Optic disc (OD) is an important part of the eye. In developing systems automatic OD detection is an important step for automated diagnosis of various serious ophthalmic diseases such as Diabetic retinopathy, Glaucoma, hypertension, etc. The variation of intensity within the optic disc and intensities close to the optic disc boundary are the major hurdle in automated optic disc detection. General edge detection algorithms are frequently unsuccessful to segments the optic disc because of this. Complexity increases due the presence of blood vessels. This paper presents a simple method for OD segmentation by using techniques such as Principal Component Analysis (PCA), Mathematical Morphology and Circular Hough Transform. PCA used for good presentation of the input image and mathematical morphology is used to remove blood vessels from image. Circular Hough Transform is used for boundary segmentation.
Keywords—Diabetic retinopathy; glaucoma; optic disc; principal component analysis.
I. INTRODUCTION
T
he OD is an important part of the human eye. Human eye receives a light and is carried through the optic nerve to the brain. The optic disc is the point from where the optic nerve exits the eye, the anatomy of the human eye is shown in the Fig.1. To respond a light there is no presence of the light sensitive rods at this point so, there is a break in the field of vision. The entry point of the blood vessels that supplies the retina is at the OD. The optic nerve head in a normal human eye carries millions of neurons from the eye towards the brain. The position of the OD is 3 to 4mm to the nasal side of the fovea. It is a slightly oval shape, with average dimensions of 1.82mm horizontally by 1.92mm vertically.
The OD is considered as one ...
... middle of paper ...
...th successful OD segmentation.
Fig. 9. Results of the proposed method with failure in OD segmentation.
This technique is applied after vessel removal. Circular Hough transform gives boundary of the OD.
IV. RESULTS
The proposed method applied on selected images from MESSIDOR database. In some of the images OD segmented properly and in some images proposed method failed to segment OD properly as shown in Fig. 8 and Fig. 9 respectively.
V. CONCLUSION
This paper presents a simple method for OD segmentation by using image processing techniques. This method applied to selected images. Accuracy depends on image acquisition and variations in image. This method will be helpful in mass screening of diabetic retinopathy and in calculating C/D ratio in glaucoma diagnosis.
ACKNOWLEDGMENT
The authors would like to thank the MESSIDOR program for facilitating their database.