A semi-automated system for Optic Nerve Head Segmentation in Digital Retinal images Sayan Chakraborty Dept. of CSE, Bengal College of Engg. & Tech, Durgapur, West Bengal, India, sayan.cb@gmail.com Debmalya Chatterjee Dept. of CSE, Bengal College of Engg. & Tech, Durgapur, West Bengal, India, bittudeb@outlook.com Suvojit Acharjee Dept. of ECE, National Inst. of Tech, Agartala, Tripura, India, acharjeesuvo@gmail.com Aniruddha Mukherjee Dept. of CSE, JIS College of Engineering, Kalyani, West Bengal, India, aniruddha.mukherjee1@gmail.com Prasenjit Maji Dept. of CSE, Bengal College of Engg. & Tech, Durgapur, West Bengal, India, maji.katm@gmail.com Nilanjan Dey Dept. of CSE, Bengal College of Engg. & Tech, Durgapur, West Bengal, India, dey.nilanjan@gmail.com Abstract—Out of the twelve pairs of cranial nerves, optic nerve is the most important cranial nerve. Optic nerve establishes a connection from eyeball to brain. The four segments present in optic are: intraocular, intra-orbital, intracanalicular, and intracranial. Optical nerve head segmentation may lead to detection of many diseases such as glaucoma, eye hypertension. We propose a semi-automated system, where independent contours are defined, and then with the help of those ground truths, a person with medical education and solid experience in ophthalmology will be able to manually set the contours. Such a process would result in extracting information regarding optic diseases like glaucoma, diabetic retinopathy, age-related macular degeneration, retinal vascular occlusion. Index Terms—Optical images; retinal images; segmentation; optic nerve head; cup to disc ratio. I.INTRODUCTION Image processing operations such as segmentation, compression, watermarking has hugely boosted the medical area. With the growth of these techniques, it is possible to detect many diseases. Previously lot of research has been done to address abnormalities, using the digital retinal images. One of them is optic nerve segmentation. Optic nerve, which is an integral part of cranial nerve, courses from the eyeball to the brain. With the segmentation of the optic nerve head in retinal images, the ratio of the cup to disc [1] can be monitored. The part of the optic nerve which is clinically visible on examination can be described by optic disc. Although it might sound inaccurate as ‘disc’ refers to a flat, 2 dimensional structures without depth, whereas ‘optic nerve head’ is very much a 3 dimensional structure which should ideally be viewed stereoscopically. Optic disc [2] can be identified by orange- pink rim with a pale centre. This pale centre is devoid of neuro- retinal tissue and is known as the cup. Cup to disc ratio is the ratio of the vertical size of this cup and the disc as a whole. A cup to disc ratio of 0.3 (i.e. the cup occupies 1/3 of the height of the entire disc) is generally considered normal. If the cup to disc ratio is higher than 0.3, then it indicates a decrease in the quantity of healthy neuro-retinal tissue. This is also known as glaucomatous change. Recently, lot of work has been done on the optic nerve segmentation [3,4] to detect disease from digital retinal images. Some of the previous works regarding optic nerve segmentation includes optic nerve head segmentation method, which was described by Li and Chutatape [1] in the year 2001. They proposed a new method to localize optic disc center. The desired regions were first chosen by clustering the brightest region or pixels in retinal images, provided that there is no abnormality in the retina image. Principal component analysis was applied to these chosen regions. The minimum distance between the original retinal image and its projection onto disk space was located as the center of optic disc. In the same year, Barrett et. al. [2] used Hough transformation system which is capable of finding geometric shapes. Therefore, the circular shape of optic disc was distinguished using Hough transform and other thresholding and morphological algorithms. Lowell et al. (2004) proposed an algorithm [3] for the localization and segmentation of the optic nerve head boundary from low-resolution images. Foracchia et al. (2004) presented a new technique [4] for localizing the optic disc