Program Information
The Effects of Inter-Cycle Respiratory Motion Variation On Dose Accumulation in Single Fraction MR-Guided SBRT Treatment of Renal Cell Carcinoma
B Stemkens1*, M Glitzner1 , C Kontaxis1 , B Denis de Senneville2,3 , F Prins1 , SPM Crijns1 , L Kerkmeijer1 , J Lagendijk1 , CAT van den Berg1 , RHN Tijssen1 , (1) Department of Radiotherapy, University Medical Center Utrecht, Utrecht, the Netherlands (2) Imaging Division, University Medical Center Utrecht, Utrecht, the Netherlands (3) IMB, UMR 5251 CNRS/University of Bordeaux, France
Presentations
SU-D-207A-7 (Sunday, July 31, 2016) 2:05 PM - 3:00 PM Room: 207A
Purpose:
To assess the dose deposition in simulated single-fraction MR-Linac treatments of renal cell carcinoma, when inter-cycle respiratory motion variation is taken into account using online MRI.
Methods:
Three motion characterization methods, with increasing complexity, were compared to evaluate the effect of inter-cycle motion variation and drifts on the accumulated dose for an SBRT kidney MR-Linac treatment: 1) STATIC, in which static anatomy was assumed, 2) AVG-RESP, in which 4D-MRI phase-volumes were time-weighted, based on the respiratory phase and 3) PCA, in which 3D volumes were generated using a PCA-model, enabling the detection of inter-cycle variations and drifts.
An experimental ITV-based kidney treatment was simulated in a 1.5T magnetic field on three volunteer datasets. For each volunteer a retrospectively sorted 4D-MRI (ten respiratory phases) and fast 2D cine-MR images (temporal resolution = 476ms) were acquired to simulate MR-imaging during radiation.
For each method, the high spatio-temporal resolution 3D volumes were non-rigidly registered to obtain deformation vector fields (DVFs). Using the DVFs, pseudo-CTs (generated from the 4D-MRI) were deformed and the dose was accumulated for the entire treatment. The accuracies of all methods were independently determined using an additional, orthogonal 2D-MRI slice.
Results:
Motion was most accurately estimated using the PCA method, which correctly estimated drifts and inter-cycle variations (RMSE=3.2, 2.2, 1.1mm on average for STATIC, AVG-RESP and PCA, compared to the 2D-MRI slice). Dose-volume parameters on the ITV showed moderate changes (D99=35.2, 32.5, 33.8Gy for STATIC, AVG-RESP and PCA). AVG-RESP showed distinct hot/cold spots outside the ITV margin, which were more distributed for the PCA scenario, since inter-cycle variations were not modeled by the AVG-RESP method.
Conclusion:
Dose differences were observed when inter-cycle variations were taken into account. The increased inter-cycle randomness in motion as captured by the PCA model mitigates the local (erroneous) hotspots estimated by the AVG-RESP method.
Contact Email: