Diabetic macular edema (DME) is a complication of diabetes. It can blur or distort patients' vision and make the blindness. It is categorized by the presence of lesions. Habitually the presence of lesions is detected by Ophthalmologists from the dilated retinal images captured by dropping chemical solution into an eye. This process peeves the patients. So, there is a need for an autonomous method to detect the presence of lesions using image processing algorithm from the non-dilated images to

more »

... lp the ophthalmologists to diagnose the disease without inconvenience and irritation to the patient and thus protects patients from vision loss. In this work, Meadian filter and Contrast Limited Adaptive Histogram Equalization used in image preprocessing. Motion pattern estimation with masking process used for segmentation. To extract the feature Grey Level Co-occurrence Matrix isused. Support vector machine used to classify the severity level for disease. The proposed algorithm has produced the sensitivity of 99.743%, specificity of 97.14% and accuracy of 97.711%.It is more helpful for ophthalmologist in the detection of DME. Since this method is automated, it detects faster and this level of accuracy in result helps the ophthalmologists to diagnose the disease very easily.