In this paper the study of fundus image segmentation using convolutional neural networks is carried out. A neural network architecture was made to classify four classes of images, which are made up of thick and thin blood vessels, healthy areas, and exudate areas. The CNN architecture was constructed empirically so as the required accuracy of no less than 96 % is ensured. The segmentation error was calculated on the exudates class, which is key for laser coagulation surgery. In the paper we utilized the HSL color model because it renders color characteristics of eye blood vessels and exudates most adequately. We have demonstrated the H channel to be most informative. We have investigated the robustness of technology to various noises. Experimental studies have shown the instability of the convolutional neural network to Gaussian white noise and resistance to impulse noise.
The paper proposes a method for selection the region of diabetic macular edema in fundus images using OCT data. The relevance of the work is due to the need to create support systems for laser coagulation to increase its effectiveness. The proposed algorithm is based on a set of image segmentation methods, as well as searching for specific points and compiling their descriptors. The Canny method is used to find the boundary between the vitreous body and the retina in OCT images. The segmentation method, based on the Kruskal algorithm for constructing the minimum spanning tree of a weighted connected undirected graph, is used to select the retina to the pigment layer in the image. Using the results of segmentation, a map of the thickness of the retina of the eye and its deviation from the norm were constructed. In the course of the research, the optimal parameter values were selected in the Canny and graph segmentation algorithms, which allow to achieve a segmentation error of 5 %. SIFT, SURF, and AKAZE methods were considered for super-imposing calculated maps of the retina thickness and its deviation from the norm on the fundus image. In cases where a picture from the fundus camera of the OCT apparatus is provided along with OCT data, using the SURF method, it is possible to accurately combine with the fundus image.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.