Separation of CARS image contributions with a Gaussian mixture model
in: Journal of the Optical Society of America A-Optics Image Science and Vision (2010)
Coherent anti-Stokes Raman scattering (CARS) gained a lot of importance in chemical imaging. This is due to the fast image acquisition time, the high spatial resolution, the non-invasiveness, and the molecular sensitivity of this method. By using the single-line CARS in contrast to the multiplex CARS, different signal contributions stemming from resonant and non-resonant light–matter interactions are indistinguishable. Here a numerical method is presented in order to extract more information from univariate CARS images: vibrational composition, morphological information, and contributions from index-of-refraction steps can be separated from single-line CARS images. The image processing algorithm is based on the physical properties of CARS process as reflected in the shape of the intensity histogram of univariate CARS images. Because of this the comparability of individual CARS images recorded with different experimental parameters is achieved. The latter is important for a quantitative evaluation of CARS images.