Introduction
Introduction
Cross sectional imaging techniques including magnetic resonance imaging (MRI) and computed tomography (CT) are increasingly being considered as alternative imaging modalities for colorectal screening [1 ]
[2 ]. Using thin section axial images and assigned software both techniques allow the generation of three-dimensional views of the colon, simulating those obtained with conventional colonoscopy. Since virtual colonography is relatively safe and minimally invasive it might become an attractive alternative to existing screening tests for colorectal cancer, the second most common cause of death from malignancy [3 ].
Currently MRI colonography is restricted by limited availability of scanners and high procedural costs. MRI as well as single-slice CT suffer from restrictions in spatial resolution and motion artefacts, which explain insufficient detection rates for masses smaller than 10 mm [2 ]. Single-slice CT requires several breath holds or a slice thickness exceeding 4 mm in order to scan the entire colon. A recent study comparing single-slice CT colonography and conventional colonoscopy suggests similar efficacy for detection of polyps 6 mm or more in diameter (82 - 91 %). However, restrictions in spatial resolution resulted in a low sensitivity for polyps smaller 6 mm (55 %) and frequent false positive findings [1 ].
Recently introduced multi-slice CT (MSCT) scanners represent a significant improvement in CT technology, combining high resolution thin slice imaging with high speed volume coverage [4 ]
[5 ]. MSCT has already been shown to improve the demonstration of colonic distention and to depict fewer respiratory artefacts [6 ]. The purpose of this prospective study was to evaluate the sensitivity and specificity of MSCT colonography in detection of colorectal polyps.
Materials and Methods
Materials and Methods
We prospectively studied 48 patients (20 women, 28 men, mean age 61.5 years) for colorectal cancer screening (low risk, no symptoms) or for evaluation of symptoms (including hematochezia, abdominal pain, weight loss, metastases of unknown primary) using a four slice CT scanner (Somatom Volume Zoom, Siemens AG, Forchheim, Germany). The study was approved by the ethics committee of the Medical Board Münster. From all patients a signed informed consent was obtained. The examination was performed after standard colonoscopy preparation (clear liquid diet and drinking of 4 L of polyethylene glycol solution the day prior to the scheduled examination), intravenous injection of butylscopolamin (Buscopan®, Boehringer, Ingelheim, Germany) and colonic distension with room air to maximum patient tolerance (average of 30 bulb compressions). Distension was assessed by scout CT images, when necessary further air insufflation was performed. We obtained images in prone and supine position [7 ]
[8 ] using a detector configuration of 4 × 1 mm, a table feed of 5 mm/rotation at 140 mAs, 120 kV, and a gantry rotation time of 500 msec. Slice thickness and reconstruction increment were 3 and 1.5 mm, respectively. Conventional colonoscopy was performed after the CT examination. The endoscopist was unaware of the CT findings. CT images were assessed by two blinded radiologists in a consensus decision using a separate workstation (Vitrea 1.1, Vital images, USA) with multiplanar and volume-rendering capabilities. The evaluation consisted of initial review of the magnified 2D transverse, sagittal, and coronare CT images in both supine and prone positions followed by endoluminal 3D images in problem solving situations. The transverse and reformatted coronal and sagittal 2D CT images werde displayed alongside the endoluminal images in a four-quadrant display format. Lesion size was determined with computed caliper measurement of the largest diameter on the 2D images (CT colonography) and with in vivo comparison to an open biopsy forceps (conventional colonoscopy). Size and location of polyps were recorded by using a segmental classification scheme for conventional and CT colonography (rectum, sigmoid, descending colon, left flexure, transverse colon, right flexure, ascending colon, and coecum). Using conventional colonoscopic findings as the gold standard, CT colonography findings were analyzed by direct polyp matching concerning size (size difference less than 5 mm) and location (same or adjacent segment). Insufficient air distension or bowel preparation was documented for each colonic segment. Examination, data transfer, and image interpretation took about 45 minutes per case.
Results
Results
Concerning polyps conventional colonography was normal in 33 patients. A total of 30 polyps and one carcinoma were identified in 15 patients. MSCT colonoscopy detected 23 polyps (77 %) and the carcinoma. All polyps greater than 10 mm were detected. 6 of 7 polyps from 9.9 to 5.1 mm and 14 of 20 polyps smaller than or equal to 5 mm were identified. 13 false positive findings affected 8 patients (Table [1 ]). In 4 of these patients conventional colonscopy revealed no polyp at all. A follow-up conventional colonoscopy in one patient with three true positive and one false positive findings revealed the “false positive” lesion as a true 4 mm polyp.
