Picosecond spectral coherent anti-Stokes Raman scattering imaging with principal component analysis of meibomian glands

Abstract

Abstract. The lipid distribution in the mouse meibomian gland was examined with picosecond spectral antiStokes Raman scattering (CARS) imaging. Spectral CARS data sets were generated by imaging specific localized regions of the gland within tissue sections at consecutive Raman shifts in the CH 2 stretching vibrational range. Spectral differences between the location specific CARS spectra obtained in the lipid-rich regions of the acinus and the central duct were observed, which were confirmed with a Raman microspectroscopic analysis, and attributed to meibum lipid modifications within the gland. A principal component analysis of the spectral data set reveals changes in the CARS spectrum when transitioning from the acini to the central duct. These results demonstrate the utility of picosecond spectral CARS imaging combined with multivariate analysis for assessing differences in the distribution and composition of lipids in tissues. C 2011 Society of Photo-Optical Instrumentation Engineers (SPIE).

Identification of cholesterol crystals in plaques of atherosclerotic mice using hyperspectral CARS imaging

“…In this study, we utilized PCA to decompose the hyperspectral data in terms of a manageable set of principal components that best describes the spectral variations in the original data. Detailed descriptions on the implementation of PCA for image analysis can be found in several reviews ( 46,47 ), and its applications for hyperspectral CARS imaging has been recently demonstrated ( 32,36 ).…”

“…Because of the undistorted reproduction of the original CARS spectra, the spectral comparison can be performed directly on the CARS dataset without the need for converting the data to linear Raman spectra with phase retrieval algorithms ( 32 ). To enhance the contrast among different constituents in the image, the RGB scoremap is generated based on new axes, Z p , where the spectra from the same colored pixels are averaged and normalized to the peak value.…”

“…In this study, we demonstrate that, with PCA, major constituents can be discriminated based on the correlated spectral variations within the dataset itself, thereby providing an unsupervised and effi cient assessment of chemical composition. On this note, the application of PCA for dimensionality reduction, in lieu of other nonlinear techniques, has been recently highlighted ( 34 ), and its utility for spectral decomposition and image reconstruction of hyperspectral CARS datasets has been illustrated in several studies ( 35,36 ), including our own work ( 32 ).…”

“…Supplementing coherent Raman images with spectral information enables the visualization of several lipid classes selectively ( 31 ). Our method is based on hyperspectral CARS imaging, which was recently successfully used to unveil the compositional lipid variations in gland tissues ( 32 ). Here we use hyperspectral CARS in combination with principal component analysis (PCA) to extract spectrally signifi cant variations of molecular distributions throughout the lesion.…”

Accumulating advantages, reducing limitations: Multimodal nonlinear imaging in biomedical sciences – The synergy of multiple contrast mechanisms

“…Hence, a variety of techniques for multispectral data analysis have been applied to spectral CRS and fluorescence data. Principal component analysis (PCA) known from linear Raman spectroscopy has been applied to spectral CARS data to visualize differences in the lipid distribution in meibomian glands [46], and to identify cholesterol crystals in both SRS and CARS spectral data [37,47]. Even more sophisticated methods are required to extract the Raman spectrum from multiplex CARS data, e.g., the maximum entropy method or the Kramers Kronig relations [48][49][50].…”