Generalized Constant Solid Angle ODF and Optimal Acquisition Protocol for Fiber Orientation Mapping

Recent advances in diffusion MRI have allowed for improved understanding of the white matter connectivity. Models like the Diffusion Tensor, diffusion Orientation Distribution Function (ODF) with a monoexponential signal decay have shown good fiber reconstruction accuracies. More complex radial signal decay models, like the bi-exponential model, have been shown to better approximate the in-vivo diffusion signal. In this paper, we generalize the Constant Solid Angle ODF (CSAODF) algorithm to handle any q-space sampling and exploit the biexponential model. Simulation results to optimize the reconstruction and acquisition parameters are described. Finally, the algorithm is validated on human brain data. Our generalized CSA-ODF model performs optimally with 200 q-space data points distributed over three shells acquired at b = 1000, 2000s/mm and in the range [3000, 6000]s/mm for the third shell. Crossings up to about 30 degrees can be recovered, and fiber orientations can be detected with a precision of about 1 degree.