Self-Modelling Curve Resolution of near Infrared Imaging Data

Abstract

ISSn: 0967-0335 © IM publications llp 2008 doi: 10.1255/jnirs.773 all rights reserved chemical imaging refers to a spectroscopic technique which allows the identification of the chemicals in a sample as well as their distributions. a moving single-point detector or a fixed array detector acquires thousands of spectra spatially located. three-dimensional data are generated and may be viewed as a stack of images where two dimensions depict the spatial information and one dimension features the spectral information of the analysed sample.1 Imaging techniques have found several applications in the pharmaceutical industry such as monitoring of blend homogeneity2 and influence of process parameters on the physical properties of tablets.3,4 It has also been applied to process troubleshooting,5 quality control6 and is a potentially useful apparatus for process analytical technology (pat).7 pat is an initiative launched by the food and drug administration which promotes the development of methods to monitor and fully understand pharmaceutical processes. chemical imaging techniques generate a huge amount of data, up to 80,000 near infrared (nIr) spectra, which must be processed to extract the relevant information such as chemical signatures and distribution maps of the compounds. the simplest way to find out information on distribution maps is to consider the absorbance maps at a specific wavelength of one compound. nevertheless, this demands prior knowledge of the absorption bands, which might not be known in all cases. Moreover, when the number of compounds increases, Self-modelling curve resolution of near infrared imaging data