Computational optimization of phase shaped CARS

We explore strategies for optimizing selectivity, specificity, and sensitivity in broadband CARS by precalculating pulse shapes using an evolutionary algorithm. We show the possibility of selective excitation of a single constituent in a test case of a mixture of five resonant compounds. The obtainable contrast ratio for a test case of PMMA in a mixture of five resonant compounds is predicted to be 2000:1, and is related the uniqueness of the complex vibrational response of the compound of interest compared to that of the surrounding molecules. Furthermore we investigate how the effects of homodyne mixing in the focal volume affect the obtainable contrast ratio and how noise affects the optimization. We also show preliminary results of experimental optimization of the CARS signal from PMMA microspheres, resulting in high contrast imaging, free of non-resonant background signal.