An Integrated Microfluidic System for One-Stop Multiplexed Exosomal PD-L1 and MMP9 Automated Analysis with Deep Learning Model YOLO.

While immune escape and physical invasion are two key pathways in tumor development, traditional methods for analyzing their exosomal markers are often complex and face identification bias. Microfluidic technology offers significant advantages for non-invasive liquid biopsy, particularly in the analysis of tumor progression markers carried by exosomes. Here, we developed an integrated microfluidic system for the sensitive, accurate, totally on-chip exosome isolation and automatic quantification of tumor progression markers PD-L1 and MMP9. This platform leverages microfluidic design principles for efficient sample mixing and monodisperses microbeads for precise analysis, allowing for complete processing within 40 min. The system's high efficiency and precision are further enhanced by a lightweight YOLOv5-based positional migration strategy that automates fluorescence quantification. Validation using four different cell lines demonstrated distinct exosomal protein signatures with a low detection limit of 12.58 particles/μL. This innovative microfluidic chip provides a sensitive and easy-to-handle tool for exosomal marker analysis, holding great potential for cancer identification and personalized therapy guidance in the era of point-of-care testing (POCT).