[Colonoscopy with computed tomography with volume reconstruction. The results and a comparison with endoscopy and surgery].

PURPOSE: Virtual CT colonoscopy is a novel technique whose diagnostic accuracy and clinical yield are currently investigated. Several studies have shown its capabilities in detecting colon and rectal cancers. We report the results of a preliminary experience with the volume rendering technique and compare CT colonoscopy with endoscopy and surgery.

MATERIAL AND METHODS: Our series consisted of 25 patients with colon cancer confirmed at endoscopy and/or surgery. All examinations were carried out with a spiral CT scanner Philips Tomoscan AVE1. Intestinal preparation was adequate in all patients, consisting in gas insufflated immediately before acquisition, after the injection of 100 mL iodinated contrast agent. All examination were performed in prone position using axial 5 mm slices with 5 mm table feed (pitch 1) and 2 mm reconstruction index; 120 kV and 200-250 mA were used. Images were transferred to a workstation (Easy Vision, Philips, release 4.2.1.1) for processing. We acquired multiplanar (MPR) and virtual endoscopic images with volume rendering; the selected threshold was -250 to -600 Hounsfield Units. Virtual endoscopic images could be obtained in 23 of 25 cases. The results of the radiological study were compared with endoscopic and surgical findings in 25 and 17 cases, respectively.

RESULTS: Endoscopy and surgery detected 46 lesions: 29 were malignant and 17 benign. Axial CT and MPR alone detected 35 lesions (76%), 29 of them malignant and 6 benign. The 11 benign lesions missed by axial CT ranged 6-8 mm in diameter. There were no false positives. CT colonoscopy alone detected 66 lesions, but 22 of them were false positive due to residual stool in the colon (21 cases) and to residual barium in the colon (1 case). Combining CT colonoscopy and axial and MPR images enabled to correct the false positive diagnoses made by CT colonoscopy alone and to decrease the false negative ratio of axial and MPR images. Forty-one of 44 lesions (93%) were detected. Thirty-seven lesions were found in the 17 surgical patients; 34 of them were correctly identified combining CT colonoscopy and axial and MPR findings, while endoscopy detected only 31 lesions. Thus, CT had 92% sensitivity, versus 83% of fiberoptic endoscopy.

CONCLUSION: In this preliminary experience volume rendering CT colonoscopy exhibited high sensitivity in detecting colon cancers, but their correct evaluation and characterization can be obtained if axial and MPR studies are combined. Further investigation and technological developments are necessary to define the yield of this new technique.