Decoupled illumination detection in light sheet microscopy for 4D observation of spermatozoa at high-resolutions

We present the use of wavefront coding (WFC) combined with machine learning in a light sheet fluorescence microscopy (LSFM) system. We visualize the 3D dynamics of sperm flagellar motion at an imaging speed up to 80 volumes per second, which is faster than twice volumetric video rate. By using the WFC technique we achieve to extend the depth of field of the collection objective with high numerical aperture (NA=1) from 2.6 μm to 50 μm, i. e., more than one order of magnitude. To improve the quality of the final images, we applied a machine learning-based algorithm to the acquired sperm raw images and to the point spread function (PSF) of the generated cubic phase masks previous to the deconvolution process.