Automatic optic disc segmentation with peripapillary atrophy elimination.

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Vision

Background:

• Accurate optic disc segmentation is crucial for automated glaucoma diagnosis.
• Existing methods can be hindered by structures like peripapillary atrophy.
• Glaucoma diagnosis relies on features derived from the optic disc and cup.

Purpose of the Study:

• To propose an optic disc segmentation method that effectively eliminates peripapillary atrophy.
• To improve the accuracy of optic disc segmentation in retinal fundus images.
• To provide a reliable method for calculating features like the cup-to-disc ratio for glaucoma diagnosis.

Main Methods:

• A novel optic disc segmentation approach utilizing peripapillary atrophy elimination.
• Implementation of edge filtering, constraint elliptical Hough transform, and peripapillary atrophy detection.
• Exclusion of non-disc structure edges, particularly peripapillary atrophy, to refine segmentation.

Main Results:

• The method was evaluated on a dataset of 650 manually annotated retinal fundus images.
• Achieved average performance metrics of m(1) = 10.0%, m(2) = 7.4%, and m(VD) = 4.9%.
• Demonstrated improved accuracy in optic disc segmentation by removing confounding structures.

Conclusions:

• The proposed method offers a robust solution for optic disc segmentation in retinal images.
• Effective peripapillary atrophy elimination leads to more accurate segmentation results.
• The method facilitates the computation of key features for automated glaucoma diagnosis systems.