Encrypted login | home

Program Information

Validation Phantom of Dosimetric and Deformable Accuracy for Purposes of Commissioning An MR-Guided Online Adaptive Radiotherapy (ART) Program

K Mittauer

K Mittauer*, P Hill , J Bayouth , University of Wisconsin, Madison, WI


TU-D-FS4-1 (Tuesday, August 1, 2017) 11:00 AM - 12:15 PM Room: Four Seasons 4

Purpose: To investigate the dose calculation and deformation accuracy of an online adaptive radiotherapy (ART) program for a clinical MR-guided radiotherapy (MRgRT) system.

Methods: An in-house deformable phantom was comprised of the following anatomically-relevant materials: pork-ribs (ribcage), 27 lbs of ground-chuck/beef (body), cow-femur bone (spine), grapes (nodes), banana (pancreas), tangerine (kidney), orange (kidney), lemon (duodenum), and air-filled balloons (bowel). TLDs were affixed within each regions-of-interest (ROI). A CT scan and 3D-volumetric MR scan were acquired. CT was deformed to MR for electron density. Planning was performed on MR for 5 Gy/5 fractions. For each fraction an online ART was executed. Prior to each fraction, a deformation was applied to the phantom using Styrofoam blocks. Ground truth was known for ROI’s volume, TLD position, and magnitude of ROI deformation (i.e., interfraction centroid difference from fractional to simulation MR scan). Dose accumulation was performed in MIM software. The TLD-to-calculated dose results for this deformable phantom were compared to a similar study on a rigid homogenous phantom (IROC head-and-neck).

Results: ART CT-MR deformation showed good agreement (mean dice similarity coefficient (DCS) 0.92±0.02, volume ratio 1.11±0.11, Hausdorff distance (HD) 1.49±0.61 cm) with larger variation observed for MR-MR deformation (mean DCS 0.78±0.15, volume ratio 0.91±0.16, HD 2.84±1.34 cm). Mean differences of calculated-to-measured TLD with a 5-fraction ART course to a heterogeneous deformable phantom (7.25%±4.63%) were similar for a homogeneous rigid phantom with non-ART (3.67%±0.36%). Agreement for targets (low gradient regions) with heterogeneous deformable phantom remained within clinical trial limits (calculated-to-measured TLD dose within ±7%). Greater differences were observed in MIM-reported accumulated dose (22.48%±22.95%).

Conclusion: The dosimetric and deformation accuracy of an online ART program has been validated. Greater differences were observed in MIM-reported dose accumulation statistics due to greater uncertainty with MR-MR deformation. Caution should be used when performing cumulative dose calculations from multiple fractions.

Contact Email: