Evaluating the Impact of Liver Vasculature Model Complexity for Estimating Dose to Circulating Blood During Radiation Therapy.

Purpose: To assess the impact of liver model complexity on the estimated radiation dose to circulating blood during radiation therapy.

Methods And Materials: Six patients with hepatocellular carcinoma (HCC) were selected covering a range of clinical treatment volume (CTV) sizes and locations. Photon and proton treatment plans were generated for each patient. Planning computed tomography, CTV contours, and dose distributions were deformably registered to the reference livers provided by the International Commission on Radiological Protection report. Three vasculature models were considered: (1) main vascular tree (MVT), (2) coarse vascular tree (CVT) of 1045 vessels, and (3) detailed vascular tree (DVT) of 2041 vessels. Blood dose-volume histograms (bDVH, bDVH, and bDVH) and the mean circulating blood dose (μ, μ, and μ) were estimated using Monte Carlo simulations for all 3 models. The effect of varying blood velocity (v) in HCC tumors on dose estimation was also evaluated through increasing the tumor v by 1.5, 2, and 4.2 times.

Results: For the 3 lesions located in the left lobe, the estimated μ was lower than μ by an average ± standard deviation of (6 ± 4)% and (17 ± 7)% for photon and proton treatments, respectively. Smaller differences were found for lesions in the right lobe, where μ was on average (2 ± 1)% lower than μ for photon and (3 ± 1)% lower for proton treatments. More pronounced difference between μ and μ was seen in lesions with smaller CTV sizes. We also found that considering the elevated tumor v led to a reduction of estimated dose to circulating blood, with a maximum reduction in the estimated μ of 39% and 8% for CTV of 603 and 249 mL, respectively.

Conclusion: Our study revealed that the impact of liver vasculature model complexity on the estimated dose to blood depended on lesion-specific characteristics. For lesions with larger CTV size on the right liver lobe treated with photons, modeling only major vessels could generate bDVHs that are dosimetrically comparable with bDVHs of more complex vascular models. Increased tumor v resulted in a reduction of the estimated blood dose.