Novel Approach For Detection Of Early Diabetic Retinopathy - link
The early detection of Diabetic Retinopathy is necessary to prevent blindness. Retinal imaging is a common clinical procedure used to record the visualisation of the retina. The main difficulty in image capture of the ocular fundus is image quality which is affected by factors, such as medial opacities, defocus or presence of artefact. Micro Aneurysms (MA) are the earliest clinical sign of Diabetic Retinopathy. The main problem in diagnosis of eye diseases is the appearance and structure of blood vessels in retinal images. The detection of blood vessels is difficult in the automatic processing of retinal images. In proposed method, for segmentation of blood vessels, the contrast-limited adaptive histogram equalisation (CLAHE) is used along with canny edge technique for exact edge location. The method is tested on publicly available Digital Retinal Images for Vessel Extraction (DRIVE) database. Comparative results of all basic Edge detection techniques with respect to the quality metrics parameter such as energy, entropy, mean etc is analysed. It is noted that proposed method performs well in extracting the vascular pattern than traditional edge detection techniques.
Link to the matlab code: link