Robust real-time bronchoscope electromagnetic tracking using an adaptive hybrid registration approach without fiducials

This paper proposes a new approach of adaptive hybrid registration without fiducials to electromagnetically track a bronchoscope in real time. Bronchoscope tracking seeks to obtain the position and orientation of the bronchoscope tip in a reference coordinate system. An electromagnetic tracker with a sensor attached to the bronchoscope tip is usually used. However, the spatial transformation between the electromagnetic tracker and reference coordinate systems must be determined. Although current fiducial-free methods work well, they are still constrained on their initialization in optimization and an assumption of operating the bronchoscope along the bronchial centerlines. To address the constraints, we explore an adaptive hybrid alignment framework that uses adaptive point selection and a hybrid objective function during optimization. The experimental results demonstrate that our proposed framework is much more accurate than previous methods. The current tracking error was reduced from 5.4 to 2.5 mm, which further approximates the clinical requirement of 2.0 mm.