基于保持参数的DTI各向异性滤波

弥散张量磁共振成像(DTI)易受噪声干扰,且现有DTI去噪方法仅对各方向通道的DWI数据进行单独处理,未充分考虑各通道之间相关性以及DTI数据本身结构和形态特点.针对这些不足,作者根据DTI数据结构及形态特征,综合DWI不同方向通道间相关性,重构统一特征值和特征向量的扩散张量,并增加保持参数对迭代滤波过程进行约束,实现DTI各向异性滤波去噪并降低了对DTI数据结构和形态特征的影响.将所提出的方法分别用于合成弥散张量数据仿真和真实脑部DTI数据实验.仿真和实验结果表明,该方法能有效减少噪声对DTI数据的影响,较之现有去噪方法效果更好.

DTI Anisotropic Filtering Based on Preserving Parameter

To overcome the drawback that the diffusion tensor magnetic resonance imaging (DTI) is vulnerable to noise and the current method deals with the DWI separately in different directions which ignores the correlation of the structure information and characteristics of the DTI, a novel anistropic filtering method synthetically considering the structure information and characteristics of the DTI was proposed based on the correlation of DWI in different directions. The smoothing structure tensor with unified eigenvalue and eigenvector was reconstructed and the iteration process was constrained by the preserving parameter. By this method, denoising simulations using a synthetic DTI dataset and experiments using an in vivo brain DTI dataset were done. The results of the simulations and experiments demonstrated that this method could remove the noise significantly and is more effective compared with the traditional methods.