Due to the sequential, one-dimensional mechanical scanning of optical coherence tomography (OCT), a rolling shutter acquisition mode is induced that produces geometric artefacts due to the relative motion between the imaged object and the OCT probe. In this paper, we propose a technique that exploits motion artefacts in OCT volumes, to estimate both the 3D shape and motion of an arbitrary sample. Our proposed method begins by mathematically modelling motion artefacts in OCT in the 3D Cartesian space SE(3). Subsequently, we develop a methodology to estimate the motion using a volume similarity method that compares a reference OCT volume with a motion-distorted OCT volume, as a whole, rather than pointwise through 3D-to-3D point matching. The estimated motion is then utilised to correct any artefacts present in the motion-distorted volume. This methodology is then enhanced to estimate both the shape and motion from only one OCT volume acquired using the rosette non-Cartesian scan trajectory (without prior knowledge of the shape). Our results reveal that accurate shape and motion estimation are achievable using both numerically simulated and experimental OCT data, demonstrating the robustness of our method.
Osinde et al. (Sun,) studied this question.