APPLICATION OF LINEAR DISCRIMINANT ANALYSIS FOR CLASSIFICATION OF CYANOBACTERIA BY INTRINSIC FLUORESCENCE SPECTRA
A technique for cyanobacterial species classification by in-vivo single-cell fluorescence spectra was elaborated using special methods of mathematical statistics. The analysis of several hundred spectra for 20 different cyanobacterial strains, obtained by means of confocal laser scanning microscopy, was carried out. A specific procedure for input data processing and the order of extraction of such parameters as fluorescence spectra shape and the ratio of individual peaks were elaborated. Statistical methods were used for determination of a limited set of key parameters sufficient for classification. To solve the classification problem a standard multivariate discriminant analysis was used. As an example, the proposed classification algorithm was applied for the differentiation of three cyanobacterial strains, belonging to two genera.
Authors: T. R. Jhangirov, А. S. Perkov, A. A. Liss, N. Yu. Grigoryeva, L. V. Chistyakova
Direction: Informatics, management and Computer Technology
Keywords: Cyanobacteria, discriminant analysis, statistical methods for biology, fluorescence spectra
View full article