Study of component distribution in pharmaceutical binary powder mixtures by near infrared chemical imaging

• Published in: Journal of spectral imaging Vol. 3; p. a2
• Main Authors: Bautista, Manel, Cruz, Jordi, Blanco, Marcelo
• Format: Journal Article
• Language: English
• Published: IM Publications Open 12-12-2012
• Subjects: concentration maps; correlation coefficient; homogeneity; NIR chemical imaging; pharmaceutical samples
• ISSN: 2040-4565; 2040-4565
• DOI: 10.1255/jsi.2012.a2

Near infrared chemical imaging (NIR-CI) has recently emerged as an effective technique for extracting spatial
information from pharmaceutical products in an expeditious, non-destructive and non-invasive manner. These features
have turned it into a useful tool for controlling various steps in drug production processes. Imaging techniques provide a
vast amount of both spatial and spectral information that can be acquired in a very short time. Such a huge amount of data
requires the use of efficient and fast methods to extract the relevant information. Chemometric methods have proved
especially useful for this purpose. In this study, we assessed the usefulness of the correlation coefficient (CC) between
the spectra of samples, the pure spectra of the active pharmaceutical ingredient (API) and we assessed the excipients to
check for correct ingredient distribution in pharmaceutical binary preparations blended in the laboratory. Visual information
about pharmaceutical component distribution can be obtained by using the CC. The performance of this model construction
methodology for binary samples was compared with other various common multivariate methods including partial least
squares, multivariate curve resolution and classical least squares. Based on the results, correlation coefficients are a
powerful tool for the rapid assessment of correct component distribution and for quantitative analysis of pharmaceutical
binary formulations. For samples of three or more components it has been shown that if the objective is only to determine
uniformity of blending, then the CC image map is very good for this, and easy and fast to compute.