Particle identification is a critical and challenging task in high-energy physics experiments, particularly for future collider facilities such as FCC-ee and CEPC. The conventional dE/dx method is limited by significant uncertainties in total energy deposition, which constrain its particle separation capability. The cluster counting (dN/dx) technique, instead, exploits the Poisson nature of primary ionization, providing a statistically robust approach to infer particle mass. Simulation studies performed with Garfield++ and Geant4 demonstrated that dN/dx can achieve up to twice the resolution of dE/dx in helium-based drift chambers. This improvement, however, requires precise detection of electron peaks and identification of ionization clusters—tasks complicated by signal overlap in the time domain. In this work, advanced algorithms were developed to enable accurate identification of electron peaks and ionization clusters. These methods were validated using beam test data collected at CERN with several drift tubes operated with different helium gas mixtures. The analysis results and a comparative study of the resolutions achieved using dN/dx and dE/dx methods are presented and discussed.