Click here to


Are you sure ?

Yes, do it No, cancel

Robustness Evaluation of KV-Streaming Based Real-Time Vertebrae Tracking During Spine SBRT

A Qin1*, J Liang1 , J Zhou1 , M Lachaine2 , D Yan1 , (1) William Beaumont Hospital, Royal Oak, MI (2) Elekta Ltd, Montreal, QC


(Sunday, 7/29/2018) 5:05 PM - 6:00 PM

Room: Room 202

Purpose: To develop a real-time motion monitoring platform for Spine SBRT based intra-treatment KV-streaming and to evaluate the robustness of the rigid-registration based vertebrae tracking algorithm.

Methods: The XVI 5.0 came with the capability of streaming KV projections to a client workstation during delivery. An in-house software tool was developed to receive the KV-streaming for real-time target tracking. Briefly, a set of projections from pre-treatment CBCT is pre-processed to enhance the vertebral target and used as reference after applying couch correction. During VMAT delivery, the KV projections were received at about 5 Hz. In-plane two-Dimensional translational registration was conducted between the streamed and pre-treatment projections on a predefined rectangular region around isocenter. The output displacement is considered to be the intra-treatment target motion. The feasibility and accuracy of the system was verified on an Alderson Rando Phantom with manual couch movement. A comprehensive robustness evaluation was then carried out on seven spine SBRT patients with various anatomic target positions from T6 to L5. Specifically, for each projection angel, 10×10 artificial translations with step size of 0.518 mm was simulated on the streamed projections. The distribution of the resulting tracking error was analysis according to the target anatomic position and projection angle.

Results: The tracking algorithm was able to process the acquired projection in real-time (>5Hz) and got high accuracy (0.0 ± 0.1mm) on the phantom. For patient data, the overall accuracy is (0.56±0.37, 0.17±0.23)mm. Larger detecting uncertainty was found for a small portion of projections with diaphragm in tracking window and for few lateral projections with poor bony contrast.

Conclusion: The robustness of the rigid-registration based vertebral tracking depends on the target anatomic position, as well as projection angle. A robustness evaluation can be utilized to estimate the tracking uncertainty and exclude the small portion of angles with poor accuracy.

Funding Support, Disclosures, and Conflict of Interest: Supported by Elekta Research Grant


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email