Related Experiment Video
Updated: Jun 12, 2026
09:10
Multimodal Analysis of Microplastics in Drinking Water using a Silicon Nanomembrane Analysis Pipeline
Published on: June 13, 2025
Improving the classification performance of microplastics by noise reduction and baseline correction of Raman spectra
Optics Express
|June 11, 2026
Summary
A new SE-ResUNet model significantly improves microplastic identification from Raman spectra. This AI approach enhances signal quality, achieving over 15-fold SNR improvement and boosting accuracy to 96.90% under challenging conditions.
Area of Science:
- Environmental Science
- Analytical Chemistry
- Spectroscopy
Background:
- Microplastics are pervasive global pollutants.
- Raman spectroscopy is key for microplastic detection.
- Conventional preprocessing methods for Raman spectra have limitations.
Purpose of the Study:
- To develop an advanced method for microplastic Raman spectral analysis.
- To overcome limitations of traditional preprocessing techniques.
- To enhance microplastic identification accuracy under non-ideal conditions.
Main Methods:
- Application of a ResUNet model with Squeeze-and-Excitation (SE) blocks.
- Denoising and baseline correction of microplastic Raman spectra.
- Comparison with traditional Wavelet Threshold Denoising and AirPLS methods.
Main Results:
- Achieved over a 15-fold improvement in signal-to-noise ratio.
- Significantly enhanced microplastic identification accuracy from 35.13% to 96.90% under stringent conditions.
- Outperformed traditional methods, which yielded 55.70% accuracy.
Conclusions:
- The SE-ResUNet model effectively enhances spectral quality for microplastic analysis.
- This AI-driven approach optimizes post-processing outcomes in Raman spectroscopy.
- The method shows high potential for accurate microplastic identification in environmental monitoring.
