Techniques for extraction of depth-resolved in vivo human retinal intrinsic optical signals with optical coherence tomography.

PURPOSE: To demonstrate acquisition and analysis methods for depth-resolved observation of slow retinal physiology induced changes in infrared backscatter in vivo.
METHODS: A dark-adapted human was briefly subjected to a localized photobleach. For 20 min before and 30 min after the stimulus, volumetric optical coherence tomograms were collected partially overlapping the bleached region. Tomograms were segmented into retinal layers by a newly described algorithm exploiting information in adjacent B-scans. En face fundus images extracted from major intraretinal layers were laterally registered manually. Time series summarizing the observed backscatter in selected layers for the bleached and unbleached areas are shown with a variety of corrections and normalizations applied: tomograms were corrected for inherent sensitivity roll-off, and the ratio between other layers and an assumed unchanging layer (retinal pigment epithelium), as well as the ratio of the stimulated area to the unstimulated area, were calculated.
RESULTS: Adjacent B-scan information allows a simpler segmentation algorithm to be used. Sensitivity roll-off correction reduces signal variability due to eye motion. After normalizations, the signal correlated with the stimulus appears strongest at the photoreceptor inner-outer segment junction.
CONCLUSIONS: Demonstrated methods manage data complexity and reduce uncorrelated signal variability. This single trial warrants further investigation of intrinsic optical signals to observe slow physiologic responses.