Noothalapati Hemanth, Iwasaki Keita, Yamamoto Tatsuyuki
Raman Center for Medical and Biological Applications, Shimane University.
Anal Sci. 2017;33(1):15-22. doi: 10.2116/analsci.33.15.
Biological specimens such as cells, tissues and biofluids (urine, blood) contain mixtures of many different biomolecules, all of which contribute to a Raman spectrum at any given point. The separation and identification of pure biochemical components remains one of the biggest challenges in Raman spectroscopy. Multivariate curve resolution, a matrix factorization method, is a powerful, yet flexible, method that can be used with constraints, such as non-negativity, to decompose a complex spectroscopic data matrix into a small number of physically meaningful pure spectral components along with their relative abundances. This paper reviews recent applications of multivariate curve resolution by alternating least squares analysis to Raman spectroscopic and imaging data obtained either in vivo or in vitro from biological and medical samples.
生物样本，如细胞、组织和生物流体（尿液、血液），包含许多不同生物分子的混合物，在任何给定的点上，所有这些生物分子都会对拉曼光谱产生贡献。纯生化成分的分离和鉴定仍然是拉曼光谱学中最大的挑战之一。多元曲线分辨是一种矩阵分解方法，是一种强大而灵活的方法，可用于带有诸如非负性等约束条件，将复杂的光谱数据矩阵分解为少量具有物理意义的纯光谱成分及其相对丰度。本文综述了通过交替最小二乘法进行多元曲线分辨在从生物和医学样本体内或体外获得的拉曼光谱和成像数据方面的最新应用。
