Neural information processing efforts to restore vision in the blind

Abstract

Retina Implants belong to the most advanced and truly 'visionary' man-machine interfaces. Such neural prostheses for retinally blind humans with previous visual experience require technical information processing modules (in addition to implanted microcontact arrays for communication with the remaining intact central visual system) to simulate the complex mapping operation of the 5-layered retina and to generate a parallel, asynchronous data stream of neural impulses corresponding to a given optical input pattern. In this paper we propose a model of the human visual system from the information science perspective. We describe the unique information processing approaches implemented in a learning Retina Encoder (RE), which functionally mimics parts the central human retina and which allows an individual optimization of the RE mapping operation by means of iterative tuning using learning algorithms in a dialog between implant wearing subject and RE. © Springer-Verlag Berlin Heidelberg 2004.