Reconstruction Of Super-resolution Microscopy With Structured Illumination

Fluorescence microscopy is a diffraction-limited system,in which the spectrum outside the cut-off frequency is lost during imaging,leading to a resolution limit in the acquired images.When the distance between two fluorophores is less than the resolution limit,two points will merge into one in the image.Structured illumination microscopy(SIM),which can overcome the diffraction limit and enhance lateral resolution to 100 nanometers,has been gradually developed in the past 20 years.However,SIM still suffers shortcomings of low time resolution and short-term time-lapse sequence.These disadvantages can be softened by methods such as reducing the excitation power or shortening the exposure time.However,these methods would reduce the signal-to-noise ratio of the raw images,leading to a potential incorrect estimation of parameters and reconstruction artifacts.This thesis proposes a reconstruction algorithm of the Hessian structured illumination microscopy(Hessian SIM),which can be used to reconstruct low signal-to-noise ratio raw images collected at low excitation power and short exposure time.Through the pre-averaging and phase-only method,the parameters of the SIM microscopy can be calculated robustly and accurately from these raw images.At the same time,the spatiotemporal Hessen penalty proposed in this thesis can effectively suppress noise-induced artifacts in the reconstructed image.Compared with traditional SIM,Hessian SIM has fewer artifacts caused by error parameters and noise in reconstructed images,a higher temporal resolution up to 188 Hz,and a longer consecutive imaging frames up to 6,800 frames.All these advantages make the Hessian SIM microscopy suitable for super-resolution imaging of living cells.This thesis further analyses the causes of various artifacts in the SIM reconstructed images,summarizes corresponding solutions,and proposes an optimized SIM reconstruction protocol that integrates these solutions into the traditional reconstruction SIM protocol.Reconstructed images from different raw data prove that the optimized reconstruction protocol always has high quality.To observe the three-dimensional structure of the cell membrane,which is located hundreds of nanometers above the glass slide,this thesis proposes a new kind of reconstruction algorithm based on the deconvolution of multi-angle total internal reflection microscopy(DMA-TIRF).Considering the point spread function(PSF)of TIRF microscopy,this algorithm can reconstruct a high-resolution three-dimensional TIRF image,reaching a lateral resolution of 200 nanometers and an axial resolution of 50 nanometers.To further improve the lateral resolution of the DMA-TIRF microscopy,a multi-angle total internal reflection fluorescence microscopy with structured illumination(MASIM)is proposed in this thesis.With the structured illumination in the multi-angle TIRF microscopy,MASIM microscopy gathers the raw images which contain the signals' spectrum outside the cut-off frequency.Then,from these raw images,a three-dimensional super-resolution image,which achieves 100-nanometer lateral resolution and50-nanometer axial resolution,can be reconstructed through the reconstruction algorithm proposed in this thesis.