Deep learning based 2.5D flow field estimation for maximum intensity projections of 4D optical coherence tomography

摘要

In microsurgery, lasers have emerged as precise tools for bone ablation. A challenge is automatic control oflaser bone ablation with 4D optical coherence tomography (OCT). OCT as high resolution imaging modalityprovides volumetric images of tissue and foresees information of bone position and orientation (pose) as well asthickness. However, existing approaches for OCT based laser ablation control rely on external tracking systemsor invasively ablated artificial landmarks for tracking the pose of the OCT probe relative to the tissue. Thiscan be superseded by estimating the scene flow caused by relative movement between OCT-based laser ablationsystem and patient.Therefore, this paper deals with 2.5D scene flow estimation of volumetric OCT images for application inlaser ablation. We present a semi-supervised convolutional neural network based tracking scheme for subsequent3D OCT volumes and apply it to a realistic semi-synthetic data set of ex vivo human temporal bone specimen.The scene flow is estimated in a two-stage approach. In the first stage, 2D lateral scene flow is computed oncensus-transformed en-face arguments-of-maximum intensity projections. Subsequent to this, the projections arewarped by predicted lateral flow and 1D depth flow is estimated. The neural network is trained semi-supervisedby combining error to ground truth and the reconstruction error of warped images with assumptions of spatialflow smoothness. Quantitative evaluation reveals a mean endpoint error of (4:7 - 3:5) voxel or (27:5 - 20:5) μmfor scene flow estimation caused by simulated relative movement between the OCT probe and bone. The sceneflow estimation for 4D OCT enables its use for markerless tracking of mastoid bone structures for image guidancein general, and automated laser ablation control.