Hybrid Respiratory Motion Tracking Using X-Ray Imaging and Integrated Infrared Cameras In a Mobile Cone-Beam CT System
Abstract
Purpose
To assess the agreement in motion tracking at a mobile cone-beam computed tomography (CBCT) X-ray imaging system integrating optical cameras, at varying gantry angle and controlled breathing periods.
Methods
A metal marker was attached to the moving rod in the CIRS dynamic motion phantom, with a cos4 motion trace and breathing periods set to 2.55, 3, 4, and 5 s. Eight CBCT datasets were acquired at 80 kVp and 11.9 frames per second for a total duration of 60 s with full 360° gantry rotation using two filtering windows for the optical cameras: 350 ms, to generate smoother motion trajectories, and 80 ms, to capture higher temporal resolution. A quantitative evaluation was performed using the root mean square error (RMSE) with respect to the nominal motion trace. An image-enhanced template-matching (TM) algorithm was used to track the centroid of the metal marker in CBCT projection images.
Results
Marker trajectories derived from CBCT projections demonstrated maximum deviations of 0.7 mm, in good agreement with the accuracy of the optical tracking system (0.75 mm). The X-ray source rotation in CBCT acquisition introduced a sinusoidal component in the TM-derived motion signal, which was corrected by detecting the plateau regions in the cos4 breathing trace. The 350 ms filtering window resulted in reduced mismatch (up to 0.25 mm difference) between tracked and reference trajectories for the optical cameras, due to decreased noise in comparison to the 80 ms window. RMSE increased inversely to the breathing period, due to higher frequency motion.
Conclusion
Mobile CBCT enables tumour motion tracking with accuracy comparable to infrared-based tracking under realistic breathing conditions. Tracking accuracy degrades for optical tracking cameras in faster breathing motion, highlighting the importance of data filtering for clinically robust motion tracking.