自适应图像全变差约束的有限角度CT重建算法

针对有限投影角度的CT图像重建问题，提出一种改进的基于自适应图像全变差（Total p Variation, TpV）约束的代数迭代重建算法。改进算法采用两相式重建结构，首先利用代数重建技术（ART）算法重建中间图像并做非负修正，然后利用自适应TpV正则项约束图像稀疏特性，进一步优化重建结果，其中正则项可根据图像区域特性自适应的调整决定平滑强度的参数p，两项交替进行直到满足收敛要求。本文应用经典的Shepp-Logan体模对改进算法进行仿真重建，以重建图像及其局部放大图作为主观分析依据，以profile图和归一化绝对距离值作为客观评估标准，与经典的ART-TV算法进行比较，对比分析重建结果发现:本文算法重建图像不仅与真实体模更接近，重建误差更小，而且能更好地保护图像的边缘特性。

An adaptive total p variation-constrained CT image reconstruction method for limited angle data

An improved algebraic iterative reconstruction algorithm based on the total p variation (TpV) was proposed for the reconstruction of CT images with limited projection angles. Two-phase reconstruction structure was applied in the improved algorithm. The algebraic reconstruction technique was firstly adopted to reconstruct image that meet the identity of projection data and applied to correct non-negativity. Subsequently, adaptive TpV regularization was used to constrain the image sparse features, further optimizing the reconstruction results, in which the parameter p was adjusted according to the image region characteristics. Two phases alternated with each other until meeting the convergence criteria. The classical Shepp-Logan phantom was applied to perform a simulation reconstruction for the improved algorithm. The reconstructed images and its local zoom-in views were used for subjective analysis, while the profile graphs and the normalized absolute distance were regarded as the objective evaluation criterions. The results demonstrated that compared with the images reconstructed by conventional algebraic reconstruction technique-total variation algorithm, the images reconstructed with the improved algorithm were not only close to the real Shepp-Logan phantom with smaller reconstruction errors, but also achieve a better protection for the edge features of images.