基于预扫描正则化与稀疏约束重建的低剂量CT灌注成像

CT灌注成像中患者会受到长时间的X线照射,在灌注前预先扫描一幅正常剂量图像,在后续灌注过程中进行低剂量采集,将所获低剂量图像与参考图像做减影并进行滤波处理以获得灌注信息,然后叠加到正常剂量图像中,可以非常显著地降低CT灌注成像中辐射剂量,然而,当剂量非常低的时候,重建图像会受到噪声与伪影干扰。本文基于类似的预扫描正则化数据采集方案,分别进行正常剂量和低剂量预扫描,然后利用相同的低剂量进行后续灌注过程的扫描,将所采集低剂量数据与预扫描的低剂量数据在投影域作差,然后在重建中引入稀疏性约束,以获得更准确的灌注信息重建,重建后的灌注信息最后同样叠加到正常剂量图像中。本文采用一套人脑CT灌注图像进行模拟实验,结果表明在同样剂量下,本文所提新方法重建结果所包含的灌注信息更准确,时间衰减曲线与正常剂量情况的吻合度更高,而且平均过渡时间可重复性好。

Low-dose CT perfusion imaging based on pre-scan regulation and on reconstruction with sparsity constraints

The long-period CT perfusion imaging leads to an excess amount of radiation dose to the patient. However, the radiation dose could be significantly reduced if a previous normal-dose image is acquired before a set of low-dose scans of perfusion, and a filtering processing is performed on the differences between the current low-dose images and the previous normal-dose image, then the results are added to the previous image. But the selection of plenty of parameters makes the algorithm complicated. This paper proposes an innovative approach performed in sinogram domain instead of in image domain. First a normal-dose image and a set of low-dose projection data are acquired before the perfusion. Second the perfusion information is commendably reconstructed with sparsity constraints of the differences between current low-dose perfusion sinograms and previous low-dose sinogram. Finally, the reconstructed perfusion information is added to the previous normal-dose image. The proposed method was validated by simulated experiments with a set of brain CT perfusion images, which showed that the new method provided more accurate perfusion information; the time-attenuation curve was more close to that for normal-dose scan and the mean transit time more repeatable.