Active Sensing for the Enhancement of Reading Capabilities in Simulated Prosthetic Vision

Current retinal prostheses offer a promising solution to people with degenerative diseases of the outer retina. Notwithstanding, the obtained visual acuity is still severely limited and offers a very basic vision restoration. In this work, we study the enhancement of prosthetic reading performance through "active sensing".We generated prosthesis vision simulation by converting gray scale images to low resolution quantized gray-level images in which each pixel intensity is represented by a Gaussian phosphene. To investigate reading capabilities, various phosphenes pixilated word images were displayed at different phosphene densities and contrasts, as well as scaling. We implemented two sensing paradigms :"active sensing" modality by allowing the subject to actively scan or resample the scene using the mouse; and passive scanning produced by horizontal movements of words at 0.2-0.5 degrees. We measured reading capabilities and analyzed the scanning path chosen by the subjects. We have revealed a clear perference to active sensing modalitiy in the relevant phosphenes' densities and an improvement of up to a factor of ten when the zoom option was enabled. In addition we have revealed that active sensing modality was preferable in lower phosphenes' densities, implying that it is naturally used in attempt to comensate for lack of resolution.

* This research was carried out towards the Master degree in Electrical Engineering at Bar-Ilan University, under the supervision of Prof. Zeev Zalevsky and Dr. Yossi Mandel