Abstract: Generalized Auto-calibrating Partially Parallel Acquisition(GRAPPA) algorithm assumes that the reconstruction kernel is shift invariant in the whole K-space, it often generates artifacts and noise in the clinic. Aiming at this problem, this paper proposes an GRAPPA reconstruction algorithm based on anisotropic diffusion. It is used to remove the noise and the singular point in K-space data, while maintaining the phase information, making the reconstruction more accurate. Vivo experimental reconstruction demonstrats that the proposed algorithm can reduce noise and artifacts at any under-sampling condition, and gets higher quality data reconstruction.

Key words: Magnetic Resonance Imaging(MRI), parallel imaging, Generalized Auto-calibrating Partially Parallel Acquisition(GRAPPA) algorithm, K-space, Auto-calibration Signal(ACS), anisotropic diffusion

摘要： 全局自动校准部分并行采集(GRAPPA)算法假设插值核在整个K空间内具有平移不变性，在实际应用中容易引起重建伪影和噪声放大。为此，提出一种基于各向异性扩散的GRAPPA重建算法。利用偏微分方程设计各向异性扩散重建模型，对GRAPPA算法合成后的数据进行各向异性扩散，在保证相位信息正确的情况下，去除K空间中的噪声和奇异点，从而提高重建图像的准确率。对活体实验数据的重建结果表明，该算法能减少噪声和伪影，提高重建图像的信噪比。

关键词: 磁共振成像, 并行成像, 全局自动校准部分并行采集算法, K空间, 自动校准信号, 各向异性扩散

CLC Number: 

• TP911.73