Abstract
Purpose::
Visual neuroprostheses will likely be complex systems involving a large number of channels with different bio-electrical parameters for stimulation that will require a lengthy tuning process for every implanted patient, and for every channel. We have developed a research platform that (a) automatically estimates values for stimulation, (b) performs psychophysical tests to characterize the nature of the evoked percepts, (c) corrects possible deformations of the evoked patterns and d) can be used to test how parameters of phosphene-based vision affect visual task performance.
Methods::
The system is composed of a real-time retina-like encoder which contains a highly flexible retinal model coupled to a PC that can send configuration and stimulation commands to any kind of visual neuroprosthesis. The platform includes a module that can be used to perform psychophysical experiments in sighted volunteers by means of a second PC with a tactile screen and a Head-Mounted Display (HMD).
Results::
The platform has been successfully tested in the following experiments: brightness sensitivity, spatial resolution, temporal resolution, stimulus orientation and simple pattern discrimination. These tests were performed with sighted volunteers, including non-computer literate, elderly patients, and one with a transplanted cornea and glaucoma. All the volunteers performed well in their first trials after verbal instructions, and easily understood the feedback mechanisms. The spatial remapping procedures significantly reduced the errors associated with representation of visual space.
Conclusions::
We have developed a patient-controlled system that will coordinate and spatially remap the perceptions that are expected to be generated by a visual prosthesis. This platform automates all the procedures, provides a vehicle for multiple psychophysical experiments and will facilitate the design and testing of any visual neuroprosthesis.
Keywords: clinical research methodology • visual cortex • retina