An improved beam splitting method for intensity modulated proton therapy.

2020 
Pencil beam algorithm (PBA) has become the predominant dose calculation method in proton therapy due to high efficiency. However, density heterogeneity decreased the accuracy of PBA. To improve PBA's accuracy, beam splitting method was used which divides the original scanning beam into multiple thinner beamlets. Beam splitting should ensure that the beamlets' summed fluence is as close to the original beam fluence as possible, while keeping the spatial variance of beamlets small and minimizing the number of beamlets. In this work, the generalized differential evolution (GDE) algorithm was utilized for the optimal scheme. Under reasonable constraints, several schemes were optimized by GDE algorithm. To achieve a trade-off between the accuracy and calculation speed, three hexagons based schemes, which split the original beam into 7, 13 and 19 beamlets respectively, were proposed and compared with the scheme of Raystation 4.5. The schemes with 13 beamlets and 19 beamlets were more accurate than the scheme of Raystation which has 19 beamlets, with maximum absolute difference between the summed beamlets fluence and the original beam fluence was 2.12% and 0.93% respectively. Furthermore, beam splitting schemes are implemented into a proton dose calculation algorithm which based on the KylinRay-IMPT TPS. These schemes based on the algorithm were compared with Monte Carlo program TOPAS 3.2 in slab geometry with lateral heterogeneity. The dose, calculated by the algorithm with the scheme of 13 beamlets, showed a good agreement with the dose from TOPAS. Also, an abdominal geometry is used for further verification. The Gamma analysis passing rates are greater than 99.7% with the 2% /2 mm criterion. Thus, the accuracy and effectiveness of the improved beam splitting method were preliminarily verified.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    0
    References
    1
    Citations
    NaN
    KQI
    []