Label-Free Single-Cell Activity-Based Cell Type Classification Using Optical Resonant Waveguide Grating Biosensor Data

Béres, Bálint and Horváth, Róbert (2025) Label-Free Single-Cell Activity-Based Cell Type Classification Using Optical Resonant Waveguide Grating Biosensor Data. In: 16th IEEE International Conference on Cognitive Infocommunications CogInfoCom 2025. Proceedings - International Conference on Cognitive Infocommunications, CogInfoCom . IEEE, Piscataway (NJ), pp. 137-142. ISBN 9798350356922; 9798350356915; 9798350356939

Text Label-Free_Single-Cell_Activity.pdf - Published Version Restricted to Registered users only Download (4MB) | Request a copy

Abstract

Accurate, label-free identification and classification of single cells based on functional properties is becoming increasingly vital in areas such as immunotherapy and diagnostics. This study introduces a novel label-free method for high-precision single-cell analysis by integrating high-resolution phasecontrast microscopy with dynamic adhesion measurements from a resonant waveguide grating (RWG) biosensor. The workflow incorporates the Cellpose model for image-based segmentation, the Practical Maximum Clique (PMC) algorithm for automatic alignment, and four CNN-based models for cell classification. Validation across six cell types on two types of functionalized surfaces (Fibronectin and Noncoated) yielded an maximum classification accuracy of nearly 95%. This integrated approach preserves cellular function and enables direct assessment of biological activity, making it well-suited for automated diagnostics and downstream applications.

Actions (login required)

Edit Item