Motion-predictive retinal optical coherence tomography during large-scale head and body movement

Motion is the longstanding enemy of high-quality ophthalmic imaging with optical coherence tomography (OCT). To suppress motion, subjects are restricted to static, seated postures; patients unable to satisfy such requirements frequently cannot undergo OCT imaging. Leveraging our prior robotic OCT work, we propose a predictive motion planning pipeline that uses face and pupil trackers and autoregressive filters to complete imaging despite large-scale head movements. We evaluate its prediction capability through retinal imaging of human subjects during whole body movements. Our algorithm demonstrates fast and robust motion prediction capabilities for subjects in different motion profile, recovering full retinal scans.