Hannover 2010 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Downloads | Hilfe
MO: Fachverband Molekülphysik
MO 20: Experimental Techniques I
MO 20.6: Vortrag
Donnerstag, 11. März 2010, 12:15–12:30, F 142
Hyperspectral data processing for chemoselective MCARS microscopy using principal component analysis — •Christoph Pohling1,2, Tiago Buckup1,2, and Marcus Motzkus1,2 — 1Physikalisch-Chemisches Institut, Ruprecht-Karls-Universität, D-69120 Heidelberg, Germany — 2Physikalische Chemie, Philipps-Universität, D-35043 Marburg, Germany
Multiplex Coherent anti-Stokes Raman Scattering (MCARS) microscopy is a labelling free imaging technique, which has been steadily improved during the last decade [1]. Important for future application in medicine is the capability of providing chemoselective image contrast in case of biological samples. In this context, the Raman lineshape should be retrieved from the coherent CARS signal [2] and the unknown sample components must be labelled clearly. We have implemented a processing tool for MCARS microscopy that applies principal component analysis (PCA) to generate chemical contrast. Initially, the PCA calculates the eigenvectors from the MCARS hyperspectral data set. Later the image is recomposed automatically from the main eigenvectors received from PCA. The last step is realized either by matrix multiplication or by using an evolutionary fitting algorithm. We discuss the sensitivity regarding line width, line separation and sample concentration with simulated data. Furthermore, we show the effect of the extraction of the RAMAN line on the sensitivity of this approach and demonstrate its capability in biological samples. [1] von Vacano, B. et al., J. Raman Spectr., 7 (2007) 916. [2] Liu, Lee, Cicerone, J. Raman Spec., 40 (2009) 726.