Imaging reconstruction software that "unfolds" the colon wall flat could greatly shorten reading time for CT colonography. Researchers from Argentina have found no significant differences in diagnostic accuracy between the "filet view" and conventional CTC in the first study comparing both virtual colonoscopy techniques.
 |
| Filet view software provides several displays for simultaneous comparison of conventional axial, 2D, and 3D CT colonoscopy. (Provided by Diagnóstico Maipú) |
Just like conventional CTC, unfolded haustra or filet view CTC is noninvasive and more comfortable method for patients than colonoscopy. It also allows the evaluation of the total colon lumen in nearly all cases, does not require sedation or anesthesia, and poses very low risk, said principal investigator Dr. Patricia Carrascosa, director of research at the Diagnóstico Maipú imaging center in Buenos Aires.
The filet view technique provides further advantages.
"Data are easier and faster to reproduce and analyze with filet view than with conventional virtual colonoscopy," she said.
Carrascosa and colleagues imaged 23 patients with suspected colon cancer who had signs of rectal bleeding, constipation, or weight loss. All patients underwent conventional CTC, unfolded haustra CTC, and conventional optical colonoscopy.
The investigators found no significant differences in diagnostic yield between conventional and filet view CTC analysis in correlation to colonoscopy. The filet view technique significantly reduced the average time of interpretation for CTC, however, from 15 minutes to eight.
The Argentine researchers presented their findings at the 2005 RSNA meeting in Chicago.
Patients underwent all imaging studies one day after conventional preparation. Radiologists read blinded studies and organized results into groups of normal (A) and pathologic (B) findings. The investigators subdivided group B findings according to size: lesions below 5 mm, lesions between 5 and 9 mm, and lesions larger than 9 mm.
Colonoscopy detected 35 elevated lesions. Both conventional and filet-view CTC respectively found 29 of these lesions and erroneously characterized six as false negatives.
Preparation allowed the evaluation of all colonic segments. Conventional CTC recorded average sensitivity, specificity, and positive and negative predictive values of 85.3%, 96%, 82.9%, and 96.7%, respectively. Comparable filet view CTC values were 82.9%, 97.4%, 87.9%, and 96.2%, respectively.
Filet view CTC sensitivity and specificity for lesions smaller than 5 mm, between 5 mm and 9 mm, and larger than 9 mm were, in turn, 81.8% and 97.6%, 88.9% and 98.8%, and 100%, respectively. The scanning protocol included 16-slice scanning at 1-mm slice thickness, 0.5-mm reconstruction interval, and 120 kVp/50 mAs.
Filet view allows simultaneous evaluation of supine and prone acquisitions after adequate colonic cleansing and bowel distention. Software reconstructs the colon lumen in 3D, then draws a center-dividing line from cecum to rectum. It virtually bisects and spreads out the colon along its longitudinal axis. The system can depict the unfolded colonic haustra with a 10° overlap at the edges as a safety margin, in cine mode, and without the need for bidirectional fly-throughs.
Diagnóstico Maipú radiologists are already using filet view in all patients undergoing CTC. The technique helps clinicians reduce postprocessing and reading time and allows them to provide patients with their results immediately after the exam, Carrascosa said.
The Argentine healthcare system currently covers virtual colonoscopy in patients with a personal or family history of colorectal cancer or polyps. It also covers CTC for related clinical conditions, such as rectal bleeding, or patients with positive fecal occult blood tests. Nonsymptomatic patients pay out-of-pocket fees for screening studies, she said.
For more information from the Diagnostic Imaging archives:
3D postprocessing tools shed light on hidden polyps
Pediatric CT colonography sheds light on screening
Prepless CT colonography hides lesions, hinders read
