ABSTRACT
We report the development and implementation of an epi-detected spectral-focusing hyperspectral stimulated Raman scattering (SRS) imaging technique for label-free biomolecular subtyping of glioblastomas (GBMs). The hyperspectral SRS imaging technique developed generates SRS image stacks (from 2800 to 3020 cm-1 at 7 cm-1 intervals) within 30 s through controlling the time delay between the chirped pump and Stokes beams. SRS images at representative Raman shifts (e.g., 2845, 2885, and 2935 cm-1) delineate the biochemical variations and morphological differences between proneural and mesenchymal subtypes of GBMs. Multivariate curve resolution (MCR) analysis on hyperspectral SRS images enables the quantification of major biomolecule distributions in mesenchymal and proneural GBMs. Further principal component analysis (PCA) and linear discriminant analysis (LDA) together with leave-one SRS spectrum-out, cross-validation (LOOCV) yields a diagnostic sensitivity of 96.7% (29/30) and specificity of 88.9% (28/36) for differentiation between mesenchymal and proneural subtypes of GBMs. This study shows great potential of applying hyperspectral SRS imaging technique developed for rapid, label-free molecular subtyping of GBMs in neurosurgery.
