Detection and Classification of Retinal Diseases in Spectral Domain Optical Coherence Tomography Images based on SURF descriptors

Optical Coherence Tomography (OCT) is a non-invasive eye-imaging modality for detecting macular edema both in its early and advanced stages. The main aim of this work is to present the automatic detection of edema of the retinal layers particularly around the macula in diabetic patients. After detection and extracting certain features in the OCT retinal images a classification of the type of Diabetic Macular Edema is done. In this method during preprocessing stage we remove the speckle noise followed by flattening and cropping of the image is done. Then this is followed by Speeded up robust feature extraction. The extracted features are then classified using Support Vector Machine binary classifier as normal or abnormal and thus having Diabetic Macular Edema. This technique has been applied for 25 normal and 45 abnormal OCT images. The results show that this method accurately detected edema diseases in between the layers in the retinal. Then we could classify them using Support Vector Machine as normal or abnormal. Experimental results shows that an average retinal disease detection accuracy of 99% for Support Vector Machine (SVM) classifier. Thus, this algorithm can be used by ophthalmologists in early detection of Macular Edema.