Topology Regulated Background Extraction (TRBE) method for eye fundus images

Area of Science:

• Computational ophthalmology
• Medical image analysis
• Digital fundoscopy

Background:

• Preprocessing digital fundoscopy images requires identifying relevant pixels.
• Background extraction is a key step to reduce computational load.
• Existing methods may struggle with noise and suboptimal image quality.

Purpose of the Study:

• To present a novel background extraction method for digital fundoscopy images.
• To improve the efficiency of computer-aided diagnosis (CAD) systems.
• To enhance the focus on relevant retinal information.

Main Methods:

• Utilizing computational topology to analyze binarized fundoscopy images as cubical complexes.
• Extracting 1 and 2-dimensional homological groups.
• Identifying luminescence values to create a binarization mask for background segmentation.

Main Results:

• Successfully segmented the background of digital fundoscopy images.
• Generated a mask to isolate informative pixels (eye fundus).
• Demonstrated robustness against image noise and quality degradation.

Conclusions:

• The computational topology-based method effectively extracts backgrounds from fundoscopy images.
• This approach facilitates subsequent analysis by focusing on relevant pixels.
• The method is suitable for preprocessing stages in computational ophthalmology and CAD systems.