Intraoperative 3-D Reconstruction and Geometric Modeling Using Sensorized Microsurgical Instruments

Precise micrometer-scale tissue manipulation is essential for successful microsurgery. Despite advancements in intraoperative sensing and instrument tracking, providing accurate real-time feedback on instrument positions relative to critical tissue structures remains challenging. This research aims to reliably reconstruct fine-grained anatomical surfaces intraoperatively using a smart instrument with fiber-based distance sensing. First, instrument-integrated optical coherence tomography sensor readings are integrated with instrument tracking to reconstruct accurate surface point clouds. Subsequently, radial basis functions, ordinary Kriging, and B-splines are utilized for surface modeling, employing tailored basis functions and smoothing parameters. The proposed methods are validated in the context of vitreoretinal surgery through simulations and with ex vivo studies involving porcine and human cadaver eyes. In simulations, radial basis function interpolation demonstrates the highest robustness across varying noise levels and eye shapes. In ex vivo studies, ordinary Kriging outperformed, yielding root mean square errors of 14.7 μm for the porcine eye and 38.1 μm for the human eye, with 99.7% and 96.5% of absolute errors below 100 μm. The proposed methods effectively model featurerich micro-anatomical surfaces and enable model-based assistance and automation, ultimately enhancing surgical precision in delicate ophthalmic procedures.