Sub-millimeter spine position monitoring for stereotactic body radiotherapy using offline digital tomosynthesis

Purpose: Spine stereotactic radiotherapy (SBRT) requires intrafraction motion <1–2 mm. We evaluated the accuracy and precision of digital tomosynthesis (DTS) in combination with triangulation for spine position tracking. Materials/methods: Single-slice DTS images were generated from kV cone beam CT (CBCT) projection images. They were registered to reference DTS images reconstructed from the planning CT-scan to determine 2D shifts between actual patient position and treatment plan position. 3D spine position was obtained by triangulation of each registration with a previous registration, for every 1° of data. For 7 patients who underwent spine SBRT, the standard deviation (SD) of DTS + triangulation over one entire dataset was evaluated for different DTS angles (2–16°) and triangulation angles (1–46°). For 32 CBCT datasets, acquired before or after treatment of the 7 patients, using 4° DTS and 18° triangulation angle, SDs were determined and average positions were compared to clinically performed CBCT registrations. Results: Mean SDs were 0.29 ± 0.10 mm for lateral (range 0.1–0.55 mm), 0.14 ± 0.08 for longitudinal (0.05–0.39) and 0.24 ± 0.10 for the vertical direction (0.10–0.57). Lateral and vertical SDs for thoracic spine were higher than for lumbar spine. Differences between clinical CBCT registration and DTS + triangulation were 0.1 ± 0.26, 0.02 ± 0.33 and −0.07 ± 0.21 mm. Conclusion: The combination of DTS and triangulation allows for monitoring spine position with sub-mm accuracy and precision