Collaborative Surface Reconstruction and Tracking with Infrared Coding Multiplexing for Image-Guided Surgery

Rapid and high-precision preoperative patient-To-image registration is a crucial challenge in surgical navigation. Existing navigation systems acquire point clouds using marker-based probes or scanners with attached optical markers. However, these methods inevitably cause skin deformation and human factor interference, reducing the efficiency and stability of the registration. Moreover, the heavy reliance on preoperative Computed Tomography (CT) images or scanners for surface reconstruction significantly increases medical costs. To address these issues, this paper proposes a novel collaborative framework for surface reconstruction and tracking that integrates the 2D Gaussian Splatting (2DGS) technology. 2DGS enables high-quality and information-rich surface reconstruction using only a visible light camera, thereby decreasing the reliance on CT or scanners. We propose infrared coding multiplexing technology, which integrates an infrared digital projector with optical tracking system to achieve tracking of the patient and surgical instruments without multisensor calibration. Additionally, we recover the scale of the 2DGS reconstruction model using a Scale-Adaptive Iterative Closest Point registration algorithm. We validated the effectiveness of the system through a phantom head model experiment. While providing richer texture information, the surface target localization error of the 2DGS model was 1.14 mm, which is comparable to the scanner model. This demonstrates the potential value of our method for partial open surgeries and skin surgeries.