Center for Virtual Imaging Trials and Carl E. Ravin Advanced Imaging Laboratories, Department of Radiology, Duke University School of Medicine, Durham, North Carolina, USA.
Department of Physics, Duke University, Durham, North Carolina, USA.
Med Phys. 2022 Aug;49(8):5439-5450. doi: 10.1002/mp.15798. Epub 2022 Jun 21.
The gold-standard method for estimation of patient-specific organ doses in digital tomosynthesis (DT) requires protocol-specific Monte Carlo (MC) simulations of radiation transport in anatomically accurate computational phantoms. Although accurate, MC simulations are computationally expensive, leading to a turnaround time in the order of core hours for simulating a single exam. This limits their clinical utility. The purpose of this study is to overcome this limitation by utilizing patient- and protocol-specific MC simulations to develop a comprehensive database of air-kerma-normalized organ dose coefficients for a virtual population of adult and pediatric patient models over an expanded set of exam protocols in DT for retrospective and prospective estimation of radiation dose in clinical tomosynthesis.
A clinically representative virtual population of 14 patient models was used, with pediatric models (M and F) at ages 1, 5, 10, and 15 and adult patient models (M and F) with body mass index (BMIs) at 10th, 50th, and 90th percentiles of the US population. A graphics processing unit (GPU)-based MC simulation framework was used to simulate organ doses in the patient models, incorporating the scanner-specific configuration of a clinical DT system (VolumeRad, GE Healthcare, Waukesha, WI, USA) and an expanded set of exam protocols, including 21 distinct acquisition techniques for imaging a variety of anatomical regions (head and neck, thorax, spine, abdomen, and knee). Organ dose coefficients (h ) were estimated by normalizing organ dose estimates to air kerma at 70 cm (X ) from the source in the scout view. The corresponding coefficients for projection radiography were approximated using organ doses estimated for the scout view. The organ dose coefficients were further used to compute air-kerma-normalized patient-specific effective dose coefficients (K ) for all combinations of patients and protocols, and a comparative analysis examining the variation of radiation burden across sex, age, and exam protocols in DT, and with projection radiography was performed.
The database of organ dose coefficients (h ) containing 294 distinct combinations of patients and exam protocols was developed and made publicly available. The values of K were observed to produce estimates of effective dose in agreement with prior studies and consistent with magnitudes expected for pediatric and adult patients across the different exam protocols, with head and neck regions exhibiting relatively lower and thorax and C-spine (apsc, apcs) regions relatively higher magnitudes. The ratios (r = K /K ) quantifying the differences air-kerma-normalized patient-specific effective doses between DT and projection radiography were centered around 1.0 for all exam protocols, with the exception of protocols covering the knee region (pawk, patk).
This study developed a database of organ dose coefficients for a virtual population of 14 adult and pediatric XCAT patient models over a set of 21 exam protocols in DT. Using empirical measurements of air kerma in the clinic, these organ dose coefficients enable practical retrospective and prospective patient-specific radiation dosimetry. The computation of air-kerma-normalized patient-specific effective doses further enables the comparison of radiation burden to the patient populations between protocols and between imaging modalities (e.g., DT and projection radiography), as presented in this study.
在数字断层合成（DT）中，估计特定于患者的器官剂量的金标准方法需要针对特定协议的辐射传输进行解剖精确的计算体模中的蒙特卡罗（MC）模拟。尽管准确，但 MC 模拟计算成本很高，导致模拟单个检查的周转时间约为核心小时。这限制了它们的临床应用。本研究的目的是通过利用患者和协议特定的 MC 模拟来克服这一限制，为扩展的 DT 检查协议中的成年和儿科患者模型虚拟人群开发一个全面的空气比释动能归一化器官剂量系数数据库，用于回顾性和前瞻性估计临床断层合成中的放射剂量。
使用了一个具有临床代表性的 14 个患者模型的虚拟人群，其中包括儿科模型（M 和 F）的 1、5、10 和 15 岁以及成年患者模型（M 和 F）的 BMI 在第 10、50 和 90 百分位数的美国人口。使用基于图形处理单元（GPU）的 MC 模拟框架模拟患者模型中的器官剂量，结合临床 DT 系统（VolumeRad，GE Healthcare，Waukesha，WI，USA）的特定扫描仪配置和扩展的检查协议集，包括用于成像各种解剖区域（头部和颈部、胸部、脊柱、腹部和膝盖）的 21 种不同的采集技术。器官剂量系数（h）通过将器官剂量估计值归一化为在源在 scout 视图中距 70cm（X）处的空气比释动能来估计。使用 scout 视图中估计的器官剂量来近似投影放射摄影的相应系数。进一步使用器官剂量系数计算所有患者和协议组合的空气比释动能归一化患者特定有效剂量系数（K），并进行了比较分析，研究了在 DT 中以及与投影放射摄影相比，不同性别、年龄和检查协议的辐射负担变化。
开发了包含 294 种不同患者和检查协议组合的器官剂量系数数据库，并公开发布。观察到 K 的值产生与先前研究一致的有效剂量估计值，并且与不同检查协议中儿科和成年患者的预期幅度一致，头部和颈部区域的幅度相对较低，胸部和 C 脊柱（apsc、apcs）区域的幅度相对较高。定量比较 DT 和投影放射摄影之间空气比释动能归一化患者特定有效剂量差异的比值（r=K/K）对于所有检查协议均接近 1.0，除了覆盖膝盖区域的协议（pawk、patk）。
本研究为 14 个成年和儿科 XCAT 患者模型的虚拟人群开发了一个器官剂量系数数据库，涵盖了 21 种 DT 检查协议。使用临床空气比释动能的实际测量值，这些器官剂量系数能够实现实用的回顾性和前瞻性患者特定放射剂量测定。空气比释动能归一化患者特定有效剂量的计算进一步能够比较不同协议和成像方式（例如，DT 和投影放射摄影）之间的患者人群的辐射负担，如本研究所示。
