Imaging-Based Assessments of Response

项目总结(见说明)： 基于图像的分析作为一种改进放射治疗的手段，以根据个别患者的预后或早期治疗反应定制治疗，是该计划项目成功的关键研究领域。项目3研究的假设是，鉴于肿瘤和正常器官的潜在异质性及其对治疗的反应，解剖、代谢和生理图像(如本计划的临床项目中获得的)最适合用于个体化治疗，方法是强有力地识别比单个体素或整个肿瘤/正常器官更具预测性的子体积，并将这些子体积作为辐射剂量再分配的空间指南，例如焦点增强。目标1将改进生理图像的重建。目的2是研究识别预测反应的亚体积的方法。目的3将提高亚体积映射到患者的准确性，以指导放射治疗剂量分布的修改。将开发用于子体积识别的模式识别技术，并将其与结果(例如，局部故障位置)相关联。 将研究来自不同成像方法(例如，胶质瘤的11C甲硫氨酸PET和DCE MRI)用于局部反应预测的互补和冗余信息，以及成像信号和提取的子体积(从测试-重新测试数据)的重复性。方法改进生理成像和分析，包括通过模式识别直接分析药代动力学，稀疏采样并行MR！将研究使用压缩传感进行重建以解决器官运动并提高时间分辨率和/或体积覆盖。为了指导个体化治疗，子体积需要被几何地映射到被治疗的患者，由于图像配准精度的限制而具有不确定性。有限元模型将作为地图集进行研究，以正则化基于强度的 用于将这些子体积放置在治疗计划CT扫描空间中的可变形对准方法，所述CT扫描是为了计划和/或修改治疗而获得的。

PROJECT SUMMARY (See instructions): Image-based analyses as a means of modifying Radiation Therapy to customize treatment based on an individual patient's prognosis or early response to treatment is a research area crucial to the success of this program project. Project 3 investigates the hypothesis that, given the underlying heterogeneity of tumors and normal organs and their response to treatment, anatomic, metabolic and physiologic images (as acquired in the clinical projects of this program) are best utilized for individualizing therapy by robustly identifying subvolumes that are more predictive than individual voxels or the entire tumor/normal organ as a w/hole, and providing these subvolumes as a spatial guide for radiation dose redistributions such as focal boosting. Aim 1 will improve reconstruction of physiological images. Aim 2 is to investigate methods of Identifying subvolumes predictive of response. Aim 3 will improve the accuracy of mapping subvolumes to patients to guide modification of radiation therapy dose distributions. Pattern recognition techniques will be developed for subvolume identification, and correlated with outcomes (e.g. sites of local failure). Complementary and redundant information from different imaging methods (e.g. 11C Methionine PET and DCE MRI in gliomas) for local response prediction will be studied, as well as the reproducibility of imaging signals and extracted subvolumes (from test-retest data). Methods to improve physiological imaging and analysis, including direct analysis of pharmacokinetics via pattern recognition, and sparsely sampled parallel MR! reconstructed using compressed sensing to resolve organ movement and improve the temporal resolution and/or volumetric coverage, will be investigated. To guide therapy individualization, subvolumes need to be mapped geometrically to the treated patient, with uncertainties due to the limits of image registration accuracy. Finite element models will be investigated as atlases to regularize the intensity-based deformable alignment methods used to place these subvolumes in the space of treatment planning CT scans acquired for the purposes of planning and/or modifying treatment.