Translator Disclaimer
9 March 2010 3-D segmentation of retinal blood vessels in spectral-domain OCT volumes of the optic nerve head
Author Affiliations +
Abstract
Segmentation of retinal blood vessels can provide important information for detecting and tracking retinal vascular diseases including diabetic retinopathy, arterial hypertension, arteriosclerosis and retinopathy of prematurity (ROP). Many studies on 2-D segmentation of retinal blood vessels from a variety of medical images have been performed. However, 3-D segmentation of retinal blood vessels from spectral-domain optical coherence tomography (OCT) volumes, which is capable of providing geometrically accurate vessel models, to the best of our knowledge, has not been previously studied. The purpose of this study is to develop and evaluate a method that can automatically detect 3-D retinal blood vessels from spectral-domain OCT scans centered on the optic nerve head (ONH). The proposed method utilized a fast multiscale 3-D graph search to segment retinal surfaces as well as a triangular mesh-based 3-D graph search to detect retinal blood vessels. An experiment on 30 ONH-centered OCT scans (15 right eye scans and 15 left eye scans) from 15 subjects was performed, and the mean unsigned error in 3-D of the computer segmentations compared with the independent standard obtained from a retinal specialist was 3.4 ± 2.5 voxels (0.10 ± 0.07 mm).
© (2010) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Kyungmoo Lee, Michael D. Abràmoff, Meindert Niemeijer, Mona K. Garvin, and Milan Sonka "3-D segmentation of retinal blood vessels in spectral-domain OCT volumes of the optic nerve head", Proc. SPIE 7626, Medical Imaging 2010: Biomedical Applications in Molecular, Structural, and Functional Imaging, 76260V (9 March 2010); https://doi.org/10.1117/12.844330
PROCEEDINGS
8 PAGES


SHARE
Advertisement
Advertisement
Back to Top