Scholarlink

<p>Microplastics (MPs), as emerging pollutants in the environment, have attracted widespread attention due to the potential threats they pose to ecosystems through their accumulation in soil. With the promotion and application of biodegradable plastics, the pollution problem of their derived biodegradable MPs has become increasingly prominent. However, current detection of MPs in soil focuses on traditional non-biodegradable MPs. To address this issue, this study presents a laboratory-based proof-of-concept for the classification of MPs using Short-Wave Infrared (SWIR) hyperspectral imaging (HSI) coupled with machine learning algorithms including Support Vector Machine (SVM), Artificial Neural Networks (ANN), Convolutional Neural Network (CNN). Systematic evaluation confirmed the feasibility of distinguishing biodegradable (PLA, PBAT, PHA) and conventional (PA, PP, PVC) MPs across three distinct soil matrices. Under controlled conditions, the CNN demonstrated superior performance, achieving a peak overall accuracy of 95.98%. Compared with traditional destructive thermal analysis, this method validates the intrinsic spectral separability of biodegradable polymers in soil mixtures (SM). This study establishes a theoretical foundation for the future development of rapid, non-destructive screening technologies for soil MPs pollution.</p> Graphical abstract <p></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Classification of biodegradable microplastics in soil using short-wave infrared hyperspectral imaging and machine learning: a laboratory-based proof-of-concept study

  • 期刊论文

  • 发表日期：

    2026年3月7日

查看全文
  • Lei Chen,
  • Yukun Tian,
  • Hongyang Song,
  • Jiakang Li,
  • Rongjia Fu,
  • Wenhui Hao,
  • Jieru Hou

摘要

Microplastics (MPs), as emerging pollutants in the environment, have attracted widespread attention due to the potential threats they pose to ecosystems through their accumulation in soil. With the promotion and application of biodegradable plastics, the pollution problem of their derived biodegradable MPs has become increasingly prominent. However, current detection of MPs in soil focuses on traditional non-biodegradable MPs. To address this issue, this study presents a laboratory-based proof-of-concept for the classification of MPs using Short-Wave Infrared (SWIR) hyperspectral imaging (HSI) coupled with machine learning algorithms including Support Vector Machine (SVM), Artificial Neural Networks (ANN), Convolutional Neural Network (CNN). Systematic evaluation confirmed the feasibility of distinguishing biodegradable (PLA, PBAT, PHA) and conventional (PA, PP, PVC) MPs across three distinct soil matrices. Under controlled conditions, the CNN demonstrated superior performance, achieving a peak overall accuracy of 95.98%. Compared with traditional destructive thermal analysis, this method validates the intrinsic spectral separability of biodegradable polymers in soil mixtures (SM). This study establishes a theoretical foundation for the future development of rapid, non-destructive screening technologies for soil MPs pollution.

Graphical abstract