基于各向异性扩散滤波与非局部先验的PET图像重建算法

PET Image Reconstruction Algorithm Based on Anisotropic Diffusion Filtering and Nonlocal Prior

针对在正电子发射断层成像中,经典的惩罚最小二乘算法和基于局部先验惩罚的改进算法不能有效地抑制噪声并保护图像边缘的问题,提出了一种新的各向异性扩散滤波非局部惩罚最小二乘重建算法(PLS-PDE).新算法首先对图像进行基于非局部二次先验的最小二乘估计,接下来对此估计值进行基于双向扩散系数的各向异性扩散滤波.所提出的算法综合了非局部先验与各向异性扩散滤波的优点.实验结果表明,新算法在抑制噪声和边缘保护方面取得了良好的折中,较大程度地改善了重建图像的质量.

Aiming at the problem of not effectively suppressing noise and preserving edge in normal penalized least squares algorithm and the improved algorithm based on local prior during reconstructing positron emission tomography image, a new penalized least squares reconstruction algorithm incorporating nonlocal prior and anisotropic diffusion filtering was proposed. Firstly, the image was estimated by least squares algorithm with nonlocal prior. Then, this estimated image was filtered by anisotropic diffusion filter with bidirectional diffusion coefficient. The new algorithm synthesized the advantages of nonlocal prior and anisotropic diffusion filtering. The experimental results show that the proposed algorithm achieves a good balance between reducing noise and keeping edge. The quality of reconstruction image is significantly improved.