基于四维CT的肺通气功能成像用于放射性肺炎的预测模型研究

The incidence of radiation pneumonitis (RP) has seriously restricted the increase of the target dose, and hence the clinic efficacy of radiotherapy. The risk and severity of RP complication for patient after radiotherapy should be evaluated carefully before treatment delivery. The common method currently used in the clinic is still an experience threshold which based on the research in the 1990’s. This method is based on the assumption that the response of the whole lung against the radiation was homogeneous and the relative volume of lung received low-dose and the mean lung dose are used to predict the risk of RP complication. However, there are two problems with this approach: first, the lung volume was determined based on an instantaneous CT image while the lung volume would be different in different respiratory phase. It would induce uncertainty into the acquisition of dose-volume parameters which associated with lung volume. Second, it has been approved by many experiments that the ventilation function of lung was different in different parts of the lung. This study will introduce some advanced technologies, such as deformable registration, pulmonary ventilation function image derived from 4DCT, cluster analysis and principal component analysis technology, into the prediction of RP. The 4DCT datasets for 200 patients received radiotherapy are used to generate the pulmonary ventilation function images after deformable registration. After the collection of dose-volume and dose-function parameters, retrospective analysis would carry out to find the relationship between these dose-volume and/or dose-function parameters with the follow-up data for corresponding patients. A quantitative prediction model for RP based on pulmonary ventilation function images derived from 4DCT would respectively be established. The results will provide clinic a quantitative guidance for the prediction of the risk and severity of RP.

放射性肺炎（Radiation Pneumonitis，RP）严重制约了胸部肿瘤靶区照射剂量的提高，限制了疗效。放疗实施前需认真评估患者放疗后RP的发生概率和严重程度。目前临床常用的方法仍是基于上世纪90年代总结的经验阈值，假设全肺对照射的反应是均质同性的，使用全肺接受低剂量照射的百分体积和平均照射剂量预测RP发生的概率。该方法存在两个问题：1.该方法中肺体积是依据瞬时图像确定的，不同呼吸时相扫描得到的CT图像肺体积会有较大差别，从而影响所采集的剂量体积指标的准确性；2.实验证明肺不同部位的通气功能是不同的。本项目引入变形配准技术、基于4DCT的肺通气功能成像技术、聚类分析和主成分分析技术，基于200例患者的4DCT图像生成肺通气功能成像，采集剂量体积和剂量功能指标，建立基于4DCT的肺通气功能成像用于RP的量化预测模型。研究成果将为临床预测RP发生概率和严重程度提供明确可行的量化指导。

放射性肺炎（Radiation Pneumonitis，RP）严重制约了胸部肿瘤靶区照射剂量的提高，限制了疗效。放疗前需认真评估患者放疗后RP的发生概率和严重程度。目前临床常用的仍是上世纪90年代总结的经验阈值，假设全肺对照射的反应是均质同性的，使用全肺受低剂量照射的百分体积和平均照射剂量预测RP发生的概率。但该方法中使用的肺的体积是依据单次瞬时图像确定的，不同呼吸时相的CT 图像其肺的体积会有较大差别，影响所采集的全肺受低剂量照射的百分体积的准确性。其次肺不同部位的通气功能显然是不同的，特别是不同肺叶的活动程度，包括有无肺不张都会影响通气功能权重。.针对上述问题，本研究将基于4DCT的肺通气功能成像引入放射性肺炎的预测。首先从刚性位移层面和患者真实图像层面评价了两大变形配准程序RayStation和MIMvista的准确性，选定了更为准确的MIMvista算法用于后续生成基于4DCT的肺通气功能图像。此外本研究新提出了一种简化的基于4DCT的肺通气功能成像算法，并基于50例患者的4DCT和SPECT肺通气功能图像，量化比较了该算法与既往报道的三种常用算法，证明该算法生成的肺通气功能图像与金标准SPECT肺通气功能图像几乎高度相似，平均DSC值为0.5958，平均计算时间为15秒，准确性和计算效率显著优于既往算法（p≤0.002）。而且它不依赖于变形配准，避免了变形配准参数设定对最终结果的影响。同时该算法只需要AVG CT及其在AVG CT上定义的肺轮廓为输入数据，生成的肺通气功能图像与放疗计划中的剂量分布不存在位移偏差，无需进行配准，更便于临床使用。最后基于主成分分析、模糊聚类及BP神经网络建立了放射性肺炎预测模型，并证明使用功能体积参数预测放射性肺炎比原有的使用剂量体积参数预测更为准确。该结论有望促进以基于4DCT的肺通气功能图像为代表的肺功能图像在放射治疗中的应用。