核医学治疗中患者的体素级剂量计算

Voxelized Dose Assessment for Radionuclide Therapy in Nuclear Medicine

  • 摘要: 核医学治疗中为了制定合理的治疗方案需要对患者进行个体化剂量评估。本研究通过计算患者的体素级剂量分布,对分化型甲状腺癌患者的131I SPECT图像及CT图像进行了精准定量分析,利用体素S因子方法、直接蒙特卡罗方法和GPU蒙特卡罗方法计算了核医学患者体素级剂量分布。体素S因子方法相比于蒙特卡罗方法,在上呼吸道内部、肺部、骨骼等与水密度差异大的位置表现出剂量结果的差异,在肺和骨骼的最大差异达到40%,且在组织的交界处差异明显,在距离源分布较远位置的剂量相对于蒙特卡罗结果有一定低估。GPU蒙特卡罗方法与直接蒙特卡罗方法计算结果一致,并实现了400倍的加速比。综上所述,体素级剂量计算能获得亚器官区域和肿瘤区域的剂量不均匀分布,直接蒙特卡罗方法相比于体素S因子方法能够实现更精确的剂量计算,GPU程序能有效加速蒙特卡罗计算。

     

    Abstract: Nuclear medicine treatment requires patient-specific dose assessment as treatment plan reference. Voxelized dose results can give three-dimensional dose distribution. The conventional MIRD method assumes that the source is uniformly distributed within the organ, and the calculated dose results in an organ-averaged dose. In this study, based on molecular imaging data, the non-uniform distribution of the source term within the organ was obtained, and the voxel-level dose distribution was calculated for nuclear medicine patients using the following method. With accurate quantitative analyze of 131I SPECT images and CT images of patients, voxel S-value method and direct Monte Carlo (MC) method were used to calculate voxel-level dose distribution in nuclear medicine patients. The voxel S-factor is defined as the average absorbed dose of each radioactive decay in the source voxel to the target voxel, both of which are contained in an infinite homogeneous tissue. The voxel S-value is calculated to evaluate the dose of multiple surrounding source voxels to the target voxel. The three-dimensional dose distribution is calculated based on the source distribution obtained from the SPECT images. The MC simulation was processed based on the source distribution obtained from SPECT images and CT images. The HU value of each voxel in the patient’s body was obtained from the CT images, and the density value of each voxel was further obtained. The material of each voxel was divided according to the density to construct the voxel phantom. The voxel-level activity distribution was obtained based on the SPECT images to construct the source term distribution model. The MC simulation based on CT images was used to count the energy deposition of each voxel to obtain the accurate voxel-level dose distribution in the patient’s body. A 400x speedup was achieved using GPU MC simulation. The results of the S-value method and the MC method show a maximum difference of over 40% in the lung area. Voxel-level dose simulation can obtain the dose distribution in sub-organ and tumor, which provides a valuable reference for precision treatment of clinical nuclear medicine. The S-factor method can quickly give the dose distribution of patients, but this method ignores the tissue inhomogeneity, and the dose calculation results are not accurate in tissues and organs with large density differences, such as lung and bone, etc. The MC voxelized simulation can accurately give the dose distribution in patients, which is important for the accurate assessment of clinical dose in nuclear medicine.

     

/

返回文章
返回