Accuracy and robustness of 4D logfile-based dose reconstruction and start of clinical application

Introduction: We established a 4D logfile-based dose reconstruction for monitoring and potential intervention during intensity-modulated proton therapy (IMPT) of moving tumors. Before clinical application, we assessed the validity of reconstructed doses and the sensitivity against changes of selected input parameters by phantom experiments.
Material/Methods: A dynamic thorax phantom (CIRS, USA) with a soft-tissue target and radiochromic film insert was imaged by 4DCT and irradiated with either quasi-monoenergetic fields or 4D optimized proton plans. The surrogate signal (ANZAI, Japan) of the regular motion was recorded in synchronization with the machine logfiles. Reconstructions were performed with different dose grid resolutions (1mm/3mm), deformable image registrations (DIR; manually defined or automatically generated vector-fields) and artificial asynchronies between machine and motion logfiles.
Results: Characteristic dose patterns on radiochromic films were well reconstructed (Fig.1A). Gamma pass rates (2mm, 2%) for extracted characteristic profiles of the reconstructed and measured doses were >98% under static conditions, ranged between 99% and 86% for 5mm motion depending on applied reconstruction parameters, especially the DIR, and were about 80% for 30mm motion due to the predominant residual motion in the 4DCT (Fig.1B). Fig.1C demonstrates the robustness against potential minor asynchronies (≈5ms) between machine and motion logfiles. A workflow test during a pancreatic cancer IMPT treatment (Fig.2) revealed a data processing time of approximately 20min/fraction.
Conclusions: Due to satisfying accuracy and robustness for clinically aimed motion amplitudes (≤5mm), IMPT treatment of non-small cell lung cancer accompanied by daily 4D logfile-based dose monitoring will start in our institute within the first months of 2020.