中文摘要：

多通道欠采样非笛卡尔轨迹数据重建是当前磁共振成像的研究热点,当欠采样因子比较大时,病态问题往往使得敏感度编码（SENSE）方法重建图像信噪比严重降低,传统的解决方法是在重建方程中引入Tikhonov约束或TV约束。提出自适应约束的SENSE重建算法,由先验图像的梯度特征并借鉴PM模型的思想决定惩罚函数,在梯度幅值较大的区域使用各向异性扩散的TV约束方式,在梯度幅值较小的区域使用各向同性扩散的Tikhonov约束方式。进行8通道2.6倍欠采样可变密度螺旋轨迹人体动静脉畸形瘤动脉注射X线的仿真实验。结果表明,与平方和（SOS）重建方法、传统无约束SENSE重建方法以及TV约束SENSE重建方法相比,本算法可以有效抑制部分数据成像带来的噪声和伪影,并能较好保护图像细节尤其是小细节信息,成像效果优于传统方法。

英文摘要：

The reconstruction of non-Cartesian under-sampled data from multi-channel MRI acquisitions is of the current research focus.The signal-to-noise ratio（SNR） of SENSE reconstruction is usually degraded by the ill-conditioned problem especially at large acceleration factor.Regularization methods,based on Tikhonov and TV are the most widely used methods and have been shown to be effective in alleviating the problem.This article proposed an adaptive constraint model for SENSE that made use of the gradient feature of the prior image and combined with the PM model to decide the penalty function to deal with non-Cartesian data from multiple coils.TV based constraint,an anisotropically smoothing method is applied in regions of higher gradient while Tikhonov regularization method is applied in more ambiguous regions with lower gradient.Data were simulated using collected projection X-ray of an arterial bolus injection in a patient with an AVM with an 8-channel head coil,and the undersampling factor is 2.6.Simulation experiments show that our method has better performance in removing the noise and artifact caused by under-sampled data with a better image quality and protecting image edge information in details,compared with the other reconstruction methods such as conventional SOS、SENSE and TV constraint SENSE.