Deep learning based label-free small extracellular vesicles analyzer with light-sheet illumination differentiates normal and cancer liver cells

Abstract

Abstract Small extracellular vesicles (sEVs) are considered as potential markers for tumor detection and vehicles for tumor treatment. Here we develop a deep learning based nanoparticle analyzer that can measure label-free sEVs. Light sheet technology is adopted to illuminate single nanoparticles on chip that suppresses the background noise. A deep learning method for nanoscale particle tracking is demonstrated, which accurately obtains the particle size distribution of polystyrene beads as small as 41\xa0nm in diameter. Small extracellular vesicles from normal and cancerous liver cell linage cells, and from sorafenib drug treated cancerous cells, are analyzed label-freely. It is shown that the three types of sEVs can be well differentiated by their particle size distributions. Our deep learning based small extracellular vesicles analyzer (DeepEVAnalyzer) thus provides not only a new technique for sEV-like nanoparticle analysis, but also demonstrate the potential of using sEVs as label-free marker for cancer diagnosis.