Respiratory motion compensation for image-guided bronchoscopy using a general motion model

In this paper, we investigate the use of a general motion model for motion compensation of CT/X-ray fusion during image-guided bronchoscopy, where respiratory motion causes a misalignment between the pre-operative CT and the intra-operative X-ray images. The general motion model is learned from four-dimensional images of different patients and captures statistics of lung motion. For motion compensation, the general model can then be individualized using the pre-operative CT, while a surrogate to drive the model can be derived from the X-ray images. Quantitative evaluation was carried out on 4D-CT data. Virtual projections were generated from the 4D-CT data to simulate the intra-operative X-ray images of different breathing states. The average prediction error between maximal inhale and maximal exhale was found to be 4.06 ± 1.64 mm, which means that the model compensated for 56% of the motion amplitude.