Acuros XB、各向异性解析算法与蒙特卡罗算法在非均匀组织中剂量计算准确性对比研究

Accuracy comparison of Acuros XB algorithm, anisotropic analytical algorithm and Monte Carlo algorithm in calculating doses of inhomogeneous tissues

目的：对比Acuros XB算法（AXB）、各向异性解析算法（AAA）和蒙特卡罗（MC）算法在非均匀组织中剂量计算准确性。方法：在Eclipse计划系统上分别设置两种类型的非均匀模体（水-肺-水模体、水-骨-水模体）,并设定3个不同大小的0°照射野,源皮距=100 cm。采用AXB、AAA及MC算法进行剂量计算,提取射野中心轴百分深度剂量,以MC计算结果为基准,计算AXB和AAA两种算法与MC算法的相对偏差,提取非均匀组织及高梯度区（即4.5～15.5 cm）的数据做对比分析。结果：AXB算法3个射野相对偏差绝对值分别为4.186±1.451、0.834±0.300、0.726±0.165（水-肺-水模体）和1.694±0.374、1.325±0.328、0.343±0.244（水-骨-水模体）。AAA算法在两模体的对应值分别为6.679±4.694、4.151±1.789、4.353±2.546（肺）和3.270±0.826、5.971±1.587、2.406±0.574（骨）。采用配对样本t检验,P值均小于0.05。结论：在非均匀组织及其边界,AXB算法计算精度比AAA算法更为准确,基本接近MC算法。

Objective To compare the dosimetric accuracy of Acuros XB algorithm（AXB）, anisotropic analytical algorithm（AAA） and Monte Carlo（MC） algorithm in calculating the dose of inhomogeneous tissues. Methods Based on Eclipse planning system, two virtual inhomogeneous phantoms, water- lung- water phantom and water- bone- water phantom, were created. Three fields of various sizes were set at gantry angle of 0°, with source skin distance of 100 cm. AXB, AAA and MC algorithm were applied to calculate the percentage depth dose on the beam central axis. The relative differences benchmarked to calculated results of MC algorithm were calculated. And the data of inhomogeneous tissues and high dose gradient regions with depths from 4.5 cm to 15.5 cm were extracted for comparison. Results For AXB, the absolute values of relative differences in three fields were respectively 4.186 ± 1.451, 0.834 ± 0.300, 0.726 ± 0.165 in water- lung- water phantom and 1.694 ± 0.374, 1.325 ± 0.328, 0.343 ± 0.244 in water- bone- water phantom. The corresponding values for AAA were 6.679±4.694, 4.151±1.789, 4.353±2.546 in water-lung-water phantom and 3.270±0.826, 5.971±1.587, 2.406±0.574 in water- bone- water phantom. Paired sample t test showed all differences were statistically significant（P〈0.05）. Conclusion For the dose calculation of inhomogeneous tissues and its boundaries, the accuracy of AXB was close to that of MC algorithm and better than that of AAA.