Computerized extraction of vessel regions in retinal fundus images -- Performance evaluation using DRIVE Database

Blood vessel extraction is one important component technology of computer-aided diagnosis for the retinal fundus images. We developed a morphology filter bank for the enhancement of vessel regions in retinal fundus images. By using morphology filter bank, nodular patterns and linear patterns of various sizes can be enhanced. We applied our method to 40 images in the DRIVE database. The results indicated that mean area under the ROC curve was 0.919. Therefore, our method would be useful for the extraction of blood vessels in the retinal fundus images. Keywords Retinal fundus image, Vessel Extraction, Morphology Filter bank, Computer-aided diagnosis