Related Experiment Video
Updated: Jun 28, 2026
08:58
Fully Automated Centrifugal Microfluidic Device for Ultrasensitive Protein Detection from Whole Blood
Published on: April 16, 2016
11.1K
RaSizeNet: An interpretable multimodal deep learning model integrating SERS spectra and inertial microfluidic size
Guanwen Su1, Wenrui Yang1, Guozhao Liu1
1School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, China.
Journal of Hazardous Materials
|February 18, 2026
Summary
A new deep learning model, RaSizeNet, accurately identifies bacteria using combined spectroscopy and microfluidics. This breakthrough enhances pathogen detection in complex samples for improved public health.
Area of Science:
- Biotechnology
- Analytical Chemistry
- Artificial Intelligence
Background:
- Accurate pathogen identification is crucial for public health and infectious disease control.
- Label-free, high-throughput, and sensitive bacterial detection in complex samples remains a challenge.
Purpose of the Study:
- To develop an interpretable multimodal deep learning model for high-precision identification of bacterial populations.
- To integrate surface-enhanced Raman spectroscopy (SERS) and inertial microfluidic size distribution for bacterial detection.
Main Methods:
- Synthesized Fe@PCN-Au hybrid particles for SERS activity and magnetic responsiveness.
- Embedded particles into a dual-zone microfluidic chip for bacterial enrichment, separation, and signal enhancement.
- Developed a multimodal deep learning model (RaSizeNet) with a size-guided cross-modal attention mechanism using SERS spectra and microfluidic data.
Main Results:
- RaSizeNet achieved an average accuracy of 96.22% in complex samples.
- Outperformed unimodal (SERSNet, 82.23%) and conventional multimodal (MultiNet, 92.81%) models.
- Demonstrated superior adaptability and stability under various conditions (concentrations, flow rates, media).
Conclusions:
- The developed RaSizeNet model offers a promising framework for multimodal pathogen detection.
- Integration of microfluidic-SERS sensing with AI provides theoretical and technical support for real-time detection.
- This approach advances high-precision identification of complex bacterial populations.
