Log File-Based Dose Reconstruction to Moving Targets during Lung Stereotactic Body Radiation Therapy

Abstract

Purpose: To perform film-based verification of 4D dose reconstruction to moving targets during lung stereotactic body radiation therapy (SBRT). Introduction: Current patient-specific quality assurance measures to test deliverability of plans with dynamic intensity modulation involve delivering beams to static measurement device and comparing the planned dose to measurement. However, motion-induced dose errors are not detected with static measurement. Previous studies have investigated combining machine log data with respiratory tracking to determine moving-target dose. By combining machine log data with anatomic and density information at each breathing phase from 4D-CT, intrafraction anatomical deformation due to respiration may be accounted for. However, to our knowledge, a film-based verification of dose reconstruction using machine log data, intrafraction respiratory tracking, and 4D-CT has yet to be performed. Methods: Lung SBRT plans were anonymized for 12 patients treated at our institution. Treatment plans were copied onto known geometry (programmable respiratory phantom) and dose was computed. Each SBRT plan was delivered to the phantom twice; first using 3 sec/breath (SPB), again at 6 SPB. Respiratory traces were acquired during treatment. Logfiles were acquired after treatment and partitioned according to breathing amplitude. Next, in-house code was used to import logfile beams into the treatment planning system. Dose was computed on each 4D-CT image using the imported beams and deformably accumulated. The accumulated, planned, and measured doses for each plan and breathing rate were compared using gamma analysis. Results: Gamma passing rates (GPR) (3%, 2mm, 10% threshold) of 4D dose reconstruction vs. planned dose were >94% (mean 98.9% range 94.1%-100%) for all plans at each breathing rate. No significant difference was found between the 3 and 6 SPB GPRs (p=0.310). Overall, the 4D dose reconstructions were found to better agree with film measurement, within the tumor motion extent, than the treatment plan for both breathing rates (3 SPB: p=0.013, 6 SPB: p=0.017). Conclusions: Log file-based dose reconstruction was verified using film measurement for 12 lung SBRT plans delivered to a respiratory motion phantom. We showed that, given predictable phantom motion, 4D dose reconstruction resulted in significantly higher GPR compared with film than treatment plan to static geometry.