Welcome Vittal!
The post that happened to come in just an hour prior to this one was largely on the same question, though posed slightly differently.
The connectomestats command does permit direct statistical inference upon connectivity matrix data, with a couple of statistical enhancement algorithms to choose from. We have however been hesitant to actively advocate for the use of such given the results of our own experimentation (which unfortunately remain unpublished for now). It can be used, but it’s important to not over-interpret the results. It’s also important if directly comparing edge-wise connectivity values between subjects to consider what other factors may introduce differential scaling of such values between subjects (discussed here).
You can of course alternatively perform some higher-order derivative calculation on the individual matrices and then compare those results across subjects. MRtrix3 doesn’t have a command for performing such matrix calculations but there are other software packages dedicated to such.
Rob