Voxel-level denoising of diffusion MRI using a rank-1 decomposition

A voxel-wise rank-1 decomposition in spherical harmonics of the dMRI signal allows for a denoising method with minimal assumptions on signal and noise distributions. A voxel-wise denoising strategy is compared to MP-PCA on in vivo data and simulated data. The rank-1 decomposition and MP-PCA show visually similar results on the in vivo data, indicating that a voxel-level denoising approach has potential. However, results on the simulations differ. This may have various explanations; further investigation is necessary to strengthen the credibility of a rank-1 decomposition as a denoising approach.