CSD from precomputed FA maps


Hello all!
I am new to the community and have a question regarding constrained spherical deconvolution.
I am hoping to perform CSD using precomputed FA maps. (I have performed harmonization of the maps across a number of different scanner parameters.) I see that there is a version of CSD that used FA, but I have had no luck figuring out a method of forcing in precomputed FA.
Is it possible?
Thanks so much for any advice!! :slight_smile:


Hi Julia,

I think there may be a fundamental misunderstanding regarding what CSD is / does. A Fractional Anisotropy map contains just a single scalar value per voxel; therefore there is no way that such data could be used as input to a process like CSD that estimates higher-order information in each voxel.

Perhaps where you have read that there is “a version of CSD that used FA”, what was in fact being referred to was the process of response function estimation, which is a requisite step that must be performed prior to CSD. Further, even in the case of using FA information for that process (the dwi2response fa algorithm in MRtrix3), the FA maps themselves are still not provided as input: the algorithm selects voxels from which to construct the response function based on an internal calculation of FA, but the algorithm itself still needs the full DWI data to be provided as input in order to estimate the response function parameters based on the DWI intensities.