基于主成分分析和多元曲线分辨的蓝细菌流式荧光光谱分析方法

利用流式细胞术对细胞进行多色荧光分析时，往往获得的是由多种组分荧光光谱混合的多元荧光光谱。在对蓝细菌进行光谱流式检测时，所测得的荧光光谱同时包含了多种未知荧光光谱，且存在严重的光谱混叠。为了获得蓝细菌中的主要组分光谱及其浓度，提出主成分分析和多元曲线分辨相结合的方法，对蓝细菌的流式荧光光谱进行处理。该方法通过主成分分析获得蓝细菌的主要纯组分数量，然后利用渐进因子分析寻找各组分的起始点和终止点，并估计纯组分的初始光谱，最后利用交替最小二乘结合其纯组分光谱的单峰性和非负性，对初始估计的纯组分光谱进行迭代修正，从而得到纯组分光谱及其组分浓度。仿真和实验结果表明，该方法能够准确地估计混合光谱中纯组分的个数并对其谱峰进行拟合，进而准确地估计各个组分的浓度。该方法不但适用于蓝细菌的光谱分析，还可用于其他多元混合光谱体系的解析。

Analytical Method of Cyanobacteria Flow Fluorescence Spectrum Based on Principal Component Analysis and Multivariate Curve Resolution

When flow cytometry is used to analyze the polychromatic fluorescence of cells, multiple fluorescence spectra were often obtained, mixed with multicomponent fluorescence spectra. In this paper, the fluorescence spectra of cyanobacteria including many unknown fluorescence spectra were detected by flow cytometer with serious spectral overlap. In order to extract the main components and their concentrations from cyanobacteria spectra, a method of principal component analysis combined with multivariate curve resolution was used to process the fluorescence spectra of cyanobacteria. At first, the number of main components of cyanobacteria was given by principal component analysis, and then Evolving Factor Analysis was adopted to find the starting and end position of each component and to estimate the initial spectrum of pure components, finally Alternating Least Square combined with the pure components spectral unimodality and non-negativity was used to correct the initial estimation of pure components and concentrations. In the simulation and experiment, it was proved that the method could accurately estimate the number of pure components in the mixed spectra and fit the spectral peaks, and then accurately estimate the concentration of each component. This method can not only be applied in the spectral analysis of cyanobacteria, but also used for other multiple spectral mixture analysis.