Robotic-Assisted High-Precision 3D OCT Reconstruction for Post-Anastomosis Vascular Evaluation

Optical coherence tomography (OCT), due to its high-resolution, non-invasive, three-dimensional imaging capabilities, has been extensively applied in medical diagnostics and disease assessment. Currently, OCT is recognized as the gold standard in ophthalmology for diagnosing and preventing ocular diseases. However, the limited field-of-view (FOV) inherent to traditional OCT systems restricts their application to relatively small and well-defined regions. To overcome this limitation, robotic-assisted OCT systems have emerged, significantly expanding the imaging FOV by combining robotic arm with OCT imaging. In this paper, aiming for precise post-anastomosis evaluation in vascular surgery, we propose a high-precision three-dimensional (3D) reconstruction method based on robotic-assisted OCT. We combine Doppler signals representing blood flow dynamics as supplementary input features to enhance segmentation accuracy, specifically targeting precise delineation of both the inner and outer vascular walls. A deep neural network leverages these combined features to segment vascular structures from surrounding tissue in OCT cross-sectional images. Guided by robotic-arm pose data and a fusion registration algorithm, the segmented OCT slices are co-registered and stitched to yield a wide-field, high-accuracy 3D vascular reconstruction. To validate our approach, we constructed a custom dataset comprising OCT imaging data collected from mouse femoral and tail arteries following vascular anastomosis procedures. The experimental results demonstrate the high accuracy and robustness of the proposed 3D reconstruction technique. This work presents implications for robotic-assisted vascular anastomosis, potentially enhancing intraoperative and postoperative assessments, thus contributing valuable technological advancements for future robotic surgical applications.