Measurement precision enhancement of surface plasmon resonance based angular scanning detection using deep learning.

Area of Science:

• Optoelectronics
• Biosensing
• Artificial Intelligence

Background:

• Surface plasmon resonance (SPR) is a label-free sensing technique crucial for measuring refractive index changes.
• Accurate plasmonic angle determination is vital for SPR measurement precision, with current methods facing limitations.
• Existing plasmonic angle detection methods include polynomial fitting and image processing, with precision limits around 10⁻⁵ to 10⁻⁶ RIU.

Purpose of the Study:

• To develop a deep learning-based method for precise plasmonic angle detection in SPR measurements.
• To enhance the accuracy and reduce the complexity of SPR data analysis.
• To achieve higher measurement precision compared to conventional methods.

Main Methods:

• A convolutional neural network (CNN) architecture was designed for plasmonic angle localization.
• The CNN was trained using simulated SPR reflectance spectra with added shot and speckle noise for robustness.
• The method was validated experimentally by measuring refractive indices of air and nitrogen gas mixtures.

Main Results:

• The deep learning method achieved a measurement precision of 4.23 × 10⁻⁶ RIU.
• This precision surpasses conventional methods, including cubic polynomial curve fitting (7.03 × 10⁻⁶ RIU) and 2D contour fitting (5.59 × 10⁻⁶ RIU).
• The AI-based approach eliminates the need for sophisticated post-processing and optical instrumentation.

Conclusions:

• Deep learning offers a powerful and efficient approach for enhancing plasmonic angle detection in SPR.
• The proposed AI method significantly improves measurement precision for label-free sensing.
• This technique holds promise for advancing high-precision refractive index sensing applications.