Improvements in Track Reconstruction of Multi-Particle Events in the Mu2e Detector

The Mu2e experiment at Fermilab is designed to probe charged lepton flavor violation (CLFV) via the coherent transition of a muon into an electron on an aluminum nucleus, with an expected single-event sensitivity of $R_{\mu e} < 6.2 \times 10^{-16}$ at 90\% C.L. Track reconstruction begins by grouping tracker hits into clusters based on timing and longitudinal position. The current algorithm reconstructs a single helical trajectory per cluster, which limits efficiency in events with multiple particles, such as antiproton-induced backgrounds and photon conversions ($\gamma \rightarrow e^{+}e^{-}$) from radiative pion capture. To address this, we present a data-driven pattern recognition algorithm designed to identify multiple helices within a single cluster. The method exploits the Mu2e tracker geometry, where helical trajectories become approximately linear in the azimuthal–longitudinal ($\phi$–$z$) plane, enabling the classification of hits into segments based on slope. This approach enhances track finding in complex topologies and improves identification of track candidates in calibration and physics events. We will discuss the methods and features employed in our algorithm and current evaluation results.