Segmentation of ophthalmic Optical Coherence Tomography images using graph cuts

Published

Book Section

We describe an efficient approach for the automated segmentation of pathological/morphological structures in ophthalmic Spectral Domain Optical Coherence Tomography (SDOCT) images. In this algorithm, image pixels are treated as nodes of a graph with edge weights assigned to associate pairs of pixels. The weights vary according to the distances, brightness differences, and feature variations between pixel pairs. Cuts through the graph with minimum accumulated weights correspond to morphological layer boundaries. This approach has been applied to SDOCT images with encouraging results and thus forms an adaptable framework for the segmentation of many different ophthalmic structures. © 2010 SPIE.

Full Text

Duke Authors

Cited Authors

  • Li, XT; Chiu, SJ; Nicholas, P; Toth, CA; Izatt, JA; Farsiu, S

Published Date

  • December 1, 2010

Volume / Issue

  • 7550 /

International Standard Book Number 13 (ISBN-13)

  • 9780819479464

Digital Object Identifier (DOI)

  • 10.1117/12.842299

Citation Source

  • Scopus