Flexible paper-based AuNP sensor for rapid detection of diabenz (a,h)anthracene (DbA) and benzo(b)fluoranthene (BbF) in mussels coupled with deep learning algorithms.

This study aims to develop a flexible, paper-based gold nanoparticle (AuNP) surface-enhanced Raman scattering (SERS) sensor coupled with deep learning algorithms for the rapid detection of dibenz (a,h)anthracene (DbA) and benzo(b)fluoranthene (BbF) in mussels. The SERS sensor was fabricated on a hydrophobic paper substrate and incubated with varying concentrations of DbA and BbF in spiked mussel samples before SERS spectra collection. Deep learning algorithms, including Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, were employed to analyze the complex SERS spectral data. The SERS sensor demonstrated excellent sensitivity with limits of detection of 0.09 ng/mL for DbA and 0.10 ng/mL for BbF. LSTM outperformed traditional machine learning techniques, achieving R2p > 0.99 with minimal prediction errors. Real sample analysis showed excellent agreement with HPLC-FLD methods. These findings present a promising approach for rapid, sensitive, and reliable detection of PAHs in mussels, enhancing food safety monitoring and regulatory compliance. • Developed a flexible paper-based AuNP SERS sensor for DbA and BbF detection in seafood. • LSTM network outperformed other models in analyzing SERS spectra. • Achieved low limits of detection: 0.09 ng/mL for DbA and 0.10 ng/mL for BbF. • Real sample analysis showed excellent agreement with HPLC-FLD method. [ABSTRACT FROM AUTHOR]