Accuracy and robustness of 4D logfile-based dose reconstruction


Accuracy and robustness of 4D logfile-based dose reconstruction

Spautz, S.; Peters, N.; Meijers, A.; Jakobi, A.; Knopf, A.; Troost, E. G. C.; Richter, C.; Stützer, K.

Introduction: We plan to use 4D logfile-based dose reconstruction for daily monitoring and potential intervention in PBS treatments of non-small cell lung cancer patients, restricted to limited motion (≤5mm). Here, we assessed the validity of reconstructed doses and their sensitivity to selected disturbed input parameters by dedicated phantom experiments.
Material/Methods: Quasi-monoenergetic proton fields were delivered to a dynamic thorax phantom (G.C. Technology, Germany) equipped with a 3cm soft-tissue target intersected by a radiochromic film. The surrogate signal (AZ733-V, ANZAI Medical Co.,Ltd, Japan) of the motion patterns (cos/cos4, period: 5s, peak-to-peak amplitude: 5mm and 30mm) was recorded in synchronization with the machine logfiles. 4D reconstructions with 1mm/3mm dose grid resolution were performed using 4DCTs of 12 amplitude-sorted phases and either ground truth or automatically generated deformation vector fields. Characteristic 1D profiles of the reconstructed and measured doses were compared by gamma index analyses (2mm, 2%). Maximum dose deviations due to simulated offsets between motion and machine logfiles (±1/±5/±25/±250ms) were assessed for quasi-monoenergetic and 4D optimized plans.
Results: Characteristic dose patterns were well reproduced (Fig.1). Gamma pass rates were >98% under static conditions. For 5mm motion, the pass rate of 94.2% for an ideal reconstruction with 1mm³ dose voxels dropped to 93.0% with clinically used voxel sizes (3×3×3mm³), 83.6% when using automated DIR and 78.2% for the combination of both, respectively. For the 30mm motion, the CT artifacts and residual motion were predominant and lead for 3mm dose grids to gamma pass rates of approximately 84%, irrespective of chosen DIR. Fig.2 depicts the effect of simulated logfile asynchrony.
Conclusions: The implemented method is robust against disturbed input parameters for the small, clinically aimed motion amplitudes. Reconstruction accuracy decreases with deformation-related inaccuracies and increasing 4DCT artifacts for large motion. Consistent breathing and regular control CTs are compulsory for meaningful 4D logfile-based dose reconstructions.

  • Poster
    4D Treatment Workshop for Particle Therapy 2019, 22.-23.11.2019, Kraków, Polska

Permalink: https://www.hzdr.de/publications/Publ-29781
Publ.-Id: 29781