DTI图像恢复的向量复扩散模型

Restoring DTI images by vector-valued complex diffusion model

为了减小扩散张量图像(DTI)中广泛存在的赖斯噪声影响,提出了向量复扩散模型。该模型是标量复扩散模型的推广和发展。为了评价该模型的去噪性能,对向量图像-扩散加权(DW)图像进行了恢复实验。基于模拟和真实数据进行的实验表明,相对于标量复扩散滤波器,向量复扩散滤波方法得到的PSNR和SMSE数值更高,追踪到的纤维数量更多、长度更长,故其去噪性能优于标量复扩散模型。另外,在信噪比较低情况下该模型优于实数域P&M向量滤波器。

To decrease the effects of the Rician noise introduced into the diffusion tensor images, vector-valued complex diffusion model is presented. The presented smoothing strategy is the development of the scalar typed model. To evaluate the efficiency of the proposed model in removing Rician noise of the vector valued data-DW images, experiments based on synthetic and real data are designed. All the experiment results prove quantitatively and visually the better performance of the presented filter, compared with the scalar typed filtering model, by higher PSNR and SMSE metric values and more, longer fibers. Furthermore, the presented model filters the images better at low SNR, compared with the vector-valued P＆M filter.