Glaucoma Detection from Retinal Images Using Statistical and Textural Wavelet Features.

Area of Science:

• Ophthalmology
• Medical Imaging
• Signal Processing

Background:

• Glaucoma is a leading cause of irreversible blindness, characterized by silent, progressive optic nerve damage.
• Early detection and treatment are crucial to prevent severe vision impairment.
• Automated screening systems can assist ophthalmologists by identifying glaucoma early.

Purpose of the Study:

• To propose a wavelet-based glaucoma detection algorithm for real-time screening.
• To evaluate the algorithm's performance using statistical and textural features from retinal images.
• To assess the algorithm's suitability for automated, rapid glaucoma diagnosis.

Main Methods:

• A glaucoma detection algorithm utilizing wavelet-based statistical and textural features from the optic disc region was developed.
• The algorithm was tested on two public retinal image datasets of varying resolutions.
• Performance was evaluated using metrics such as accuracy and Area Under the Receiver Operating Curve (AUC).

Main Results:

• The algorithm achieved 96.7% accuracy and 94.7% AUC on a high-resolution dataset.
• Analysis confirmed the relevance of wavelet-based features for glaucoma detection.
• The algorithm processes high-resolution retinal images in under 3 seconds, indicating real-time capability.

Conclusions:

• The proposed wavelet-based algorithm is effective for automated glaucoma detection in retinal images.
• The algorithm demonstrates high accuracy and efficiency, suitable for real-time screening systems.
• This approach can aid in the early identification of glaucoma, potentially reducing blindness.