Abstract
Purpose: Retinal prostheses may be able to partially restore visual function in the relatively near future. However it is not yet clear how useful low-resolution devices would be in patients' daily lives. We measured performance on a variety of low-vision activities of daily living (LVADL) as a function of the resolution of a simulated prosthetic device. Methods: We measured performance in sighted volunteers, using a retinal prosthetic simulator, across 23 LVADL. The resolution of the simulated prosthetic vision varied between 4′4 pixels (the resolution of current retinal prosthetics) and 32′32 pixels. We compared performance using the simulator to performance with normal vision, video vision, and without visual input. We tested 9 subjects, each subject performed each task 11 times. Performance was scored as follows: 0-unable to complete task, 1-unable to complete task without assistance, 2-completed the task within 2 standard deviations (std) of mean sighted time, 3-completed the task between 1–2 std above mean sighted time, 4-completed the task within 1 std of mean sighted time or better. Results: Performance as a function of resolution depended strongly on task. On some tasks the prosthetic device did not improve performance, regardless of resolution. On most of these tasks mean performance was relatively good (∼2.4) even without visual input. On some high-resolution tasks the prosthetic device did not improve performance until it was at a resolution of full video, whereupon performance improved from a mean level of 0.5 to a performance level of 3.75. On lower-resolution tasks performance monotonically improved with device resolution from a mean of 0.75 without visual input, to a mean of 3.7 with a 32′32 resolution simulator. Conclusions: Not surprisingly, the usefulness of a low-resolution simulator depends strongly on task. However, for many tasks, a low-resolution simulated prosthesis did significantly improve performance monotonically with resolution.
NSF - IGERT Fellowship