Abstract

Info

Title:: Intravoxel fiber structure estimation based on non-local mean smoothing

Author(s):: CHU Chunyu; College of Engineering, Bohai University, Jinzhou 121013, China

Keywords:: ; Keywords: diffusion magnetic resonance imaging; intravoxel fiber structure; spherical deconvolution; non-local mean

Abstract:: Abstract: To overcome the noise sensitivity of spherical deconvolution (SD) method on diffusion-weighted magnetic resonance image, a non-local mean smoothing-based method for intravoxel fiber structure estimation is proposed. A non-local mean method is firstly established for smoothing diffusion-weighted magnetic resonance images, and then the intravoxel fiber structure is estimated by SD method based on the smoothed diffusion-weighted magnetic resonance images, thus improving the noise immunity of SD method. The results on synthetic and physical phantom data show that the intravoxel fiber structure obtained by the proposed method not only has smaller mean angular errors than that obtained by SD method or SD combined with conventional non-local mean smoothing, but also has fewer edge blurring.