Purchase this article with an account.
Michael P Barry, Michelle Armenta Salas, Uday Patel, Varalakshmi Wuyyuru, Soroush Niketeghad, William H. Bosking, Daniel Yoshor, Jessy D. Dorn, Nader Pouratian; Video-mode percepts are smaller than sums of single-electrode phosphenes with the Orion® visual cortical prosthesis. Invest. Ophthalmol. Vis. Sci. 2020;61(7):927.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Percepts elicited by visual prostheses in response to camera input are often modeled as sums of corresponding single-electrode percepts. Prosthesis users were asked to draw video-mode percepts to investigate the validity of this model.
Four blind subjects implanted with the Orion visual cortical prosthesis participated in these experiments. All stimulating arrays were implanted over the left medial occipital cortex. Subjects sat 30 cm away from a touchscreen monitor, positioned such that the left edge of the monitor was directly in front of the subject’s midline. Subjects fixated on that edge of the touchscreen during stimulation and while responding. For each trial, the subject drew perceived phosphenes in response to single-electrode or video-mode stimulation. Normal home-use video settings were adjusted such that each electrode stimulated at its maximum normal amplitude regardless of camera input. Up to 49 single-electrode samples were taken per electrode, and at least 2 responses to video stimulation were captured for each subject. Summative models were developed by taking the union of representative single-electrode stimulation responses. Areas of video-stimulation responses and summative models were compared within each subject.
Subjects’ models, using 22, 30, 49, and 30 electrodes, predicted total percept areas of 29, 387, 436, and 700 degrees2 with 20, 8, 3, and 5 distinct regions, respectively. Conversely, the actual video percepts reported by subjects were significantly smaller with fewer distinct regions: within-subject means of 5, 32, 179, and 87 degrees2 with 4.2, 2.0, 1.7, and 1.5 distinct regions. Areas of video-stimulation percepts were on average only 20% of model predictions (range: 8–41%, p < 0.01, sign test). Distinct region counts were 33% of model predictions (range: 21–56%, p < 0.01, sign test). Distinct-region areas with video stimulation were on average 1.8 times the area of single-electrode percepts.
Percepts elicited by video-mode stimulation, using the timings and maximum amplitudes for normal home use, were smaller and simpler than predicted by a summative model based on single-electrode phosphenes. Further data collection will help inform how video-mode percepts are constructed. Both enhanced understanding of and control over video-mode percepts will be necessary to improve the experience of artificial vision.
This is a 2020 ARVO Annual Meeting abstract.
This PDF is available to Subscribers Only