Dosimetric Improvement via Real-time Dynamic Couch-based Intra-fraction Motion Tracking for 3D Conformal and IMRT Plans

To investigate the dosimetric improvement achieved via real-time motion tracking with a newly developed dynamic treatment couch. We retrospectively selected 3D conformal and IMRT treatment plans for 3 SBRT lung tumor cases. Planning was performed using the RTOG 0236 guidelines with a 7 mm margin to form the PTV. The 3D conformal and IMRT plans involved 15 and 7 beams, respectively. The prescription dose was 20 Gy per fraction using 6MV photon beams. Two 3D time-varying (and non-periodic) tumor trajectories with average peak-peak displacements of 1.5 cm and 2.5 cm were also selected. Solid water blocks, 14 cm thick, were placed on a 4D phantom capable of reproducing real 3D tumor trajectories to mimic the motion of a mobile tumor. The treatment plans were delivered under the following 3 scenarios. First they were delivered on stationary targets and film measurements performed to establish reference dose distributions. To simulate the tumor motion, the 4D phantom was made to move along selected tumor displacement trajectories in 3D. Finally, we used newly developed dynamic couch to correct for real-time phantom displacement. An infrared marker placed on the moving phantom was tracked using a camera system which provided feedback to the couch. Feedback-based dynamic couch control was employed to correct for the instantaneous phantom displacements to maintain a fixed position relative to the beam. EDR2 film was sandwiched between solid water blocks at a depth of 7 cm. To accommodate the dynamic range of the film, MUs were scaled down by a factor of 10. With 3 plans (3 SBRT and 3 IMRT) and 2 tumor trajectories we delivered a total of 30 plans. To compare the quality of plan delivered, gamma analysis was performed on film measurements under the above described 3 scenarios. We found that the mean geometric error in tracking on the first tumor trajectory (with an average peak-to-peak tumor displacement of 1.5 cm) was 1.2 mm. In comparison with the stationary case, the moving phantom case had 37% of the points with (gamma less than or equal to 1) (3%/3 mm) averaged over all the plans. With motion compensation the percentage of (gamma less than or equal to 1) increased to 89%. For the case with 2.5 cm peak-to-peak displacement, the mean geometric error was 2.1 mm. In this case compared with the stationary case, the moving phantom case with motion compensation resulted in percentage of increasing to 81% (again averaged over all the plans) from 21% with no motion compensation. Our results show that real-time target tracking using the newly developed dynamic couch improved the quality of treatment plans relative to the motion with no tracking cases. The tracking system reduced the residual geometric error to less than 10% of the original value.