选择性双向顺序压缩感知重建动态磁共振成像

压缩感知是一种新兴技术，该技术能够用远低于奈奎斯特采样频率采集的信号恢复出原始信号. 压缩感知成像方法大大提高了心脏磁共振成像的采集速度，已有的方法主要利用动态图像时间相关及心脏的周期性运动特征，如采用在时间维做傅立叶变换或求解每帧数据跟参考帧数据的差异获取稀疏数据，满足压缩感知重建的要求. 该文提出了选择性双向顺序压缩感知重建算法，利用相邻帧的差异更小的特点，获取更加稀疏的差异数据，同时利用动态图像的周期性，以目标函数积分为判据，在时间顺序和时间逆序两个方向选择效果更好的方向进行数据重建，降低图像伪影和噪声. 该选择算法，可以在不增加重建时间的情况下，选择双向顺序重建中最佳的结果. 该文对心脏磁共振图像数据进行了数据处理实验，并且跟传统压缩感知算法、参考帧差异方法及匙孔成像方法进行了比较. 结果表明：该方法无论从视觉效果还是从统计结果上，都有很大的改善.

Selective Dual-Direction Sequential Compressed Sensing for Dynamic MR Imaging

Compressed sensing (CS) is a new technique for reconstructing image from undersampled k-space data. In dynamic imaging, data sparsity can be introduced by applying Fourier transformation along the temporal dimension or calculating difference images with respect to a reference frame. In this work, a selective dualdirection sequential CS method was proposed to achieve greater data sparsity in dynamic MR imaging, by calculating difference images between adjacent frames. Both time sequential or inversed time sequential reconstruction were used for better reconstruction results, while the total computation time remained the same. Experimental results demonstrated that the proposed algorithm showed improved quality relative to the traditional compressed sensing algorithm, differencing algorithm and keyhole imaging technology.