Identification of the optic nerve head with genetic algorithms

Area of Science:

• Ophthalmology
• Medical Imaging
• Computational Intelligence

Background:

• Accurate localization and segmentation of the optic nerve head (ONH) are crucial for diagnosing and monitoring optic neuropathies like glaucoma.
• Existing methods may lack robustness or require manual intervention, highlighting the need for automated solutions.

Purpose of the Study:

• To develop and validate an automatic system for optic nerve head (ONH) detection and segmentation in eye fundus images.
• To utilize genetic algorithms for efficient and accurate ONH localization and contour approximation.

Main Methods:

• A heuristic-driven approach was employed, starting with hypothesis point generation based on image properties.
• A genetic algorithm optimized an elliptical fit to approximate the ONH contour, considering geometric constraints.
• The method was tested on 110 fundus images from patients with glaucoma and eye hypertension.

Main Results:

• The genetic algorithm successfully identified elliptical approximations of the ONH.
• The system achieved high accuracy, with 96% and 99% of images showing a discrepancy delta < 5 on internal and external datasets, respectively.
• The ONH contour was effectively represented by a non-deformable ellipse.

Conclusions:

• The proposed method offers a robust and generalizable approach for automatic ONH segmentation.
• The system directly provides clinically relevant parameters such as axis lengths, center location, and orientation.
• Results are competitive with existing literature, supporting the use of genetic algorithms in ophthalmic image analysis.