Spatial Prior Informed Constrained Spherical Deconvolution in Super Resolution Variation

Constrained Spherical Deconvolution (CSD) is a state-of-the-art technique for unraveling complex fiber configurations of white matter from diffusion MRI data. Despite its ability to identify multiple fiber orientations, it faces difficulties in resolving fiber crossings with inter-fiber angles below 40-45 degrees at low spherical harmonic orders (i.e., lmax=8) representations. On the other hand, the super-resolved CSD struggles with handling noise at high spherical harmonic orders (e.g., lmax>=12) due to its ill-posed nature, resulting in unstable estimations with spurious lobes. To address these limitations of CSD, our study introduces Spatially Regularized Super-Resolved CSD (SR2-CSD), utilizing the spatial correlations of fiber Orientation Distribution Functions to enhance the angular resolution and noise robustness.