Multivariate analysis of hyperspectral stimulated Raman scattering microscopy images

Abstract

AbstractCoherent Raman scattering microscopy is an attractive new technology for label‐free imaging. Both coherent anti‐Stokes Raman scattering (CARS) and stimulated Raman scattering (SRS) microscopy offer the possibility to record hyperspectral imaging data. While for the analysis of spontaneous Raman microscopy data multivariate methods are nowadays routinely employed, until to date most of the coherent Raman imaging data are interpreted using univariate data analysis. In this work, we report a quantitative comparison of the performance of different multivariate methods used for the analysis of hyperspectral SRS data from different model samples. Our data show for all samples that multivariate methods outperform univariate analysis. Using metrics to quantify method performance, we find that of the methods tested, multivariate curve resolution (MCR) gives the best results. We show that the combination with a selection of essential components based on first‐order autocorrelation, gives a simple workflow for the MCR‐based analysis of hyperspectral coherent Raman imaging data.