An Online Calibration Method for Microsoft Hololens2 in Surgical Navigation System

Challenges with hand-eye coordination limit display-based navigation systems, while augmented reality (AR) offers a promising alternative. However, achieving accurate and stable alignment between virtual space and real space remains a challenge. Developing a fast, accurate, and robust virtual-real calibration method is crucial for advancing the application of AR in the medical field.

We propose an automatic online calibration method for Microsoft Hololens2. First, a calibration module containing identifier image and optical spheres is designed for the automatic acquisition of the virtual-real space transformation. Then, real-time calibration combined with window sliding filtering updates and smooths the calibration results, providing the operator with a continuous, accurate, and smooth navigation screen. Finally, an AR-based surgical navigation system is developed and experimental validation is carried out.

Experiments using a 3D printed spine model were conducted to assess calibration accuracy. Results showed that the point distance error was 1.34 ± 0.43 mm, and the accuracy of the virtual-real fusion did not drift significantly over time. Two ex vivo porcine spines were used to evaluate the drilling accuracy of the AR navigation system. At the 25 planned drilling positions, the nail placement success rate was 96% (Gertzbein-Robbins grades A and B). The average placement point deviation and angular deviation were 1.09 ± 0.35 mm and 1.45 ± 0.64°, respectively.

Our calibration method helps surgeons achieve accurate and efficient virtual-real alignment through simple operations. The results of model and ex vivo animal experiments demonstrate the accuracy, efficiency, and feasibility of the proposed system.