Phosphene mapping in a 44-channel suprachoroidal retinal prosthesis

Abstract

Purpose : Visual prosthesis stimulation strategies generally assume a retinotopic mapping of the stimulating electrodes. We investigated perceived phosphene locations versus retinotopic electrode positions in retinal prosthesis recipients to validate this assumption.

Methods : Three end-stage retinitis pigmentosa (RP) recipients (S1-3) of a 44-channel suprachoroidal retinal implant (NCT03406416) performed a mapping task for a subset of electrodes. In each trial a single electrode (or a shorted pair) was activated for 500ms and the subject was instructed to move their eyes (head-fixed) to the location of the phosphene. Perceived phosphene locations were expressed as the change in eye position (° of visual arc) between stimulus onset and the end-point of the resulting saccade. Predicted phosphene locations were estimated by comparing electrode locations in fundus photos with theoretical projections in the Drasdo-Fowler schematic eye. In addition, participants performed a square localisation task (10° wide targets, n=24) on a 42″ touchscreen, and pointing precision was compared to the error between predicted and perceived phosphene locations in the mapping task.

Results : In S1 and S2, phosphene locations correlated well with retinotopic electrode locations. Variability in phosphene location for S1 and S2 was correlated with the eccentricity of the electrode from the fovea (S1: R²=0.69; S2: R²=0.81), suggesting more eccentric electrodes produced less spatially distinct phosphenes. For S3, phosphene locations did not correlate with electrode locations and there was little distinction between the measured locations of different phosphenes, indicating large or spatially indistinct phosphenes. This was associated with worse pointing precision in the target localisation task for S3 compared to S1 and S2 (Kruskal-Wallis p<0.001). Variability in phosphene location was not correlated with electrode eccentricity in S3 (R²=0.02). S3 had a more clinically advanced form of RP than S1 and S2.

Conclusions : Retinotopic position was predictive of perceived phosphene location for two of three subjects in an eye-tracker mapping task. In the third subject all phosphenes appeared in a similar region of the visual field and functional performance was worse, presumably due to the distortion of retinal morphology expected with advanced RP. Accurate mapping of spatial perception may improve camera-to-electrode mapping strategies.

This is a 2021 ARVO Annual Meeting abstract.