So I bet Andrew might also have some ideas towards this as well. But if I was doing the study, these are probably the steps I would follow:

Plot simulated hit (sum of G4 hits) and reconstructed hit (reconstructed tile) energies for (A) single muons, and (B) DIS events
For muons, we should see a nice MIP peak in the reconstructed hit energies (which we do!), and for DIS events, this peak should get washed out (which it does!).
This step has already been done, and can be seen in fig.s 8.187 of version 2.2 of the pTDR (these were prepared by Olaiya).
Then, it would be useful to see a rough "event display" for select single muon events
For one single muon event, you could make a plot that shows the reconstructed hit energy on the z-axis vs. its (eta, phi) on the (x, y) axes for each layer in the BIC and BHCal
The relevant collections would be the EcalBarrelImagingRecHits (for the BIC AstroPix layers), the EcalBarrelScFiRecHits (for the BIC scintillating fiber layers), and the HcalBarrelRecHits (for the BHCal, of course)
Examples of how to sort the BIC hits into numbered layers can be see in github.com/RuseTraveler/EpicBHCalPTDRStudies/reconstruction/macros/FillBHCalClusterCalibrationTuple.cxx
This "event display" could be visualized as either a grid of 2D heat maps, or — even better — a stack of lego plots organized by increasing radius
This could be compared against the same event display for select single electron or single pi+- events
We should mostly flat distributions + a single MIP for the muons, but very messy distributions for the electrons/pi+-
And for the electrons, of course, we should see essentially nothing in the BHCal
Next compare what we see in the BIC and BHCal against tracks using truth information
Reconstructed muon tracks can be selected using MC-track associations (CentralCKFTrackAssociations should be the relevant collection) to identify tracks associated with a muon
Next the associated BIC and BHCal projections for the muon tracks can be selected from the CalorimeterTrackProjections collection by finding projections which have a track with the same object ID as the muon tracks
Examples of how to work with the track projections can be seen in github.com/eic/snippets/Calorimetery/MatchClustersToTrackProjections
For each calorimeter, the track projections project to the face (the "surface" index will be a lower number) and to some no. of interaction lengths into the calorimeter (the "surface" index will be higher)
Then for each projection, plot the energy of the reconstructed hits within some (eta, phi) distance of the muon track projection
Care will be needed for the BIC, since we'll only want to see hits from the ScFi layer closest in radius to the BIC projection
This could be done for both single muon events, and DIS events. The former should show a strong MIP peak like before, and — I would hope — the DIS events will now also show a visible-but-weaker MIP peak
This process could be repeated for both electrons and pi+- for comparison
Finally, the last step would be try to tag reconstructed tracks as muons using MIPs
First I would try to pick out only lepton tracks using PID info (i.e. the LUTs for now)
You would start with ReconstructedChargedParticles, and select only tagged as an electron (since the PID systems won't be able to distinguish electrons vs. muons)
Then you can pick out the track associated with the reconstructed electron using its relations, and from there retrieve the relevant projections like in the previous stage
Then, based on what we saw in the previous stage, you could tag a track as being a muon if there's a hit with an energy consistent with a MIP within some (eta, phi) distance from all 4 projections in the BIC and BHCal
Naturally, there will need to be some tuning of the thresholds
And it might be useful to also fold in some sort of isolation cut on the sum of energies of all hits within that (eta, phi) distance
Finally, you can calculate the muon tagging/reconstruction efficiency and purity