polyps approximately equal in number. In the overall analysis of polyp detection in the 12 data sets, standard errors again were used to take account of repeated observations. To account for differences in the distribution of polyp sizes among orientations, polyp size (<5 or 5 mm) was included as a variable in the regression analysis. Results were expressed, where appropriate, as odds ratios (ie, the odds of detection for one variable as opposed to another variable). A P value of less than .05 was considered to indicate a statistically significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
When data on polyp detection were analyzed according to polyp size (<5 or 5 mm), there was a statistically significant interaction between the size of polyp detected and the collimation and table feed, and less significant interaction with tube current (Table 2). Furthermore, reduced table feed and increased tube current had a positive effect on detection only of polyps smaller than 5 mm (Table 2, Figs 8–10). The effect of both collimation and table feed was the same in the results of regression analysis. Thus, polyps smaller than 5 mm were optimally depicted with a collimation of 1.25 mm, pitch of 3, and tube current of 150 mA, while larger polyps were adequately depicted with a collimation of 1.25 mm combined with any of the pitch and tube current settings used.
|
|
|
|
|
|
|
|
5 mm) revealed that four (67%) of the six polyp sizes (1, 2, 3, 4, 5, and 6 mm) present in the longitudinal orientation were smaller than 5 mm, as were three (60%) of five sizes (1, 2, 3, 5, and 6 mm) in the transverse orientation and two (22%) of nine sizes (1, 3, 5, 6, 7, 9, 10, 11, and 15 mm) in the oblique orientation—indicating an uneven distribution of polyp sizes across the three orientations studied. To account for this effect, polyp size was considered in the analysis for interactions with orientation; collimation and table feed were subsequently included in this analysis because they had been shown to affect the overall polyp detection rate. The adjusted polyp detection rate was approximately three times lower in the oblique orientation because fewer small polyps were present in this segment than in the others (Table 4). There was no significant interaction between orientation or segment and polyp size, however; the differences between orientations were equivalent for polyps of differing sizes (although the transverse colon contained no polyps larger than 6 mm). Adjustments for table feed and collimation had little effect on the odds ratio of polyp detection in a particular orientation. The number of polyps observed was significantly greater in all orientations, however, with a collimation of 1.25 mm and a table feed of 3.75 mm—a finding that is consistent with the results obtained when orientation was not factored into the analysis. Again, the probability of observing more than 50% of the existing polyps of any given size was much lower for the transverse orientation than for the other two. The detection rate for polyps of 2–4 mm in each orientation, averaged across all 12 data sets, is shown in Figure 14, which illustrates the poor performance in the transverse orientation.
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
We found that both collimation and pitch had a significant effect on the polyp detection rate, which, when unadjusted for size, was 50% higher at 1.25 mm collimation than at 2.5 mm and 30% higher at pitch 3 than at pitch 6. This effect was most marked for polyps of less than 5 mm, for which a small but significant improvement in detection also occurred with increased tube current. However, for polyps of 5 mm or larger, the effect of pitch on detection was insignificant, and the benefit of decreased collimation, although significant, was small, resulting in a 7% improved detection rate. This finding is important because CT colonography, to be as effective as endoscopic colonoscopy, must depict polyps reliably at the 5-mm threshold.
Specimen orientation also seems to play an important part in polyp conspicuity. Interestingly, even after correcting for the different distribution of polyp sizes across segments, we found that significantly fewer polyps had been detected in the transverse segment. This finding has important implications, because it was suggested previously that CT colonography could be used to examine the part of the colon that is reachable with a flexible endoscope (15)—largely, the transverse colon. A possible explanation for the decrease in conspicuity in the transverse orientation is the worsening resolution along the z axis compared to that in the x-y plane. For example, the measurement of a voxel in the x-y plane will equal the field of view divided by 512—usually, about 0.6 mm. In contrast, the z-axis dimension is dependent on section thickness—for example, 1.25 or 2.5 mm, depending on the scanning protocol. Furthermore, rippling artifacts tend to occur along the z axis because of the interpolation of inconsistent data resulting from the angulation of the colon with respect to the longitudinal axis. Whiting et al (16) found increased rippling artifacts with increasing angle relative to the z axis and increasing pitch, such that the depiction of sessile polyps decreased. Although we did not quantify the effect of this artifact in the present study, it might have degraded the conspicuity of sessile polyps in the transverse orientation. The best detection rate occurred in segments running along the z axis (ascending and descending colon), and although the odds of detection were three times lower in the oblique (sigmoid) orientation, the difference in detection rate was not statistically significant.
