3205 - Dose Reconstruction for Helical Delivery Verification Using Optical Sensor Data

Purpose/Objective(s): Previous studies have shown that MV detector-derived sinograms can be used to reconstruct dose for helical delivery verification. However, potential errors arising from changes in output, MVCT density curve, and detector response can challenge accuracy. We explore here the feasibility of using kV images and MLC leaf optical sensor data, available on the new 3-D CRT and IMRT system, for helical delivery verification.

Materials/Methods: Plan and delivery data from a sample of patients treated on our 3-D CRT and IMRT system were retrospectively analyzed. For each treatment delivery, MLC leaf optical sensor data was utilized to derive MLC leaf-open-time (LOT) by measuring the duration between the mid-points of the opening & closing transition states. A delivery plan was reconstructed using synchronized gantry angle, couch position, jaw position, & LOT, together with prescription information from the treatment planning system (TPS) plan. Dose was then calculated using SureCalc (Commercially available software) and compared with reference dose using 3D gamma analysis. Three tests with different image sets for plan generation were performed. First, using raw data from 1 delivered fraction of each patient, 1 reconstructed plan was generated on top of the planning images for each of 22 patients (5 prostate, 5 female pelvis, 5 pancreas, 2 liver, 2 abdomen, 2 breast, 1 scalp). TPS dose was used as reference. Second, for 2 of the prostate patients, 1 delivery plan was reconstructed on top of each daily merged image (PreciseArt, technology company). Daily dose from PreciseArt, was used as reference. Daily variation of dose-volume parameters for bladder and rectum were evaluated. Third, reconstructed QA delivery plans were generated on a patient-specific quality assurance for 8 patients who had QA delivered on the same phantom. The SureCalc dose of each plan was compared against the TPS QA dose, as well as the measurement. Level 1 gamma criteria from TG218 were used except for comparison with QA measurement data, where 5% threshold, 3mm and 3% criteria was used.

Results: Gamma passing rates for the 3 tests are summarized in Table 1. An MLC issue was found to negatively bias the SureCalc-Meas gamma passing rate. For 1 of the prostate patients in Test 2, the average V(4000), V(6500) and V(7000 cGy) for the bladder were 18.8±4.9, 6.5±2.5 & 4.0±1.8% respectively. The average D(0.03cc) and mean dose for the bladder and D(0.03cc) for the rectum were 7836.2±5.0, 1773.9±362.5 and 7775.8±62.9 cGy, respectively.

Conclusion: Sinogram, plan and dose reconstruction based on MLC leaf optical sensor data can be reliably used for helical secondary and delivery dose verification. Daily variation of gamma passing rate and dose-volume parameters can be used to monitor patient treatment.