One potentially interesting feature of quantifying tract volume is that the measurement can be dependent on the resolution of the voxel grid to which the streamlines are mapped. E.g. Imagine that your template image is a single voxel that encapsulates the entire brain image FoV: you would count at least one streamline within this voxel, and conclude that the volume of the tract is larger than the volume of the brain 
Conversely, if the spatial resolution is too small and/or the number of streamlines is inadequate, then a basic mapping to a voxel grid followed by thresholding may yield a lot of holes within the tract and hence under-estimate volume. Food for thought…
Also for the mrthreshold command, I had difficulty choosing the option to control the thresholding?
By “difficulty”, do you mean that you could not choose which thresholding mechanism was most appropriate for your use case? If so, this is entirely commensurate with the ambiguity of the problem. You would at least not an observation of variations in tract volume across participants to have arisen solely from variations in how such thresholding was computed; so you could probably work backwards from there to think about which approaches are more or less appropriate. I certainly wouldn’t use the -top option, since it’d select the same number of voxels per participant and hence give an identical volume per participant…
BUT I was not able to calculate the volume of the tract by multiplying the volume of the voxel … So maybe one of the experts here … could correct me and add their input
Very well…
echo $(mrstats bin_out_tract.mif -mask bin_out_tract.mif -output count)*$(mrinfo bin_out_tract.mif -spacing | cut -d " " -f 1-3 | tr " " "*") | bc
Rob