Abstract
Purpose: To simplify the imagery presented to Argus II retinal implant wearers by filtering out information from background objects using head-mounted a thermal camera (T) and a stereoscopic distance filter camera (D). The functionality of these two systems was compared with normal Argus II camera (N) imagery in mobility and obstacle avoidance tasks.
Method: The data were acquired from Argus II users in a mobility test with T/N and obstacle avoidance test with D/N. In both experiments, the accuracy and response time were recorded and were used to calculate the cumulative probability of success over time. Performance using the T/D versus N cameras was compared using the Kolmogorov-Smirnov (K-S) test.
Results: Four Argus II users aged 62–87 completed the mobility task and three of them performed the obstacle avoidance task. Despite interindividual differences, K-S testing of the cumulative probability of success in both experiments showed that significantly less time was spent to respond with T/D compared to N, yet with higher accuracy (p<0.005).
Conclusion: Subjects detected obstacles and people more accurately with T/D than using N. Even though N would have allowed subjects to use cues such as target size to make the discrimination, performance with this camera was much slower and close to chance, presumably due to an interference by (non-target) phosphenes. Addition of T/D camera to N camera in the Argus II system may have significant real-life implications for personal safety and social interactions, thereby improving quality of life.