Determine the particle-type dependence of medium-composition effects on classification

Determine whether changes in the surrounding medium influence the classification of some microplastic particle types more strongly than others in RF dielectric spectroscopic cytometry with machine-learning classification.

Background

The study evaluates machine-learning classification of eight microplastic types using RF scattering-parameter measurements in deionized water and tests polyethylene terephthalate in carrier media containing up to 6.6% sea salt. PET classification remains robust under these salinity changes, but the authors note that salt water and deionized water produce different baseline S-parameters and signal magnitudes, and that salt may alter particle properties or RF sensing fields.

Because only PET is evaluated in saline media, the results do not establish whether changes in carrier-medium composition affect all microplastic classes similarly. The unresolved issue is whether medium-dependent classification degradation is specific to particular particle types, which is important for applying the RF-based method to diverse environmental water samples.