Investigators in previous phantom studies used single–detector row technology and artificial or animal colons (14, 16–18). Nonetheless, our findings are in general agreement with previously published findings. In evaluating artificially created polyps in a pig colon, Dachman et al (14) found that conspicuity decreased with increasing pitch and collimation, particularly on three-dimensional reconstructions. In contrast to our findings, phantom orientation did not affect polyp conspicuity, possibly because polyps were generally spherical and had a diameter of at least 3 mm. Investigators in another study of pig colon assessed the prevalence of adverse CT artifacts over a range of scanning parameters and found that smoothing became more evident with increasing collimation, whereas stair-step artifacts and longitudinal distortion were more dependent on increasing pitch (17). However, when they used collimation and pitch combinations similar to those used in our study, those investigators found that such artifacts were generally unimportant. Other researchers have used gas-filled plastic tubing to simulate the colon (16,18). Using this approach, with a collimation of 5 mm and pitch of 2, Beaulieu et al (18) found significantly worse detection for 2.5-mm polyps when the phantom was positioned parallel to the CT gantry—that is, in the same orientation as the transverse colonic segment in the present study. Our results suggest that this relative disadvantage persists even when narrower collimations and multi–detector row technology are used.
Although the general effects of collimation and pitch in the present study are consistent with effects found in previous phantom studies, the absolute level of detection of small polyps was lower. Only when a collimation of 1.25 mm was combined with a pitch of 3 did the rate of detection of 2–4-mm polyps increase beyond 50%. In contrast, both Dachman et al (14) and Beaulieu et al (18) found that polyps of this size were depicted, albeit faintly, with much wider collimations (7 and 5 mm, respectively). Similarly, our level of detection of small polyps with a collimation of 2.5 mm was lower than detection rates in other clinical studies, in which the rate of detection of polyps 5 mm or smaller was more than 50% with a collimation of 3 mm or more (2,4). To achieve good depiction of polyps without substantial distortion, the effective section thickness should be less than half of the polyp size. Indeed, in our own clinical practice we have frequently seen 2-mm polyps very well depicted with a collimation of 2.5 mm and pitch of 6. The relatively low level of depiction in this study is likely due to the particular colectomy specimen used. Unlike artificially produced phantom polyps and most naturally occurring adenomatous polyps, the tiny polyps in polyposis coli tend to form a shallow carpet throughout the entire colon, with larger polyps intermittently interspersed. Although our observer was aware that the specimen was from a subject with familial adenomatous polyposis, he was unaware of the number, size, and location of the polyps, and he detected 1-mm polyps in only one of the 12 data sets. If 1-mm polyps had been included in our polyp grouping for analysis, the rate of detection of polyps of 1–4 mm would have been only 17% even with the most sensitive parameter combination. The tiniest polyps, which appear as a fine mucosal nodularity, are essentially beyond the resolution capabilities of currently available multi–detector row CT scanners; such lesions, however, are of doubtful clinical importance. Similarly, many 2- and 3-mm polyps were flat and therefore lacked conspicuity. In contrast, our detection rate of polyps of 5 mm or larger approached 100%, with only sporadic failures that were due to perceptual errors and not to any shortcomings of the imaging technique. Interestingly, although the observer was experienced in the clinical use of two-dimensional images for interpretation and of three-dimensional images for problem solving, he relied almost solely on three-dimensional images when searching for the smallest polyps, which seemed more readily disclosed by the three-dimensional images. Nevertheless, the observer repeatedly missed one of three 6-mm polyps, which was well depicted on images from all 12 data sets but was partially concealed by a fold. In contrast, a nodular thickened fold that was observed on images of the ascending colon and identified as a 20-mm flat adenomatous lesion, and another feature observed on images of the transverse colon and identified as a 10-mm flat lesion, were not actually present in the specimen. These false-positive findings may have resulted from inadequate distention of or residue retained in the colon specimen.
There are few published studies of the use of multi–detector row CT for polyp detection. The results of a study performed by Hara et al (12) showed no significant difference between multi–detector row CT and single–detector row helical CT in this regard, but only polyps of 10 mm or greater were studied. Rogalla et al (10) investigated the theoretical effect of changing collimation on polyp detection by reconstructing 1-mm data sets at various section thicknesses; sensitivity for polyps measuring 3–5 mm decreased from 96% at a 1-mm section thickness to 74% at a 5-mm section thickness.
