NK and NKT cells act as major effector subsets across multiple immune heterogeneous cell populations (HCPs), and their precise identification is essential for assessing product quality and predicting clinical efficacy. Here, we present a label-free identification strategy that couples Raman spectroscopy with a convolutional neural network (CNN) to achieve rapid and accurate classification of NKT and NK cells within diverse immune HCPs, including NK and CIK HCPs. In contrast to flow cytometry, this approach requires no antibody labeling, enables in situ measurement, and markedly shortens analysis time, while deep learning substantially improves classification speed and accuracy. By combining a visualization module with spectral analysis, we identified characteristic Raman peaks at 558 and 938 cm-1 in NKT cells, which serve as robust spectral biomarkers distinguishing NKT from NK cells. These findings are of considerable significance for immune cell identification and for advancing mechanistic studies. Overall, the study addresses limitations of traditional cell-typing technologies and provides a robust new pathway for quality monitoring of immune effector cell products, offering important implications for the development of immunotherapy.
Wan et al. (Tue,) studied this question.