Model-based biological Raman spectral imaging

Abstract

Raman spectral imaging is a powerful tool for determining chemical information in a biological specimen. The challenge is to condense the large amount of spectral information into an easily visualized form with high information content. Researchers have applied a range of techniques, from peak-height ratios to sophisticated models, to produce interpretable Raman images. The purpose of this article is to review some of the more common imaging approaches, in particular principal components analysis, multivariate curve resolution, and Euclidean distance, as well as to present a new technique, morphological modeling. How to best extract meaningful chemical information using each imaging approach will be discussed and examples of images produced with each will be shown. J. Cell. Biochem. Suppl. 39: 125–137, 2002.