J Mechalakos, L Zhang, N Shah, Y Hu, P Zhang, A Ballangrud, L KUO*, J Deasy, L Cervino, Memorial Sloan Kettering Cancer Center, New York, NY

• 
  L KUO

• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  

PO-GePV-T-275 (Sunday, 7/25/2021)   [Eastern Time (GMT-4)]

Purpose: To estimate dosimetric uncertainty in registration-based dose accumulation for spine reirradiation and propose a novel method to account for this uncertainty.

Methods: Dose accumulation was performed for 3 SBRT spine reirradiation patients using 3 methods. First, a rigid vertebral registration between primary and secondary scans was used. Second, the rigid registration was independently perturbed 2-3 mm in the superior, inferior, anterior, posterior, left and right directions to estimate the effect of registration uncertainty on results. Third, an in-house API script (RAdiotherapy Dose Accumulation Routine, RADAR) was used. RADAR accounts for registration uncertainty by associating with each dose voxel in the secondary scan the maximum dose within a small search space of user specified dimensions centered on that voxel. The RADAR search dimensions were set equal to the perturbations from method 2 for comparison purposes. Results of the 3 methods were compared using maximum spinal cord (patients 1 and 2) or cauda (patient 3) dose (Dmax). RADAR was run for 2 use cases: “unrestricted”, where the entire search space was considered for each secondary voxel and “restricted”, where the search space for each voxel was limited to the portion within the cord/cauda contours.

Results: Dmax for each patient respectively was 3790/2463/3548 cGy using method 1. Dmax increased to 4062/2555/3787 cGy (SI perturbation), 4904/3808/3773 cGy (LR perturbation), and 4955/3538/3735 cGy (AP perturbation) using method 2. Using RADAR, Dmax was 4163/2595/3802 cGy (SI), 4878/3723/3751 cGy (LR), and 4878/3477/3777 cGy (AP) for the unrestricted use case and 3947/2496/3704 cGy (SI), 4846/2496/3633 cGy (LR), and 4575/2496/3777 cGy (AP) for the restricted use case. Typical computation times for RADAR were ~2-4 minutes.

Conclusion: RADAR provides an automated solution for incorporating registration uncertainty into dose accumulation for spine reirradiation. RADAR can be applied to other disease sites with adjustments to search radii, which are structure specific.

ePosters

Keywords

Not Applicable / None Entered.

Taxonomy

TH- External Beam- Photons: Dose reconstruction over deforming anatomies

Contact Email