Fly-over method provides better colonoscopy visualization
Colorectal colon cancer is the second leading cause of cancer death in the western world. Several screening tests are available, among which fiber-optic colonoscopy is the gold standard. This detects more than 90% of colorectal cancers, but it is invasive, uncomfortable, and sometimes cannot reach the cecum at the top of the colon (near the appendix). To overcome these limitations, virtual colonoscopy (VC) has emerged as a computer-based alternative to real colonoscopy. VC is non-invasive, more cost-effective than fiber-optic colonoscopy, and is free of risks and side effects. Interpreting a VC exam, however, is very labor intensive because rendered views must be compared with raw images. A more intuitive and robust method for viewing rendered images is, therefore, essential to virtual colonoscopy's widespread acceptance.
A visualization technique commonly used to simulate the view one would see during a real colonoscopy is called virtual fly-through navigation. Here, a virtual camera with a specific field of view moves along a special planned path inside the colon to render its internal views. 1 In general, fly-through methods suffer from lower surface visibility coverage: they do not show the backsides of folds in haustra––– small pouches in the colon–––or around sharp bends. Visualization techniques proposed to address those limitations include colon flattening and panoramic views. Colon-flattening methods2 visualize the entire colon as a flat sheet, but have the disadvantage that the same polyp may appear at several locations. On the other hand, panoramic-view methods3 maximize surface visibility coverage but use an unusual layout that confuses physicians. In addition, a polyp may split among several panoramic views.
We have proposed a new visualization technique for VC called virtual fly-over. The method consists of two steps. In the first step, the colon is split along its centerline into two halves, while in the second step each half is assigned a virtual camera that performs virtual fly-over navigation. To realize the first step, we divide the colon surface into consecutive strips of triangles (rings) as shown in Figure 1 (a). To realize the second step, each ring is split into two halves by an infinite clipping plane that passes along its local centerline. Finally, each set of consecutive half-rings are concatenated together to form one half of the colon as shown in Figure 1(b).
We quantitatively validated the effectiveness of the proposed method by computing the surface visibility coverage during fly-over navigation as shown in Figure 2. Experimental results on clinical datasets have shown that the average coverage was 99.59±0.2%. In Figure 2, we show captured rendered views during fly-through and fly-over navigations. While the techniques reveal an equal number of polyps with some rendered views, for others the new fly-over method finds more.
To conclude, we have presented a new virtual colonoscopy method that has several advantages over existing methods. It does not suffer from geometric distortion as do panoramic view techniques. Also, because one can control the elevation of the virtual camera over each colon half, there is no restriction on the viewing angle (e.g., >90°). This maximizes visualized surface areas and avoids the perspective distortion that plagues traditional fly-through methods. The camera viewing volume is perpendicular to each colon's half, so significant polyps (for example, those larger than 5 mm) hidden behind haustral folds are less likely to be overlooked. Finally, because the orientation of the splitting surface is controllable, the navigation can be repeated at another split orientation if a polyp is divided between the halves.