We detected a much greater number of small polyps (<5 mm) at a collimation of 1.25 mm, a table feed of 3.75 mm (pitch of 3), and a tube current of 150 mA. Unfortunately, the associated dose penalty is prohibitive, with an effective dose of 20.0 mSv for combined supine and prone scanning. Even if polyps are adenomatous, it may be safer to leave them in situ, because their malignant potential is likely outweighed by the risks associated with polypectomy (3). Furthermore, the scanning times associated with such a protocol lead to breathing artifacts. In the context of a screening program, a compromise between detection, dose, and image quality will be necessary. The results of our study suggest that a collimation of 1.25 mm, table feed of 7.5 mm (pitch of 6), and tube current of 50 mA will not degrade the detection of polyps of 5 mm or greater under ideal conditions; the effective dose for combined supine and prone scans with these parameters is 3.4 mSv, which is lower than that incurred during the acquisition of a standard abdominopelvic CT scan (6–24 mSv) (19) or barium enema study (6.4 mSv) (20). Furthermore, one could consider increasing collimation to 2.5 mm, because a collimation of 1.25 mm resulted in the detection of only 7% more polyps of 5 mm or larger; even a modest increase in detection rate, however, could be clinically important in the context of a screening program. Although we used dilute contrast material with an attenuation of 36 HU to simulate pericolonic tissues, it may be that differing body habitus would require a tube current greater than 50 mA to reproduce our results in vivo. Investigators nonetheless have found no loss of conspicuity in vivo at an effective tube current of 70 mA (21), and some have reported the successful use of a current of 10 mA (22).
Practical applications: We have shown that the detection of small polyps (<5 mm) in an optimally cleansed and distended human colon with the use of multi–detector row CT colonography is highly dependent on collimation, table feed, and, to a lesser extent, tube current. Polyps of 1 mm, however, are not well depicted with this technique. In contrast to the depiction of small polyps, that of larger polyps is less dependent on technical factors such as imaging parameters. A protocol that includes a collimation of 1.25 mm, pitch of 6, and tube current of 50 mA is a reasonable compromise between the concerns for patient dose and for polyp detection; polyps of 5 mm or larger are reliably detected with these parameters. Increases of collimation and of table feed to 2.5 and 15 mm, respectively, would reduce the occurrence of respiratory artifacts and produce only a small decrease in the rate of detection of polyps of 5 mm or larger. Small polyps in the transverse colon may be much less well depicted with multi–detector row CT colonography than those in other colonic segments.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
See also Science to Practice in this issue.
Author contributions: Guarantors of integrity of entire study, S.A.T., S.H.; study concepts, S.H.; study design and literature research, S.A.T., S.H.; experimental studies, S.A.T., I.C.T., N.F., B.P.S.; data acquisition, S.A.T., P.R.M., I.C.T., N.F.; data analysis/interpretation, S.A.T., S.H., P.R.M.; statistical analysis, S.A.T., S.H., P.R.M.; manuscript preparation, S.A.T.; manuscript definition of intellectual content and manuscript revision/review, S.A.T., S.H., C.I.B., W.A.; manuscript editing, S.H.; manuscript final version approval, all authors
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Related Article
This article has been cited by other articles:
|
|
 |
|
  S. H. Kim, J. M. Lee, C.-I. Shin, H. C. Kim, J.-G. Lee, J. H. Kim, J. Y. Choi, H. W. Eun, J. K. Han, J. Y. Lee, et al. Effects of Spatial Resolution and Tube Current on Computer-aided Detection of Polyps on CT Colonographic Images: Phantom Study Radiology, August 1, 2008; 248(2): 492 - 503. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Jensch, A. H. de Vries, J. Peringa, S. Bipat, E. Dekker, L. C. Baak, J. F. Bartelsman, A. Heutinck, A. D. Montauban van Swijndregt, and J. Stoker CT Colonography with Limited Bowel Preparation: Performance Characteristics in an Increased-Risk Population Radiology, April 1, 2008; 247(1): 122 - 132. [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]
|
|
|
|
 |
|
 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 SLATER Alteration in bowel habit Imaging, December 1, 2006; 18(4): 208 - 217. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Graser, B. J. Wintersperger, C. Suess, M. F. Reiser, and C. R. Becker Dose reduction and image quality in MDCT colonography using tube current modulation. Am. J. Roentgenol., September 1, 2006; 187(3): 695 - 701. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A S Lowe and C L Kay Recent developments in CT: a review of the clinical applications and advantages of multidetector computed tomography Imaging, June 1, 2006; 18(2): 62 - 67. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. T. Johnson, C. D. Johnson, J. G. Fletcher, R. L. MacCarty, and R. L. Summers CT Colonography Using 360{degrees} Virtual Dissection: A Feasibility Study Am. J. Roentgenol., January 1, 2006; 186(1): 90 - 95. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. G. Mang, C. Schaefer-Prokop, A. Maier, E. Schober, G. Lechner, and M. Prokop Detectability of Small and Flat Polyps in MDCT Colonography Using 2D and 3D Imaging Tools: Results from a Phantom Study Am. J. Roentgenol., December 1, 2005; 185(6): 1582 - 1589. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
|
), which is given with 95% CI, is the odds of detection at a collimation of 1.25 mm in comparison with the baseline odds of detection at a collimation of 2.5 mm. The odds of polyp detection were significantly greater with a 1.25-mm collimation for both polyp groups, although the effect was much larger for polyps of less than 5 mm.
