Hand-Camera Coordination Varies over Time in Users of the Argus® II Retinal Prosthesis System

Introduction: Most visual neuroprostheses use an external camera for image acquisition. This adds two complications to phosphene perception: (1) stimulation locus will not change with eye movements; and (2) external cameras can be aimed in directions different from the user’s intended direction of gaze. Little is known about the stability of where users perceive light sources to be or whether they will adapt to changes in camera orientation. Methods: Three end-stage retinitis pigmentosa patients implanted with the Argus II participated in this study. This prosthesis stimulated the retina based on an 18° × 11° area selected within the camera’s 66° × 49° field of view. The center of the electrode array’s field of view mapped within the camera’s field of view is the camera alignment position (CAP). Proper camera alignments minimize errors in localizing visual percepts in space. Subjects touched single white squares in random locations on a darkened touchscreen 40 or more times. To study adaptation, subjects were given intentional CAP misalignments of 15–40° for 5–6 months. Subjects performed this test with auditory feedback during (bi-)weekly lab sessions. Misaligned CAPs were maintained for another 5–6 months without auditory feedback. Touch alignment was tracked to detect any adaptation. To estimate localization stability, data for when CAPs were set to minimize errors were tracked. The same localization test as above was used. Localization errors were tracked every 1–2 weeks for up to 40 months. Results: Two of three subjects used auditory feedback to improve accuracy with misaligned CAPs at an average rate of 0.02°/day (p < 0.05, bootstrap analysis of linear regression). The rates observed here were ~4000 times slower than those seen in normally-sighted subjects adapting to prism glasses. Removal of auditory feedback precipitated error increases for all subjects. Optimal CAPs varied significantly across test sessions (p < 10−4, bootstrap multivariate analysis of variance (MANOVA)), up to 21–29° within subjects over the observed period. Across subjects, optimal CAPs showed an average rate of change of 0.39°/day (SD 0.36°/day). Conclusions: Optimal CAPs varied dramatically over time for all subjects. Subjects displayed no adaptation to misaligned CAPs without feedback. Regular recalibration of CAPs may be required to maintain hand-camera coordination.