Label-Free Lymphocyte Subtype and State Classification by Stimulated Raman Scattering Microscopy
En palabras de los autores
Accurate immune cell subtype and functional state profiling are essential for studying immune regulation, disease progression, and therapeutic response. Established single-cell analysis techniques rely heavily on either endpoint, destructive measurements or complex labeling that can potentially affect cell functions. Stimulated Raman scattering (SRS) microscopy offers quantitative, label-free imaging of cells and tissues by providing contrast based on subcellular concentrations of intrinsic biomolecules at high spatial and temporal resolution. Here, we demonstrate that Vision Transformer based hyperspectral SRS imaging classification can accurately and non-destructively classify various lymphocyte phenotypes, including lymphocyte lineage (T and B cells), T cell subset and differentiation state (CD4+ and CD8+ T cells, effector, memory, and naive for both), and functional states (naive T, activated T, exhausted T) using fluorescent markers for training. By leveraging attention-based spatial-spectral feature learning, the approach captures biomolecular patterns associated with immune cell identity and state while preserving cells for downstream or longitudinal analysis. This provides a potential platform for label-free immune profiling, live-cell analysis, and longitudinal studies of immune cell type and state transitions.
Apareció: miércoles, 18 de noviembre. bioRxiv. Preprint, todavía sin revisión por pares.