High-throughput Isotropic Light-sheet Microscopy

Light-sheet fluorescence microscopy(LSFM)has recently become an emerging technique for three-dimensional(3D)imaging.As compared with conventional epi-illumination,its selective plane excitation greatly reduces the photo-bleaching and photo-toxicity.Therefore,LSFM has been widely used in modern biomedical research.However,most of current LSFM methods,such as Gaussian and Bessel light sheets,suffer from several problems: the trade-off between resolution and field of view(FOV)and inhomogeneity in conventional Gaussian beam;the insufficient image quality caused by the side lobes of Bessel beam,leading to the 3D anisotropy resolution and limited imaging throughput.This dissertation mainly focuses on the method development for achieving 3D isotropic imaging over a large FOV of large samples at extremely high imaging speed.It mainly includes following contents:Firstly,based on conventional static Gaussian light-sheet,we propose two kinds of axially scanned light-sheet microscopies(ASLM),which contain a voice coil motor(VCM)and a customized spinning-disk(SD)lens,respectively.VCM-and SD-ASLM scan thin light-sheets with short confocal range in its propagation direction periodically and synchronize the moving sheet with a two-dimensional(2D)array of active pixels of scientific complementary metal oxide semiconductor(s CMOS)camera to effectively suppress the out-of-focus fluorescence.The VCM-ASLM generates a uniform light sheet over a larger FOV.Combined with dual-stage3 D super-resolution network(DSR),the DSR-SD-ASLM achieves 4-times resolution enhancement in each dimension and up to 64-times throughput enhancement while reaching the frame rate limit of the state of art of s CMOS camera,which breaks the limit of imaging throughput with nearly 2-order-of-magnitude.Taken together,we demonstrate seconds-timescale,stitching-free 3D imaging of entire mouse brain at isotropic single-cell resolution(1.46-μm voxel)under a 1.1× low magnification.The effective optical throughput is further improved to ～30 Gigapixels per second.Cell body counting,segmentation of neurons and brain areas can be applied based on the rapidly reconstructed whole-brain data.Secondly,a digital scanned laser light-sheet microscopy(DSLM)setup on a conventional inverted wide field microscope is designed and combined with a deep-learning technique.DSLM enables a more uniform and controllable Gaussion light-sheet illumination and the high throughput of microfluidic chip control together make the 3D imaging of living animal possible.Restored by denoising and isotropic-resolving network,the signal-to-noise ratio(SNR)can be increased by ～3 times,enabling 3D isotropic imaging(1.63-μm voxel)of freelymoving drosophila larva in microfluidic chip at a volumetric rate up to 20.We further demonstrate the quantitative analysis of the neural activities and their correlations with the forward and backward locomotion of the freely moving worm.At last,an open-top,non-diffraction light-sheet microscopy is developed by a light droplet illumination,whose conjugated plane is focused on a double-ring phase mask.The static non-diffraction light sheet finally achieves near 3D isotropic resolution(3.3-μm voxel)under a wide FOV because the side lobes of sheet are suppressed by interfering Bessel beams of specific k-vectors,which greatly reduces the degration of image quality,photo-bleaching and photo-toxicity.At the same time,the oblique open-top design can largely expand the scale of biological sample when imaging,which remarkably minimizes the time in sample preparing and changing operation.The performance of our system is validated by fluorescent beads,tumor cells and brain slice in cover glass,microfluidic chip and micro-chamber.Finally,the large-scale high-throughput 3D imaging of tumor cells and other biological samples in 96-well plate is accomplished within 5 hours.Overall,above three light-sheet microscopies can be applied to large cleared tissue,living model animal and large-scale specimen respectively,which solve the problems of 3D anisotropy imaging and limited imaging throughput.The techniques proposed in this dissertation can provide support for the study of more biological problems in the future.