Stereoscopic endoscopy: Is one eye better or two?

A prototype stereoscopic (stereo) endoscope that employs a double charge-coupled device system has been shown to be highly accurate in measuring size via computer generated derivations. Whether stereovision allows an endoscopist to become more accurate and adept has not been determined. Aims: 1) To modify an existing prototype endoscope so that it provides an endoscopist with stereovision. 2) To evaluate whether stereovision enhances endoscopic performance. Methods: 19 subjects (6 Gl attendings, 6 Gl fellows, 7 non-endoscopists) were fitted with a headset designed for virtual reality applications in which video signals were fed via separate monitors to each eye. Subjects were tested using both the stereo system and a monocular (mono) videoendoscopic system fitted with the identical headset. Subjects were randomly assigned to either stereo or mono systems and then crossed over to testing in the other system at a minimum of 15 days between testing sessions (mean = 41 days). Standardized spatial, motor-dexterity, visual acuity and binocular eye exams were performed in each subject. Testing was performed within a 2′× 3′ model with objects and tasks placed at different positions within the system. Performance of a specific manual task, accuracy of size estimates, depth perception and speed of performance were evaluated. Results: Visual tests revealed no significant abnormalities in acuity; however, 3 subjects had minor abnormalities in binocular vision in the near field. There was no correlation between visual, spatial or dexterity testing and endoscopic performance. Comfort, resolution and color reproduction of the head-mounted display optics was acceptable to most participants although a small number of participants (4 stereo, 5 mono) experienced nausea during the testing procedure. There was no significant difference between stereo and mono testing with regards to accuracy of performance of a specified task (knocking wooden-blocks over), estimates of size (12 objects, range from 3.1 mm to 37.7 mm), depth perception (7 objects) or speed (total and task-related) of performance. Conclusions: 1) The current prototype stereoendoscope can be interfaced with a lightweight, head-mounted display providing fused stereo images. 2) Performance on standardized spatial, manual-dexterity or binocular visual tests did not correlate with endoscopic performance. 3) As compared to monocular viewing, stereoendoscopy did not yield a significantly superior accuracy of performance of a specified task, estimate of size, depth perception or speed of performance.