Hi experts,

I have a set of dwi data (b=0, 500, 1500, 2500, 3500) in a single series and want to extract the b=1500 and b=2500 and generate trace weighted images (geometric mean?) for DKI estimation later.

I’m very new to Mrtrix3. So far only manager to extract the individual shell data, but the mrmath mean doesn’t seem to match with my goal.

Thank you guys in advance!

ChatGPT helps me find an way to compute the trace weighted images.

1.Compute the Logarithm:

mrcalc input.mif -log log.mif

computes the logarithm of each voxel in each volume of the DWI data.

2.Calculate the Arithmetic Mean:

mrmath log.mif mean log_mean.mif -axis 3

calculates the mean of these logarithms along the 4th dimension (axis 3).

3.Exponentiate the Result:

mrcalc log_mean.mif -exp trace.nii

exponentiates the mean logarithms to obtain the geometric mean.

This approach will yield the geometric mean of the signal intensities, which corresponds to the trace-weighted image.