Structure-Preserving Guided Retinal Image Filtering and Its Application for Optic Disk Analysis

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Science

Background:

• Retinal fundus photography is crucial for diagnosing ocular diseases like glaucoma and diabetic retinopathy.
• Image quality degradation, caused by factors like cataracts, hinders accurate retinal image analysis.
• Automated computer-aided diagnosis systems require high-quality retinal images for reliable performance.

Purpose of the Study:

• To develop a novel image filtering technique to restore degraded retinal fundus images.
• To address challenges posed by image attenuation and scattering in retinal imaging.
• To improve the accuracy of subsequent automated analysis tasks for ocular disease diagnosis.

Main Methods:

• Approximated retinal image degradation as a combination of human-lens attenuation and scattering.
• Developed a structure-preserving guided retinal image filtering (SGRIF) method.
• SGRIF incorporates global structure transferring and edge-preserving smoothing.

Main Results:

• SGRIF effectively improved retinal image contrast, validated by histogram-based metrics.
• Enhanced image quality was demonstrated through improved histogram flatness, spread, and local luminosity variability.
• The filtering method significantly benefited downstream tasks, including optic cup segmentation and cup-to-disk ratio (CDR) computation.

Conclusions:

• The proposed SGRIF method successfully restores degraded retinal images by modeling attenuation and scattering.
• Improved image quality using SGRIF leads to more accurate optic cup segmentation and CDR measurements.
• This technique has the potential to enhance the reliability of computer-aided diagnosis for various ocular conditions.