In vivo stimulated Raman scattering flow cytometry

Conventional in vivo flow cytometry relies on fluorescent labeling for target cell detection, which often causes phototoxicity, photobleaching, and impaired cellular viability. Meanwhile, some label-free alternatives typically suffer from limited sensitivity, specificity, and applicability. Stimulated Raman scattering (SRS) flow cytometry offers high throughput and intrinsic chemical specificity without exogenous labels but has thus far been restricted to ex vivo applications. To address this limitation, we developed a spectral focusing approach combined with a resonant-galvanometer-driven rapid-scanning optical delay line (RSODL), enabling microsecond-scale spectral acquisition and high-speed line scanning across vascular cross-sections, thereby acquiring spectral information from flowing cells. Based on these spectra, reconstructed images of flowing cells were generated. We first applied instance segmentation to obtain accurate masks of individual cells and extracted their morphological features for preliminary cell type classification. Subsequently, using the spatial coordinates provided by these masks, we retrospectively retrieved the average spectrum of each cell from the reconstructed images. This dual-modality approach, combining morphological and spectral information, was used to achieve recognition of cells with high precision and specificity. In a live nude mouse tumor model, we successfully detected and distinguished red blood cells, white blood cells, and circulating prostate cancer cells and accurately quantified the temporal depletion kinetics of cancer cells. This study represents the first demonstration of in vivo SRS flow cytometry.