Analysis of Confocal Microscope Images from Retina Detachment Experiments Using Texture Bases Features

Retinal detachment, the separation of retina from the retinal pigmented epithelium caused by eye injury, has been studied extensively for decades [1]. It causes many cellular changes inside the retina tissue, which could significantly affect the structure, state and function of the retinal tissues, thus may cause the loss of vision. In order to devise therapies for retinal detachment, extensive research has been done to model these cellular changes and to study how the changes can be recovered by various treatments. The goal is to, after various treatments, reattach the retina and recover cells to normal states, and therefore to return the vision. In this paper, we apply image processing techniques to extract numerical features based on image textures, and then using this feature vector as representation of the image, we statistically correlate the image features to the biological parameters, such as experimental conditions. We then study the behavior of various groups of images taken under different experimental conditions and perform classification between these images classes.