Wider 3D viewing angle may aid polyp detection

By Eric Barnes, AuntMinnieEurope.com staff writer

September 9, 2008 -- Virtual colonoscopy (VC or CT colonography) readers saw a greater percentage of the colonic mucosa by increasing the field-of-view on their 3D software to match that of a standard colonoscope. However, the wider viewing angle may have also reduced polyp conspicuity and increased distortion, with unknown effects on detection sensitivity.

In recent years, researchers have tried various methods of improving the visualization of colonic mucosa while improving reading times in 3D analysis of VC data, explained study authors Dr. James East, Dr. Brian Saunders, Dr. David Burling, and colleagues from St. Mark's Hospital and University College Hospital in London.

But novel 3D viewing methods, including the unfolded cube, virtual dissection, and filet view displays, have failed to gain widespread acceptance amid concerns that surface distortion, too much visual information presented to the reader, and other drawbacks may actually reduce detection accuracy.

Their study, published this month in European Radiology, aimed to determine if a wider viewing angle might reduce the percentage of missed mucosa during bidirectional 3D endoluminal fly-throughs of VC data, while maintaining detection accuracy.

"While 3D endoluminal displays are designed to simulate an endoscopic perspective, most software restricts the field-of-view to 90°," East and colleagues wrote. "This is considerably less than the standard 140° used in modern optical colonoscopes. A narrower field-of-view may, however, improve polyp conspicuity by reducing distortion. A unidirectional fly-through leaves approximately 25% of the colonic surface unseen and bidirectional viewing is, therefore, mandatory, but doubles reading time" (European Radiology, September 2008, Vol. 18:9, pp. 1910-1917).

Without the aid of a visualization tool to pinpoint missed regions, even bidirectional viewing misses 6% of the mucosa on average, they noted. The pilot study sought to determine if unidirectional viewing at a wider angle might reduce the need for missed region software compared to bidirectional viewing at a narrower angle. It also assessed whether the wider viewing angle was more efficient and whether it maintained adequate polyp conspicuity. A promising outcome would justify the initiation of larger studies with direct clinical outcome measures, they wrote.

The retrospective study examined high-quality prone and supine CT colonography (CTC) datasets from 20 patients (mean age, 66 years; range 41-81) selected from a previous study. All subjects had undergone virtual colonoscopy on a four-slice MDCT scanner (LightSpeed Plus, GE Healthcare, Chalfont St. Giles, U.K.) following cathartic bowel preparation and automated insufflation of CO₂ (ProtoCO₂L, Bracco Diagnostics, Princeton, NJ). Images were acquired at 1.25 or 2.5-mm collimation, 50-100 mA, and 50% section overlap, East and colleagues wrote.

Two relatively new readers who had undergone formal CTC training, a gastroenterologist with experience in 30 CTC cases and a radiologist with experience in 50 CTC cases, compared the results of unidirectional and bidirectional fly-throughs using 140° versus 90° viewing angles in terms of mucosal coverage. They also analyzed reader times for unidirectional 140° versus 90° fly-through, and analyzed digital "snapshots" of the mucosa taken from both viewing methods to compare lesion conspicuity.

All interpretation was performed on a commercial 3D workstation (V3D-Colon, Viatronix, Stony Brook, NY) that had software modified by the manufacturer to accommodate the different viewing angles.

"The study was performed in three stages -- missed region measurement, reading efficiency, and polyp conspicuity -- performed one month apart," the team wrote. "... For each viewing angle, care was taken to achieve identical colonic segmentation for each dataset series. All fly-throughs were performed using constant speed of 120" (e.g., 60% of the maximum speed of the software). Polyp conspicuity was evaluated and scored in a blinded, randomized fashion in both viewing angles.

The results showed that bidirectional 140° viewing reduced the number of missed areas between eight- and 40-fold, depending on the polyp size category, compared to standard 90° bidirectional fly-through without the additional software tool (p < 0.001).

"In keeping with earlier studies, we found a relatively large number of medium and large missed areas after 90° bidirectional fly-through: on average over four areas > 1,000 mm² were unseen, any of which could potentially harbor a large polyp," they wrote. "In contrast, the mean number of large missed areas after wide-angle bidirectional fly-through was less than 0.3 per case," sufficient to potentially eliminate the need for the missed area software.

A comparison of unidirectional 140° fly-through with standard bidirectional 90° fly-through found the unidirectional method to be faster by a mean of 3.8 minutes (9.3 minutes versus 5.5 minutes, p < 0.0001), while achieving the same surface visualization.

"The review time benefits seen at 140° may also be attenuated if the review speed needs to be slowed due to the increased amount of information presented," they wrote.

When viewed together as pairs, however, polyps were found to be more conspicuous when visualized in the 90° field-of-view (p = 0.03) compared to 140° viewing.

"Wide-angle CTC can reduce both numbers of missed areas and review times," East and colleagues wrote. "However, this may be at the expense of polyp conspicuity."

"The advantages in terms of time efficiency and mucosal visualization of routinely increasing the field-of-view to 140° must be weighed against the potential to distort colonic anatomy and potentially reduce polyp conspicuity," they wrote.

Larger studies that compare the strategies in terms of polyp detection sensitivity and specificity, as well as different viewing software applications, will be needed to validate the initial results, the authors noted.

"In reality, the optimum method of time-efficient 3D analysis, be it virtual dissection or a combination of widened endoluminal field-of-view, uni- or bidirectional analysis, and use of the missed region tool, remains to be defined and may come down to individual user preference, software access, and level of diagnostic confidence required," they added.

By Eric Barnes
AuntMinnie.com staff writer
September 9, 2008