A Smart Projective Imaging and Navigation System for Oral and Maxillofacial Surgery

In oral and maxillofacial surgery (OMS), surgical outcomes often depend significantly on the subjective experience of the surgeon. While conventional navigation systems, such as those using guide plates or fiducial markers, offer some guidance, they are hampered by instability, obtrusive hardware, and limited accuracy. Markerless navigation presents a promising alternative, but current approaches are constrained by a reliance on specific textures, scarce training data, and inefficient visualization.

This paper introduces a Smart Projective Imaging and Navigation System (SPIN) that integrates synthetic data-driven learning, coaxial augmented-reality (AR) projection, and real-time structured-light 3D scanning for markerless dynamic surgical navigation. Our approach utilizes no more than two teeth as natural markers to enable high-precision pose estimation, even in low-texture regions. To overcome the data scarcity challenge, we develop a preoperative platform that generates a diverse synthetic dataset for robust AI model training, eliminating the dependency on real surgical imagery.

Experiments on dental phantoms, patient-specific phantoms, and representative clinical cases demonstrate that the SPIN system achieves submillimeter accuracy and exhibits strong robustness.

The proposed SPIN system provides an accurate, intuitive, and markerless navigation solution for OMS.

This work establishes a novel and clinically viable framework for dynamic surgical navigation, demonstrating significant potential for broad adoption in real-world surgical environments.