Visible-light optical coherence tomography oximetry based on circumpapillary scan and graph-search segmentation

Abstract

Visible-light optical coherence tomography (vis-OCT) enables retinal oximetry by measuring the oxygen saturation of hemoglobin (sO₂) from within individual retinal blood vessels. The sO₂ calculation requires reliable estimation of the true spectrum of backscattered light from the posterior vessel wall. Unfortunately, subject motion and image noise make averaging from multiple A-lines at the same depth position challenging, and lead to inaccurate sO₂ estimation. In this study, we developed an algorithm to reliably extract the backscattered light's spectrum. We used circumpapillary scanning to sample the vessels repeatedly at the same location. A combination of cross-correlation and graph-search based segmentation extracted the posterior wall locations. Using measurements from 100 B-scans as a gold standard, we demonstrated that our method achieved highly accurate measures of sO₂ with minimal bias. In addition, we also investigated how the number of repeated measurements affects the accuracy of sO₂ measurement. Our method sets the stage for large-scale studies of retinal oxygenation in animals and humans.