Comparing the LPCA and MPPCA denoising approaches for diffusion MRI using simulated human data

In this study, we investigate two denoising methods for diffusion MRI: the local PCA approach and Marchenko-Pastur (MP) PCA approach. Ground-truth diffusion-weighted images of the human brain are developed and used for noise simulation. Two diffusion-weighting b-values and two noise levels are generated as input data for both denoisers. Metrics of diffusion tenor imaging (DTI) and neurite orientation distribution and density (NODDI) are computed after denoising and compared between denoise methods.