Superpixel-Based Optic Nerve Head Segmentation Method of Fundus Images for Glaucoma Assessment

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Vision

Background:

• Glaucoma is a leading cause of irreversible blindness globally, characterized by progressive optic neuropathy.
• Elevated intraocular pressure is a primary risk factor for glaucoma development and progression.
• Current clinical diagnosis relies heavily on manual examination of fundus retinal images, specifically the optic nerve head (ONH).

Purpose of the Study:

• To develop an unsupervised, automated method for segmenting the optic nerve head (ONH) in retinal fundus images.
• To enable accurate ellipse fitting for ONH contour detection.
• To provide a reliable tool for glaucoma screening.

Main Methods:

• An unsupervised superpixel-based approach was employed for ONH segmentation.
• Linear iterative clustering was utilized for automatic ONH contour detection via ellipse fitting.
• The method was validated using a public dataset with expert-annotated ground truths and a glaucoma-classified database.

Main Results:

• The automated segmentation method achieved comparable ellipse fitting results to expert ground truths.
• Results fell within the statistical range of inter-observer variability, demonstrating reliability.
• The developed tool provides fast and user-friendly results for ONH analysis.

Conclusions:

• The proposed unsupervised method offers an accurate and efficient approach for ONH segmentation in retinal images.
• This automated tool can assist clinicians in objective glaucoma screening and diagnosis.
• The program provides a reliable and accessible solution for analyzing fundus images for glaucoma detection.