Retina modeling by artificial neural networks

The objective of this article is to provide a theoretical framework for the structure and function of the retina. The first focus of this article is the examination of the physiological aspects of the retina. The given study proposes a model that utilizes artificial neural networks to create the model's structure. This approach is motivated by the resemblance between the behavior of the model and that of retinal cells. The neural network receives as input the intensity of light that is incident onto the retina and produces as output the activity of the retinal ganglion cells. A comparison has been conducted between the data derived by the model and the biological data. This comparative analysis demonstrates that neural networks can adequately simulate the behavior of ganglion cells. However, the effectiveness of the network is contingent upon its architectural configuration, the number of hidden layers used, and the specific learning method utilized. The experimental findings demonstrate that including the output from earlier iterations as input to the neural network results in the system exhibiting memory. This approach enhances the model's efficiency and mitigates the occurrence of periodic behavior. The model, as mentioned above, has potential applications in the development of artificial retinas, serving as a hardware implementation to restore some visual capabilities in individuals with visual impairments.