Two-dimensional segmentation of the retinal vascular network from optical coherence tomography

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Vision

Background:

• Automated retinal vascular network segmentation is crucial for diagnosing eye diseases.
• Existing methods primarily focus on traditional imaging modalities and the optic nerve head region.
• Optical coherence tomography (OCT) presents unique challenges for vascular segmentation due to resolution, contrast, and noise.

Purpose of the Study:

• To develop and validate a supervised-learning method for automatic retinal vascular segmentation using standard OCT data.
• To address the complexities of segmenting the macular vascular network from OCT images.

Main Methods:

• Extraction of features from standard 3D OCT macular scans.
• Application of a supervised-learning algorithm (support vector machines) for segmentation.
• Validation on OCT scans from healthy subjects and diabetic patients, including the optic nerve head region.

Main Results:

• The proposed method achieved 98% accuracy, 99% specificity, and 83% sensitivity for macular vascular segmentation.
• Similar high performance was observed when tested on OCT data of the optic nerve head region.

Conclusions:

• The developed feature-based, supervised-learning approach effectively segments the retinal vascular network from OCT macular scans.
• This method offers a robust solution for analyzing retinal vasculature in both healthy and diseased eyes using OCT imaging.