Assessment of papilledema using fundus images

Bio-medical imaging is playing an important role in the laboratory research, clinical practice, diagnosis and treatment of various diseases. Papilledema is an optic disc swelling, which is occurred due to increased Intracranial Pressure (ICP) of cerebrospinal fluid. Papilledema is the only initial guess for many underlying diseases, therefore, early detection of edema is very essential in emergency situations. The objective of this study is to design a computer based automated system for detection as well as observance of papilledema. We have used 46 fundus images of STARE database, which includes 36 images with no edema (normal images set) and 10 images with optic disc swelling. New method has been developed for the classification of images into papilledema and normal images, which extracts 10 features (vascular related features = 6 and GLCM textural features = 4) of 46 images and detected the changes occurred in the images due to papilledema. In order to classify the images into diseased and normal images, a Support Vector Machine (SVM) Classifier with its Radial Basis Function (RBF) kernel is used for classification of images. Textural and vascular related features are used in this study, which shows improved results. Classifier achieved the accuracy of 95.65 %. It demonstrates that this system can be used for robust and automatic diagnosis of papilledema in normal and emergency environments.