Smart Polarization and Spectroscopic Holography for Real-Time Microplastics Identification
Yanmin Zhu,Yuxing Li,Jianqing Huang,Edmund Y. Lam
DOI: https://doi.org/10.1038/s44172-024-00178-4
2024-01-01
Communications Engineering
Abstract:Optical microscopy technologies as prominent imaging methods can offer rapid, non-destructive, non-invasive detection, quantification, and characterization of tiny particles. However, optical systems generally incorporate spectroscopy and chromatography for precise material determination, which are usually time-consuming and labor-intensive. Here, we design a polarization and spectroscopic holography to automatically analyze the molecular structure and composition, namely smart polarization and spectroscopic holography (SPLASH). This smart approach improves the evaluation performance by integrating multi-dimensional features, thereby enabling highly accurate and efficient identification. It simultaneously captures the polarization states-related, holographic, and texture features as spectroscopy, without the physical implementation of a spectroscopic system. By leveraging a Stokes polarization mask (SPM), SPLASH achieves simultaneous imaging of four polarization states. Its effectiveness has been demonstrated in the application of microplastics (MP) identification. With machine learning methods, such as ensemble subspace discriminant classifier, k-nearest neighbors classifier, and support vector machine, SPLASH depicts MPs with anisotropy, interference fringes, refractive index, and morphological characteristics and performs explicit discrimination with over 0.8 in value of area under the curve and less than 0.05 variance. This technique is a promising tool for addressing the increasing public concerning issues in MP pollution assessment, MP source identification, and long-term water pollution monitoring.