الفهرس | Only 14 pages are availabe for public view |
Abstract Animal experiments were performed to record and analyze neural activities from the LGN of anesthetized rats corresponding to a specially engineered external visual and electrical stimuli. The datasets obtained were further studied to gain more understanding of the LGN neurons properties. Computational models were developed to address the neural coding (encoding and decoding) problems in the LGN neurons with the aid of the experimentally recorded datasets. A proposed model for thalamic visual prosthesis was also introduced. Results have shown the efficacy of the proposed computational models. The thesis is organized as follows: Chapter 1 gives an introduction and states the main contributions in the research. Chapter 2 discusses the theoretical backgrounds related to the thesis. Chapter 3 explains the animal experiments procedures, the data preprocessing and the statistical analyses performed to obtain the stimulus-driven neural activities datasets. Chapter 4 introduces the usage of the Artificial Neural Network techniques in solving the neural coding problems of the neuronal activities of the LGN in rats. It also describes the proposed approach for building a thalamic visual prosthesis system. Chapter 5 shows the utilization of deep convolutional neural networks in solving the neural encoding problem and shows the performance enhancement using such techniques. Finally, Chapter 6 concludes the thesis and discusses potential future work. |