Splat Feature Classification with Retinal Hemorrhage Detection in Diabetic Patients

First Author M. Kayalvizhi* Post Graduates Student P.G. Department of Computer Science Govt . Arts College, Melur 625 106 Email :[email protected] Second Author T.Balaji* Assistant Professor P.G.Department of Computer Science, Govt . Arts College, Melur 625 106. Email:[email protected] -------------------------------------------------------------------------Abstract-------------------------------------------------------------A novel splat feature classification method is presented with application to retinal hemorrhage detection in fundus images. Reliable detection of retinal hemorrhages is important in the development of automated screening systems which can be translated into practice. Automated detection of diabetic retinopathy (DR), as used in screening systems, is important for allowing timely treatment and thereby increasing accessibility to and productivity of eye care providers. Because of its cost-effectiveness and patient friendliness, digital color fundus photography is a prerequisite for automated DR detection. Patients with images that are likely to contain DR are detected and referred for further management by eye care provides. Under this supervised approach, retinal color images are partitioned into non overlapping segments covering the entire image. Each segment contains splat, pixels with similar color and spatial location. A set of features is extracted from each splat to describe its characteristics relative to its surroundings and responses from a variety of filter bank, interactions with neighboring splats, shape and texture information. An optimal subset of splat features is selected by a filter approach followed by a wrapper approach. The general characteristics of training data are used to select features. The wrapper method searches for an optimal feature subset tailored to a particular algorithm and a domain. Improvement in accuracy is achieved for some datasets for the two families of induction algorithms used: decision trees and Naive-Bayes. In addition, the feature subsets selected by the wrapper are significantly smaller than the original subsets used by the learning algorithms, thus producing more comprehensible results.