الفهرس 
Human Visual SystemOnly 14 pages are availabe for public view
 |  | from | 149 |  |  |
| | | from | 149 | | |
Abstract
Visual prostheses hold hope of vision restoration for millions with retinal degenerative diseases. The idea of visual prosthesis is to integrate into the functional part of the visual system, replacing the damaged part, hence restoring vision. The first main contribution of the thesis is the description of the development of a probabilistic model for a network of neurons in the Lateral Geniculate Nucleus (LGN), which is an important structure in the visual pathway. Testing the model has shown that neuronal firing properties in the model are similar to those reported in studies based on real data. The second main contribution of the thesis is developing a method for estimating the proper electrical stimulus. This leads to enhancing the performance of visual prostheses by using two different approaches: Kalman filters and autoencoders. The thesis also describes the training of both these approaches to learn how the brain encodes information and imitate that code. The goal of the developed approaches is to estimate an electrical stimulus that is equivalent to a certain visual stimulus, such that they both elicit similar responses from the LGN model. Results have shown that both approaches succeeded in estimating the equivalent electrical stimulus, with mean correlation between visually- and electrically-evoked responses of up to 0.6885 for optimal electrode-placement, and 0.413 for random electrode-placement. This indicates the efficacy of the proposed approaches in estimating an electrical stimulus equivalent to a specific visual stimulus.