Purpose: To assess the impact of liver model complexity levels on the dose calculation accuracy to circulating blood during photon (IMRT, VMAT) and proton beam therapy (PBS, SOBP).
Methods: Six HCC patients were selected covering a range of clinical target volumes (CTV) sizes and locations. Patient CT images, CTV contours and dose distributions were deformably registered to the ICRP reference liver phantoms. Three vasculature models of increasing complexities including major vessels only, coarse (1045 vessels) and detailed (2041 vessels) vascular trees were developed for the reference liver phantoms. Blood DVHs and the mean dose to circulating blood μ(b,dose) over 15 fractions were computed using Monte Carlo simulations. The effect of varying blood velocity ν(b) in HCC tumors on dose estimation accuracy is also evaluated by increasing the tumor v(b) by 1.5, 2 and 4.2 times.
Results: Dose under-estimation was observed primarily for lesions in the left/central liver when the liver model included only major blood vessels. μ(b,dose) was under-estimated by (6±4)% and (17±7)% for photon and proton treatments respectively. Less difference was found for lesions in the right liver, where μ(b,dose) was under-estimated by (2±1)% for photon treatments and (3±1)% for proton treatments. More pronounced μ(b,dose) under-estimation was also seen in lesions with smaller CTVs. Furthermore, our simulation demonstrated that elevated tumor ν(b) as previously reported in HCC led to a reduction of dose to circulating blood. This effect grew with increasing tumor ν(b) and CTV size, with a maximum reduction in μ(b,dose) of 39% and 4% for CTV of 603ml and 249ml, respectively.
Conclusion: Our study suggested that the impact of model complexity on dose calculation accuracy depended on lesion-specific characteristics. Accuracy was predominately affected by increased tumor ν(b) for lesions with greater CTV, whereas vasculature complexity played an important role for small lesions situated on the left/central liver.
Funding Support, Disclosures, and Conflict of Interest: This work was funded by NIH R01 CA248901, R21 CA248118, R21 CA241918 and P01 CA261669