Extracting process-related information from pharmaceutical dosage forms using near infrared microscopy

Near-infrared (NIR) microscopy is a powerful tool for the analysis of pharmaceutical samples. It provides the opportunity to chemically understand the sample, and also to gain knowledge on how ingredients arrange when mixed together. Along with spectral information of samples, which is beneficial for identification of crystal form, it is also possible to determine the spatial distribution and cluster size of ingredients in a mixture of materials. This paper describes two different methods for extracting process-related information from NIR microscopy data cubes. Principal component analysis with data reconstruction was used a means to extract pare ingredient spectra from a pharmaceutical tablet at the micro-scale. Partial least squares analysis with particle statistic investigations was utilised to compare multiple samples on a numerical level-despite data originating from an image. Both methods allow visual representations of the pharmaceutical dosage form, chemical images, describing the distribution of the different chemical components within the matrix. Either using the chemical images alone or after extraction of size data on each component, it was possible to correlate the NIR microscopy result to the final performance of the pharmaceutical product. (C) 2003 Elsevier B.V. All rights reserved.