RESUMO
BACKGROUND: With rapid evolutions of fast and sophisticated calculation techniques and delivery technologies, clinics are almost facing a daily patient-specific (PS) plan adaptation, which would make a conventional experimental quality assurance (QA) workflow unlikely to be routinely feasible. Therefore, in silico approaches are foreseen by means of second-check independent dose calculation systems possibly handling machine log-files. PURPOSE: To validate the in-house developed GPU-dose engine, FRoG, for light ion beam therapy (protons and carbon ions) as a second-check independent calculation system and to integrate machine log-file analysis into the patient-specific quality assurance (PSQA) program. METHODS: Spot sizes, depth-dose distributions, and absolute dose calibrations were configured into FRoG and a set of nine regular-shaped targets in combination with more than 170 clinical treatment fields were tested against pinpoint ionization chamber measurements. Both the treatment planning system DICOM RTplans and machine treatment log-files were used as input for the dose kernel in water, and a 3D local γ (1 mm/2%) index was used as the main evaluation metric. RESULTS: Calculated configuration data matched experimental measurements with submillimetric agreement. For regular-shaped targets, the unsigned average relative difference between calculated and measured dose values was less than 2% for both protons and carbon ions. The mean γ passing rate (PR) was around 98% for both particle species. For clinical treatment beams, DICOM-based recalculations showed a γ-PR more than 99% for both particle species. The same level of agreement was preserved for protons when moving to log-file-based recalculations. A score of around 95% was registered for carbon ion beams, once excluding low-quality machine log-files. Unsigned average relative difference against acquired data was less than 2% also for real clinical beams. CONCLUSIONS: FRoG was proven as an accurate and reliable tool for PSQA in scanning light ion beam therapy. The proposed method allows for an extremely efficient workflow, without compromising the quality of the plan verification procedure.