The main objective of multi-fluorophoric sample spectroscopy is to find the relative concentrations of individual components. If the spectra of fluorophores have broad complex shapes and significantly overlap, direct analysis is not possible. For various reasons, regression to model spectra attributed to known substances might not be easy either. Here, we describe a method of estimating the spectra of pure fluorophores in complex samples of unknown nature. This method is based on principal component analysis. Principal components themselves cannot be determined uniquely, so the method includes subsequent finding of suitable linear combinations of those principal components. Two physically reasonable assumptions were used to estimate individual fluorophore spectra: their intensities should always be non-negative and with the smallest possible spectral widths. The method’s features were tested by numerical simulations using the random generated spectra of three component mixtures, experiments with optical phantom, and analysis of previously collected spectra from real tissue samples. The accuracy of the method depends on the spectral features of the samples. If spectra intersect pairwise, components could be obtained precisely. In other cases reconstructed spectra closely match the original ones, allowing them to be attributed to possible sample components. This approach makes this method convenient for various applied diagnostic tasks. It provides not only quantitative data for sample comparison, as straight principal component analysis does, but also makes data representation more demonstrative, allowing for the creation of qualitative conclusions. The method provides a unique solution dependent only on the shapes of the fluorophores’ spectra.
https://www.opotek.com/wp-content/uploads/2018/08/opotek-logo-v4-4-325x100-300x93.png 0 0 Katy https://www.opotek.com/wp-content/uploads/2018/08/opotek-logo-v4-4-325x100-300x93.png Katy2021-11-18 23:00:152021-11-18 23:00:16Decomposition of multi-component fluorescence spectra by narrow peak method based on principal component analysis