Smart Control of Supercontinuum Generation by Machine Learning Towards Multiphoton Microscopy Applications

Multi-photon excitation (MPE) microscopy has played a significant role in biological imaging techniques since it allows to observe living tissues with further depth and excellent sectioning effect. Modalities of MPE depend on the properties of both the selected fluorophore and the laser excitation (e.g., wavelength, pulse duration, peak power). However, versatile and tunable laser excitation for MPE still constitutes a technical challenge, limited by e.g. the narrow bandwidth of laser gain medium or by the achievable tunability of parametric conversion [1]. As an alternative, supercontinuum (SC) generation has been implemented for multi-photon imaging, as it can provide broadband excitation covering numerous fluorophore multi-photon absorption peaks and can enable multiplexed spectroscopy without wavelength scanning or filtering the excitation light.