A Physiologically Based Model And Parallel Simulation For Retinal Outer Plexiform Layer

The main responsibility of outer plexiform layer is to obtain outer information for human visual system, makes preprocessing of visual information and forwards the processed visual information to inner plexiform layer, and finally transmits it to the brain for recognition. Outer plexiform layer is composed of three layers of different kinds of cells:cone layer, horizontal cell layer, and bipolar cell layer. The key point of this thesis is to build a physiological based model for retinal outer plexiform and carries on some paralleled simulations on the basis of the physiology data and physiological characteristics of retina.According to physiological data, in this thesis we finish the following models for outer plexiform layer:cone continuous distribution, cone light adaptation, spatial filtering model, horizontal cell spatial filtering, and bipolar receptive field. Model of cone continuous distribution is based on the facts that the density of cone is very high in fovea and that with the increase of the eccentricity of retinal cone cell the density decreases rapidly. Cone light adaptation is composed of bright adaptation, dark adaptation, and homeostatic adaptive of cone, making people see objects clearly in different light conditions. Model of horizontal cell spatial filtering is built on the basis that H1, H2, H3horizontal cell selectively receive signals from R, G, B cone cells and the differences among spatial gauss filter weights of the three types of cone cells. We build the model for midget bipolar and diffuse bipolar cell according to the types of bipolar cells, and the model of bipolar spatial filtering on the basis of the increase of bipolar receptive field with the increase of eccentricity.We design the software for the simulation system of outer plexiform layer according to the built model of outer plexiform layer. In this thesis, according to the requirements of the simulation system, we choose RGB bayer EXG50C of Baumer Company to obtain outer information and graphics card GEFORCE310M from NVIDIA Company as the chip to process the visual information. In order to overcome the problem of large calculation caused by the large population of the cells in simulation, we use compute unified device architecture (CUDA) parallel algorithm to realize the parallel processing of the visual information with NVIDIA graphics cards. Simulation results prove that it can meet the requirements of real-time simulation.Though the simulation system proposed in the thesis cannot be applied into clinical care, the model can verify some hypothesis in physiology. Meanwhile, it also can be the guideline for the perfection of visual prosthesis and the establishment of bionic eye system.