Three-dimensional cosmic muon tomography reconstruction of reinforced concrete using Geant4 simulations and machine learning event reduction

Cosmic muon scattering tomography is an imaging technique capable of identifying and reconstructing dense objects, typically made of high-Z materials, in different scenarios like the inspection of decommissioned nuclear reactors and shipping containers at ports. To date, this technique has not found significant applications in the maintenance of civil engineering structures. In this paper, we demonstrate through Geant4 simulations that it is possible to reconstruct the load-bearing structures within reinforced concrete via cosmic muon scattering tomography. Using an ad-hoc developed muon source emulator capable of reproducing the cosmic muon spectrum at the Earth's surface, we achieved three-dimensional imaging of internal structures in concrete blocks with excellent spatial resolution. Finally, we illustrate how the use of modern neural networks can significantly reduce the exposure time required to reconstruct these images without compromising their resolution.