Rapid Analysis of Raman Image Data Using Two-Way Multivariate Curve Resolution

The application of standard two-way curve resolution methods is reported for analysis of three-way Raman image data. Two current curve resolution methods are described: principal factor multivariate curve resolution (PF-MCR), which uses principal factor analysis (PFA) combined with varimax rotation and alternating least-squares optimization (ALS), and orthogonal projection multivariate curve resolution (OP-MCR), which uses a Gram–Schmidt modified orthogonal projection approach (OPA) followed by ALS. The OP-MCR technique is shown to be an extremely rapid method of analysis producing results equivalent to those of PF-MCR in one-third to one-fourth the time. The results from MCR analysis using either method provide the number of chemical species present in the sample, the spectrum of each species for identification, and the concentration image for each species. The additional benefit of image noise reduction also results from the MCR techniques. A brief description of the theory is presented followed by analysis and comparison of results for two real Raman image data. A discussion is given addressing the rapid analysis aspects of OP-MCR and the relative merits and drawbacks of the technique in comparison to PF-MCR. The use of data subsampling is also discussed as a way of decreasing analysis time without loss in accuracy or performance.