This action addresses the problem of imaging deep inside scattering samples using three-photon (3P) excited fluorescence signals. In biomedical imaging, multi-photon microscopy serves as a tool to investigate structures and processes inside living tissues non-invasively. In this context, 3P microscopy got a lot of traction recently, due to its superior performance at large depth. However, light scattering within the tissue still poses the main limitation to successfully image deep tissue layers. While seemingly random, these scattering processes are coherent and can be reversed via wavefront shaping of the excitation light. While some initial studies showed that this can be done for low-order large-scale distortion of the wavefront, at the time this action was conceived, scattering correcting wavefront shaping for 3P microscopy was out of reach.
Particularly in neuroscience, the enhanced performance at depth lead to a fast adaptation of 3P microscopy. In order to image deep layers of the brain of awake mice, for example, without removing or compromising superficial layers one has to find optical solutions. Wavefront shaping can provide this solution to push the boundaries of in-vivo imaging deeper into the brain and thereby open new ways to understand the intricate functioning of the mammalian brain.
The objective of this action was to develop tools for scattering correcting wavefront shaping in 3P microscopy. In proof-of-principle experiments, we demonstrated that wavefront optimisation based on the total 3P fluorescence signal can be achieved with a simple continuous optimisation algorithm, even when the initial point-spread-function is strongly scattered. This enables imaging in situations where no image could be formed before. Another conclusion of the action is that the higher non-linearity of the 3P process leads to larger signal enhancements during wavefront optimisation but is not required for converging to a focus within volumetric fluorescent samples. This refutes claims made in the literature that 2P fluorescent feedback is insufficient to focus inside homogeneously dyed volumes. Finally, the action concludes that there is a largely untapped potential to employ computational techniques in multi-photon microscopy. In two separate techniques we used random illuminations paired with a reconstruction algorithm to form an image purely computationally.
