Bonferroni Mean Pre-aggregation Operator Assisted Dynamic Fuzzy Histogram Equalization for Retinal Vascular...

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Vision

Background:

• Automated retinal vasculature segmentation is crucial for diagnosing eye diseases.
• Current methods struggle with variations in vessel structure, low contrast, and pathologies.
• Existing deep learning approaches often overlook vessel geometry, leading to segmentation inaccuracies.

Purpose of the Study:

• To introduce a novel unsupervised method for retinal vessel segmentation.
• To improve accuracy and efficiency in segmenting vascular features from fundus images.
• To overcome limitations of existing hand-crafted and deep learning-based segmentation techniques.

Main Methods:

• Proposed BMPDFHESeg method utilizes an unsupervised approach based on interrelationship handling and a Bonferroni mean pre-aggregation operator.
• Employs dynamic fuzzy histogram equalization for vessel feature enhancement.
• Fuses color channels to extract vascular information and identify vessel direction.

Main Results:

• BMPDFHESeg demonstrated enhanced efficacy and computational speed in qualitative and quantitative assessments.
• The method achieved accurate vessel segmentation on DRIVE, STARE, and HRF datasets.
• Ophthalmologists validated the segmentation accuracy and diagnostic usefulness for retinal disorders.

Conclusions:

• The proposed BMPDFHESeg method offers an effective unsupervised solution for retinal vasculature segmentation.
• This technique improves upon existing methods by incorporating geometric characteristics and enhancing feature extraction.
• BMPDFHESeg shows significant potential for aiding in the early diagnosis and management of retinal diseases.