AI-assisted interferometric nanotweezers for label-free EV detection

Developed an AI-assisted interferometric nanotweezer platform for label-free trapping and analysis of extracellular vesicles (EVs) and nanoscale particles. The system integrates nanophotonic trapping with interferometric scattering (iSCAT/COBRI) imaging, enabling real-time detection of weak optical signals from sub-diffraction-limit objects.

A major focus of this work is the development of an AI-driven image-processing pipeline for reliable particle detection and characterization under low signal-to-noise conditions. Implemented deep learning–based segmentation (U-Net) to accurately identify nanoscale particles, combined with advanced contrast enhancement and background normalization techniques to improve visibility of interferometric signals.

Designed automated data-processing workflows to extract particle trajectories, intensity fluctuations, and diffusion dynamics. These features are further used to estimate particle size and physical properties through quantitative analysis of scattering contrast and Brownian motion.

This work demonstrates strong expertise in AI-based image segmentation, optical signal enhancement, scientific image processing, and data-driven analysis, bridging nanophotonics with modern machine learning techniques for label-free biosensing.