Research output: Contribution to journal › Article
Images produced by rod-lens telescopes used in minimally invasive surgery are brightest in the central region and darker at the periphery. To enable a clear view of the darker regions of the image, the intensity of light from the source usually is set to a high level. This often causes substantial reflection and glare from surgical tools and some tissue surfaces, which can be disturbing for the surgeon. This study investigated digital image processing methods in an attempt to reduce glare without introducing other adverse qualities into the images.
Two methods of reducing glare in local high-brightness regions of the image were evaluated. The first method reduced intensity by a fixed amount while also optionally introducing a slight color. The second method combined a proportion of the original intensity value with a proportion of a lower intensity value, again with an optional color bias. Two surgical video clips were modified with each of 13 different glare-reduction variants using these methods. These and the original sequence were played to a group of 10 experienced surgeons for subjective assessment.
The pixel-based methods both showed statistically significant improvements over the original version. The incorporation of a slight yellow bias was preferred to a straightforward gray-level reduction. The simple approach of using a lower level of brightness alone was found to be unacceptable. Both new methods work in real time at normal video speeds.
Antiglare methods have been found that reduce the perception of glare and are otherwise unobtrusive. This encourages further work to refine the preferred methods and to test them with a larger group over a wider range of video sequences. Clinical trials then will follow.