Multivariate Statistical Methods for Biomarker Discovery from LAESI Mass Spectra of Small Cell Populations and Single Cells

摘要

We showed the utility of a multivariate statistical method for the selection of a small number of metabolites responsible for most of the spectral variance between cell subpopulations. The metabolites suggested by OPLS-DA are biomarker candidates that require further verification. We demonstrated the feasibility of using single cells as voxels for molecular imaging, thus following the distributions and gradients of biochemical species in a chemical image. Number of pixels in cell-by-cell imaging can be increased by automating the recognition of cells for ablation in the tissue and by preventing sample drying by introducing an environmental chamber for humidity control.