Discussion
Discussion
MSCT has multiple technical advantages over single-slice CT as already shown for CT angiography or lesion detection in the liver [9 ]
[10 ]. As published previously, MSCT improves colonic distention and reduces respiratory artefacts [6 ]. So far results of performance in detection of small colorectal polyps with multi-slice CT are still lacking.
In the largest prospective study to date by Yee et al. [12 ] the sensitivity of single-slice CT colonoscopy for polyp detection was 90 % for polyps 10 mm or larger, 80.1 % for polyps 5 - 9 mm and 59.1 % for polyps smaller than 5 mm (overall sensitivity 69.7 %). Our data show that MSCT colonography reliably identifies polyps greater than 5 mm. Furthermore, the detection rate of small colorectal polyps (Fig. [1 ]) in particular is improved compared to results previously published for single-slice CT and MRI colonography [1 ]
[2 ]
[12 ]
[13 ]
[14 ]
[15 ] and approaches the reported rate for detection of polyps smaller than 6 mm by conventional colonoscopy [11 ]. So far it is not clear whether these small lesions would justly performance of a conventional colonoscopy. Nevertheless we believe that virtual colonography should be able to identify these lesions with acceptable and comparable sensitivity to be competitive to conventional colonoscopy.
Although sensitivity for polyp detection is improved, a limitation of better size resolution with MSCT colonography is the number of false positive findings, with the majority of these lesion smaller than or equal to 5 mm (10 lesions < 6 mm in 5 patients). In our group this would result in an approximate rate of 10 % of unnecessary conventional colonoscopies, consequently raising cost, and radiation exposure. However, results of follow-up colonoscopy in one patient suggest that the specificity of MSCT colonography may be higher than reported. Indeed conventional colonoscopy misses 24 % of polyps smaller than 5 mm, 27 % of polyps 6 - 9 mm and 13 % of polyps 10 mm or more [11 ].
In cases of missed polyps, reevaluation of the CT-data set revealed colonic collapse or inadequate bowel preparation as the most common cause for false negative results, suggesting, that observer independent factors predominantly limit accuracy of CT colonography. However, the question whether MSCT improves colonic distention has not been subject of this study.
Multiple image display techniques are available to view CT colonography data. We found virtual flights as primary diagnostic method a time-consuming approach that requires more time than the 2D approach. According to our experience polyp detection is just as effective by using 2D as by using 3D images. However, endoluminal 3D images are useful in problem solving situations (for example differentiation of small polyps on folds from nodular folds), and for demonstration as well as documentation purposes.
MSCT offers a wide range of possible scan parameter combinations. The 4 × 1 mm detector configuration used in our study generates 3 mm slices with a slice sensitivity profile that is superior to 3 mm slices from a 4 × 2.5 mm detector configuration or single slice CT [4 ]. Whether the choice of the detector configuration influences image quality and polyp detection rate has to be addressed in future work. The need for ionising radiation is a potential disadvantage of CT colonoscopy. Decreasing the tube current or acquisition of broader slices with a wider detector configuration can reduce the effective radiation dose. Therefore optimisation of scanning protocols is necessary.
Furthermore, these results from a study of symptomatic and asymptomatic patients cannot be extrapolated to a screening population. Hence, the validity of MSCT colonography has to be established in multicenter studies in persons with average risk.
Table 1 Multi-slice CT colonoscopy compared to conventional colonoscopy: Results (no. of polyps, in brackets no of patients) according to polyp size and occurrence of cancer. Sensitivity was calculated on a lesion-per-lesion basis, specificity was a result of an analysis on a patient-per-patient basis.
Total True Positive False Positive False Negative Sensi- tivity Specificity
All sizes 30 23 (10) 13 (8) 7 (5) 77 % 78 %
≥ 10 mm 3 3 (3) 1 (1) 0 100 % 97 %
5.1 - 9.9 mm 7 6 (2) 2 (2) 1 (1) 86 % 94 %
< 5 mm 20 14 (5) 10 (5) 6 (4) 70 % 85 %
cancer 1 1 0 0 100 % 100 %
Fig. 1 Multi-slice CT colonoscopy: Axial (a ) and endoluminal (b ) view of a small 5 mm lesion (arrow) in the coecum displaying the polyp located on a fold.