Optic Disc Segmentation Using Manual Thresholding Technique

Abstract

Optic disc segmentation is a very important step in diagnosing glaucoma disease, where untreated glaucoma leads to irreversible loss of vision. Glaucoma is the second cause of blindness around the world. In this work, manual thresholding technique in optic disc segmentation is applied, evaluated, and compared with other segmentation techniques. The optic disc segmented by thresholding level 180 after vessel removal by morphological operation to improve segmentation process, and then boundary smoothed by dilating and eroding operation and border clearing; at final segmentation step, the segmented part was shaped to construct full circle to minimize segmentation error due to the main vessel and presence of parapapillary atrophy around the disc. The algorithm was evaluated on rim:one database, which has a ground truth from 5 experts, and achieved best Dice coefficient results with experts, 90%, and structural similarity, 83%. It was concluded that a novel optic disc segmentation algorithm by manual thresholding technique is easy, fast, and inexpensive computationally and obtained good segmentation results and would help in early detection of glaucoma by the ophthalmologist.