Room: AAPM ePoster Library
Purpose: beam profiles generated for small fields have spikes or noise due to several practical challenges involved in small field relative dosimetry measurements. Manufacturers of the dosimeters provide software tools equipped with various algorithms to process the measured beam profiles for smoothening. The present study is intended to quantify the variation in small field profile data processed with smoothening algorithms of SunNuclear’s dosimetry software in comparison with actual unprocessed measured beam profile data.
Methods: cross-line beam profiles for field sizes 2×2 cm², 3×3 cm² and 4×4 cm² are measured using SunNuclear’s 3D RFA at 10 cm depth in water with PTW’s pin point detector (Volume: 0.015 cc) for 6 MV FFF beam generated by TrueBeam LINAC. These profiles were analyzed with SunNuclear’s dosimetry software and smoothened using available algorithms i.e. adaptive data density, digital smoothening polynomial, iterative de-ionizing, SG fitting, spike removal, geometric mean and rolling average. The percentage dose variation of smoothened profiles with respect to un-smoothened actual beam profile are calculated in central and penumbra region of beam profile.
Results: on analysis, maximum and average percentage variation for adaptive data density, digital smoothening polynomial, iterative deionizing, SG fitting, spike removal, geometric mean and rolling average algorithms in comparison with measured unprocessed beam profile data are found to be 0.9% and 0.2%, 1.2% and 0.4%, 1.2% and 0.2%, 1.3% and 0.2%, 1.7% and 0.2%, 1.6% and 1.0%, 3.1% and 0.7% respectively.
Conclusion: results, it is evident that adaptive data density algorithm is introducing minimum variation with respect to actual measured beam profile data but this algorithm does not smoothen profile data satisfactorily at few points. Iterative deionizing and SG fitting algorithms introduce deviation slightly higher than adaptive data density algorithm but remove spikes in profiles effectively. It is observed that rolling average algorithm introduces maximum deviation from actual data.