Colonoscopy is the “gold standard” for colorectal polyp and cancer detection, but important lesions may be missed on the proximal aspect of haustral folds, rectal valves, or flexures.

Our purpose was to evaluate a prototype auxiliary imaging device that extends beyond the colonoscope’s tip, providing a continuous retrograde view to detect lesions missed by the forward-viewing colonoscope.

Three anatomic models of the colon were prepared with simulated polyps, 32% in obvious locations and 68% on the proximal aspect of folds. Six endoscopists examined each model with two methods. Method A used a standard video colonoscope. Method B involved an identical colonoscope with a retrograde-viewing auxiliary device positioned within its instrument channel. Order of testing was randomized and blinded.

Laboratory bench.

Detection rates for simulated polyps.

Of 78 “obvious” polyps, 69 (88%) and 70 (90%) were detected by methods A and B, respectively (P > .9). In contrast, of 162 polyps on proximal aspects of folds, 20 (12%) and 131 (81%) were detected by methods A and B, respectively (P < .00001).

Limitations resulted from (1) use of commercially available anatomic models in which haustral folds are less prominent and more rigid than in humans and (2) evaluation of a prototype device that had larger size and narrower angle of view than the planned production model and that was fixed in relation to the colonoscope.

In simulated testing, a retrograde-viewing auxiliary imaging device used with a standard video colonoscope significantly improves detection rates of simulated polyps and promises to enhance the diagnostic yield of colonoscopy in humans.