Encrypted login | home

Program Information

Simulation of Motion Amplitude Variations to Verify the Robustness of Treatment Plans

no image available
K Souris

K Souris1*, A Barragan Montero1 , G Janssens2 , D Di Perri1 , E Sterpin1,3 , J Lee1 , (1) Universite Catholique de Louvain, Bruxelles, Belgium, (2) IBA s.a., Louvain-la-neuve, Belgium, (3) KU Leuven, Leuven, Belgium


SU-H3-GePD-T-2 (Sunday, July 30, 2017) 4:00 PM - 4:30 PM Room: Therapy ePoster Lounge

Purpose: Large intra- and inter-fraction variations of the breathing motion amplitude have been reported by several publications, with possibly dramatic consequences on treatment quality for moving tumors. This study aims at modeling the amplitude variation as a new uncertainty scenario in the plan robustness verification/optimization.

Methods: The OpenREGGUI implementation of diffeomorphic deformable registration (Janssens 2010) was used to compute the velocity fields from a reference phase to all 4DCT phases. Combinations of these fields were then used to deform all phases to the motion Mid-Position. The Mid-Position CT image (MidPCT) is generated by taking the median in each voxel of all deformed phases. This operation reduces noise and artifacts of the MidPCT, compared to individual 4DCT phases. The MidPCT can be deformed back to generate a 4DCT of better quality. The motion amplitude of this reconstructed 4DCT can be scaled by multiplying the velocity fields with a scalar factor. The fast Monte Carlo code MCsquare was used to perform a 4D robustness verification of proton therapy PBS plans. Multiple scenarios of motion amplitudes are simulated. The resulting doses were reported in a DVH-band.

Results: A 20% increase of the initial breathing amplitude was first simulated to verify the method. Because rotations are preserved by the operations on velocity fields, the effective variation of tumor displacement, measured between both 4DCT, was 19%. A proton PBS plan was optimized for the initial 4DCT series. Its robustness was then tested with 100 scenarios of motion patterns randomly sampled from a Gaussian distribution centered on the initial motion amplitude and with a standard deviation of 40%. The test revealed a deviation of the CTV D95 ranging from 94% to 97% of the prescribed dose.

Conclusion: Our model enables a more realistic robustness verification/optimization of mobile tumor treatments and is directly usable in commercial TPS.

Funding Support, Disclosures, and Conflict of Interest: Kevin Souris is supported by a research grant from IBA s.a.

Contact Email: