ABSTRACT
The popularity of spectral images in many areas of analysis has greatly increased during the last decade due to the development of charge-coupled device (CCD) and infrared sensitive cameras. Large amounts of spatial information can be obtained in short periods of time. The general goal in analytical chemistry is to convert spectral images into chemical images, which show the spatial locations of various chemical components. Self-modeling multivariate curve resolution methods can be used to extract pure component spectra from the mixture spectra in images and produce chemical images. However, there is a difficulty in processing infrared spectral images due to large pixel-to-pixel baseline variations. Herein, a method for minimizing baseline interferences using fast Fourier transform (FFT) filtering in both the spectral and spatial domains is discussed. The methodology is demonstrated on a microscopic sample of butter contaminated with non-pathogenic E. coli and on a cross-sectional sample of rabbit aorta containing plaque. The processing to reduce baseline effects improved the spatial resolution without compromising the spectral resolution.
