Applications of principal components analyses to multidimensional FTIR microscopy data

The acquisition of multidimensional data, both multispatial and multispectral data, is now routinely accomplished using an FT-IR microscope equipped with a motorized stage. FT-IR microscope mapping generates multi-megabyte data sets with several thousand data points per spectrum, where each spectrum is a pixel in an image. Methods to reduce each infrared spectrum to a single intensity must be used to produce a pseudo 3-D image. Multivariate statistical methods such as principle components analysis (PCA) utilize the multiwavelength information acquired at each spatial location to generate this image containing new chemical information. PCA generates the image by determining independent sources of spectral variance without any knowledge of chemical composition. Since PCA can be applied as a full spectrum method, there is no requirement for any previous knowledge about the data set as is the case for other methods of data reduction.