Abstract:This study evaluates the influence of the algorithm used to calculate conditional ionisation cluster size distributions (ICSD) and nanodosimetric parameters from proton track structures obtained with Geant4-DNA simulations. The Uniform Sampling (US) and Associated Volume (AV) sampling algorithms and the DBSCAN clustering algorithm were used to calculate the ICSD conditioned on cluster sizes $\nu \ge 1$. In the AV algorithm, we explored two configurations: one where sampling volumes could overlap ("Overlap") and one where volumes could not overlap ("No Overlap"). The Overlap configuration was corrected with the weight function $w(\nu)=\nu^{-1}$ to eliminate the bias towards regions of high density of ionisations. In sampling algorithms, clusters were defined as the number of ionisations in spheres of 3.4 nm in diameter. In DBSCAN, clusters were formed according to a proximity criterion (maximum distance of 3.4 nm) between two ionisations - no additional geometrical restrictions were applied. The conditional ICSD, the mean cluster size, and a set of complementary cumulative frequencies were compared. We considered differences in the mean difference larger than two standard deviations statistically significant. Overall, differences between the No Overlap and the Overlap configurations and between the US and AV algorithms were statistically significant but relatively small, except when statistical fluctuations dominated. DBSCAN always resulted in significantly larger mean cluster sizes. We showed that the fast and efficient AV can be used instead of the relatively inefficient US to calculate conditional ICSDs without significant loss of accuracy. The differences observed between DBSCAN and the other algorithms are understood and originate in the fact that while the scoring volume is fixed for the sampling algorithms, for DBSCAN, it evolves according to the distance between ionisations.

Simulation of particle identification with the cluster counting technique

Simulation study of particle identification using cluster counting technique for the BESIII drift chamber

Cluster counting algorithms for particle identification at future colliders

Emulsion Cloud Chamber technique to measure the fragmentation of a high-energy carbon beam

Three-Dimensional Measurement of Aerosol Particle Clustering in Homogeneous Isotropic Turbulence

Improved Particle Identification Using Cluster Counting in a Full-Length Drift Chamber Prototype

Cluster Counting Algorithm for the CEPC Drift Chamber using LSTM and DGCNN

Nanodosimetric characterisation of proton track structure: a comparison of sampling and clustering algorithms using Geant4-DNA

New method of particle identification with tracker detectors

The study of intelligent algorithm in particle identification of heavy-ion collisions at low and intermediate energies

Some Progress on Interferometry Techniques Applied to Particle Measurements

Study of the single cluster response of a helium-isobutane drift chamber prototype using 8 keV X-rays

Particle identification capability of a homogeneous calorimeter composed of oriented crystals

Study on Parallel Computing of Radiation Detector Simulation Based on Geant4

Study on Non-uniformity of Particle Distribution Using Cluster Discrete Element Method

Particle Identification and Its Efficiency for Small Angle GDH Experiment in JLab Hall-A

Machine-learning-based particle identification with missing data

A Study of Particle Motion in the Presence of Clusters

Non-identical particle femtoscopy at sqrt(s_NN)=200 AGeV in hydrodynamics with statistical hadronization