Patient-specific CT calibration based on ion radiography for different detector configurations in 1H, 4He and 12C ion pencil beam scanning
Patient-specific CT calibration based on ion radiography for different detector configurations in 1H, 4He and 12C ion pencil beam scanning
复制标题
DOI：
10.1088/1361-6560/aba319
复制
发表时间：
2020-12-21
影响因子：
3.5
通讯作者：
Parodi, Katia
中科院分区：
文献类型：
--
作者：
Gianoli, Chiara;Goppel, Maximilian;Parodi, Katia
治疗计划X射线计算机断层摄影（CT）图像到相对于水的离子阻止能力的经验转换导致离子束治疗中的剂量计算不准确。通过将离子射线照相术（iRad）与经验转换的CT图像沿着估计的离子轨迹的前向投影相结合来实现CT图像的患者特异性校准。这项工作研究了患者特定的CT校准列表模式和积分模式探测器配置，参考地面实况离子CT（iCT）图像。分析模拟的理想化的碳离子和质子的轨迹在一个数字拟人化的幻影和现实的Monte Carlo模拟质子，氦和碳离子笔束扫描的头部和颈部患者的临床CT图像被认为是。将受控不准确度和噪声水平分别应用于校准曲线和iRad。研究了iRads切片和角度选择以及优化算法选择的影响。在直碳离子轨迹的分析模拟中获得了准确和鲁棒的CT校准。由于散射的非直线质子轨迹的分析模拟建议的限制，集成模式，但不是列表模式。为了充分利用积分模式，提出了一个专用的目标函数，证明了在分析模拟中足够高的质子统计所需的精度。在临床数据中，iRad与地面实况iCT图像的前向投影之间的不一致性与应用的不准确性处于相同的数量级（高达5%）。CT校正的精度在积分模式下为2%~ 5%，列表模式下为1%~ 3%。成功的患者特异性CT校准的可行性取决于检测器技术，并且主要受到上述不一致性的限制，这些不一致性由于较大的散射和射束光斑尺寸而使质子相对于氦和碳离子略微不利。
The empirical conversion of the treatment planning x-ray computed tomography (CT) image to ion stopping power relative to water causes dose calculation inaccuracies in ion beam therapy. A patient-specific calibration of the CT image is enabled by the combination of an ion radiography (iRad) with the forward-projection of the empirically converted CT image along the estimated ion trajectories. This work investigated the patient-specific CT calibration for list-mode and integration-mode detector configurations, with reference to a ground truth ion CT (iCT) image. Analytical simulations of idealized carbon ion and proton trajectories in a numerical anthropomorphic phantom and realistic Monte Carlo simulations of proton, helium and carbon ion pencil beam scanning in a clinical CT image of a head-and-neck patient were considered. Controlled inaccuracy and noise levels were applied to the calibration curve and to the iRad, respectively. The impact of the selection of slices and angles of the iRads, as well as the choice of the optimization algorithm, were investigated. Accurate and robust CT calibration was obtained in analytical simulations of straight carbon ion trajectories. Analytical simulations of non-straight proton trajectories due to scattering suggested limitations for integration-mode but not for list-mode. To make the most of integration-mode, a dedicated objective function was proposed, demonstrating the desired accuracy for sufficiently high proton statistics in analytical simulations. In clinical data the inconsistencies between the iRad and the forward-projection of the ground truth iCT image were in the same order of magnitude as the applied inaccuracies (up to 5%). The accuracy of the CT calibration were within 2%-5% for integration-mode and 1%-3% for list-mode. The feasibility of successful patient-specific CT calibration depends on detector technologies and is primarily limited by these above mentioned inconsistencies that slightly penalize protons over helium and carbon ions due to larger scattering and beam spot size.