Parallel micro-Raman spectroscopy of multiple cells in a single acquisition using hierarchical sparsity.

Parallel micro-Raman spectroscopy can significantly expand the analytical capacity of single biological cells. By positioning the Raman spectra of multiple trapped cells on a detector array along the grating dispersion direction, the throughput of single-cell analysis can be improved by orders of magnitude. However, accurate retrieval of the individual spectra from the superimposed spectrum in a single acquisition presents great challenges. In this work, we developed a hierarchical sparsity method under a compressive sensing framework. The method combined a group-selection strategy with in-group sparsity for spectral reconstruction. The performances of the developed method were demonstrated with both simulated and experimental data, and the Raman spectra of the individual trapped cells were retrieved with both high accuracy and low noises; especially, with a group-selection mechanism, the developed method successfully avoided wrong selection of the eigenspectra for spectral reconstruction. The technique is expected to find wide applications in simultaneous monitoring of long biological processes of multiple cells by Raman spectroscopy.