Automated segmentation and clustering of spICP-ToF-MS time series

Project Summary

The detection and characterization of inorganic nanoparticles (NPs) in natural matrices has relied primarily on single-particle ICP-MS (spICP-MS). In practice, the continuous introduction of an aqueous environmental sample produces element-specific high-frequency time series consisting of background noise randomly interspersed with peaks corresponding to individual NPs. With new generation ICP-ToF-MS, the large amount of data generated poses a challenge for traditional methods to effectively perform NP peak detection. This study proposes a machine learning (ML)-based method for automatic segmentation of spICP-ToF-MS time series using simulated single-channel scenarios with varying parameters to demonstrate its feasibility and robustness.

Mickaël Tharaud
tharaud@ipgp.fr

• Research engineer at IPGP
  (specialist in analytical chemistry, with a focus on the detection and characterization of inorganic nanoparticles in aquatic systems thanks to plasma source mass spectrometry)

Jiachen Zhang

Pierre Emmanuel Peyneau

Léonard Seydoux