A novel multiple subdivision-based algorithm for quantitative assessment of retinal vascular tortuosity.

Area of Science:

• Ophthalmology
• Medical Imaging
• Biomedical Engineering

Background:

• Retinal vascular tortuosity is a key indicator of morphological changes and a potential biomarker for diseases like diabetes.
• Current methods for analyzing vascular tortuosity lack consistency with expert evaluations, hindering clinical application.

Purpose of the Study:

• To develop and validate a novel algorithm for quantitative analysis of retinal vascular tortuosity.
• To improve the consistency and accuracy of tortuosity assessment compared to human experts.

Main Methods:

• A multiple subdivision-based algorithm was developed for vessel segment tortuosity analysis.
• The algorithm incorporates a learning curve function of vessel curvature inflection points to capture local and global features.
• The method was evaluated using retinal fundus images.

Main Results:

• The algorithm demonstrated high correlation coefficients: 0.931 for arteries and 0.925 for veins against clinical grading.
• The prognostic performance against expert predictions showed an area under the receiver operating characteristic curve of 0.968.
• This indicates strong consistency with expert assessments in evaluating the entire retinal vascular network.

Conclusions:

• The proposed algorithm offers a reliable and consistent method for quantifying retinal vascular tortuosity.
• This advancement has the potential to significantly aid in the early diagnosis and monitoring of diabetic retinopathy and other vascular diseases.
• The algorithm's ability to mimic expert assessment enhances its clinical utility.