Synthesis of DWI volumes from tensor

Hey Rob,

I wanted to generate a simulated DWI dataset from given tensor /eigen vectors. It would be helpful if you could please point me to the right direction to MRtrix libraries so that I can simulate a set of DWI from given Eigen vectors.

All the best,


Hi Yixin,

You’d need both a b=0 volume, and either the tensor coefficients or the combined eigenvectors and eigenvalues, in order to generate predicted DWI volumes.

Appendix B of the MRtrix3 manuscript shows a basic example of how the C++ command syntax is set up and image data are accessed. You would need to provide either a template image containing a gradient table, or a gradient table text file, from which you could pre-generate the b-matrix using this function. Then, using a functor (see example code in the manuscript appendix), one would take as input the tensor matrix and b=0 signal intensity in a voxel, and output the predicted DWI intensities, based on simply applying the tensor forward model (which I don’t think there’s an existing function for, but it’s just linear algebra, for which we use Eigen throughout MRtrix3).

Hope that’s enough to set you along the right path

1 Like