面向肺肿瘤精确放射治疗的肺4D-CT图像超分辨率重建研究

4D-CT plays an important role in lung cancer treatment because of its capability in providing a comprehensive characterization of respiratory motion for high-precision radiation therapy. However, due to the inherent high-dose exposure associated with CT, dense sampling along superior-inferior direction is often not practical, thus resulting in low inter-slice resolution. The lack of sufficient structural information along the superior-inferior direction introduces image artifacts such as lung vessel discontinuity and partial volume effects. Consequently, the correct assessment of tumor could be severely affected due to shape distortion, which might mislead dose administration in radiation therapy. The main goal of this research is to enhance the resolution along the superior-inferior direction of lung 4D-CT data without increasing the scanning dose and time. .In this project, we present a novel idea for lung 4D-CT super-resolution. Based on the idea and our previous work, we will mainly focus at theory analysis, model reconstruction and optimization. The major topics include: (1) Build a multi-prior constraints lung 4D-CT super-resolution sparse representation model, and optimization solution; (2) A combination of multi-mixed feature and manifold theory, optimize the expression of the model from the Euclidean space to the manifold space, and study the optimization solution for manifold constraints model; (3) Based on the above model, build the joint sparse model for super-sparse 4D-CT data super-resolution reconstruction. Finally, the high-precision radiotherapy for lung cancer can be achieved.

肺4D-CT图像能够很好地跟踪肺肿瘤靶区的运动，基于4D-CT图像引导的精确放射治疗已成为肺肿瘤放射治疗的重要发展方向。然而当前肺4D-CT 数据面临着一个主要问题：层间分辨率过低。这种低分辨率数据对肺肿瘤放射治疗精度将产生显著影响。因此，研究如何在不增加扫描剂量的基础上，提高肺4D-CT图像分辨率有着重要实际意义。本项目拟提出一种新的适用于肺4D-CT图像超分辨率重建思想。基于此思想，申请者在前期研究工作基础上，从理论分析、模型建立与求解、模型优化等方面展开研究，主要内容包括：（1）构建多先验信息约束肺4D-CT 超分辨率稀疏表达新模型，并优化求解；（2）结合多维混合特征、流形空间理论，建立从欧拉空间到流形空间的优化表达模型，并研究流形约束下的模型优化求解；（3）研究超稀疏肺4D-CT数据的超分辨率重建，构建联合稀疏表达模型。通过以上研究，最终实现放射治疗过程中的肺肿瘤4D精确跟踪。

肺4D-CT图像能够很好地跟踪肺肿瘤靶区的运动，基于4D-CT图像引导的精确放射治疗已成为肺肿瘤放射治疗的重要发展方向。然而当前肺4D-CT 数据面临着一个主要问题：层间分辨率过低。这种低分辨率数据对肺肿瘤放射治疗精度将产生显著影响。因此，研究如何在不增加扫描剂量的基础上，提高肺4D-CT图像分辨率有着重要实际意义。按照项目所制定的研究计划，本项目以“提高肺4D-CT图像分辨率”为核心，重点研究了（1）基于稀疏表达的肺4D-CT图像分辨率增强模型。包括理论分析、模型构建及优化求解；（2）基于非局部平均（Non-local Means）的肺4D-CT图像分辨率增强模型。包括图像的自适应分块、图像块相位间信息的传递机制，以及块之间的融合策略。（3）基于配准的稀疏肺4D-CT超分辨率重建方法，包括稀疏图像配准、图像块自适应选择、迭代重建策略等。项目在预定研究目标和学科发展趋势的导引下，课题组完成了预定的研究计划，取得了一些研究成果，完全达到预期目标。