Blood vessel segmentation with three-dimensional spectral domain optical coherence tomography

摘要

In the context of the early detection and monitoring of eye diseases, such as glaucoma and diabetic retinopathy, the use of optical coherence tomography presents the difficulty, with respect to blood vessel segmentation, of weak visibility of vessel pattern in the OCT fundus image. To address this problem, a boosting learning approach uses three-dimensional (3D) information to effect automated segmentation of retinal blood vessels. The automated blood vessel segmentation technique described herein is based on 3D spectral domain OCT and provides accurate vessel pattern for clinical analysis, for retinal image registration, and for early diagnosis and monitoring of the progression of glaucoma and other retinal diseases. The technique employs a machine learning algorithm to identify blood vessel automatically in 3D OCT image, in a manner that does not rely on retinal layer segmentation.

主权项

1. A method of training an algorithm to identify automatically a blood vessel in a three-dimensional optical coherence tomography image, comprising:
(A) obtaining at least one fundus image generated from A-scans of a plurality of three-dimensional optical coherence tomographic images containing said vessel; (B) labeling each pixel in said fundus image as vessel containing or non vessel containing; (C) generating at least one training data set by selecting randomly from said plurality a first number of said A-scans corresponding to said respective vessel containing pixels and a second number of said A-scans corresponding to said respective non vessel containing pixels; (D) for each A-scan in said training data set, extracting features comprising at least one two-dimensional feature and at least one one-dimensional feature; (E) generating an ensemble classifier by iteratively comparing said features with said corresponding labeled pixels, such that said ensemble classifier is trained to identify automatically said vessel.