| Due to the optical diffraction limit,traditional optical microscopy cannot distinguish biological structures smaller than 200 nm in size.Super-resolution imaging microscopy have been developed to overcome this optical diffraction limit,bringing the biomedical optical imaging to a brand-new era.Existing super-resolution imaging microscopy technologies developed in recent years can be generally classified into the following three types:Point-spread-function-engineering-based microscopy such as stimulated emission depletion microscopy(STED),frequency-domain-modulation-based microscopy such as(saturated)structured illumination microscopy(SIM/SSIM)and single-molecule-localization-based microscopy such as photo activation localization microscopy(PALM)or stochastic optical reconstruction microscopy(STORM).All these advanced techniques provide powerful tools for the optical imaging of tiny target beyond the optical diffraction limit in size,and thus attracting huge researching interests.As a typical single molecule localization super-resolution imaging technology,STORM can achieve a localization accuracy of 20 nm and ultra-high spatial resolution by regulating the switching state of fluorescent dye molecules.In addition to the requirement for an excellent photoblinking fluorescent probe,image reconstruction algorithms also play an important role in post-processing of high-resolution image data.The traditional algorithms for single-molecule localization microscopy require for sparsely labeling of optical probes on biological targets to avoid the severely overlapping of light spots of the fluorescent probes.Nevertheless,this will ask for thousands of photoblinking frames to reconstruct a single super-resolution image with rather low temporal resolution,which is not suitable for live cell STORM imaging.Multi-molecular localization algorithms,in which multi-Gaussian fitting is used to analyze overlapping data,can partially solve this problem.However,if the photoblinking density of fluorescent probes in each single frame overflows the upper limit with large number of overlapping spots,the multi-Gaussian fitting algorithm will not be able to provide ideal reconstructed images,probably reducing spatial resolution instead.Both methods mentioned above are difficult to carry out STORM imaging in live cells.To solve the above problems,researchers have developed a variety of localization algorithms for high-density data,but currently there are not many mature applications.This thesis has carried out study on the pre-processing algorithm for STORM super-resolution image reconstruction.A pre-processing algorithm based on the equivalent relationship between the integrated value of the modulated image and the virtual image,virtual single pixel imaging(VSPI)algorithm,is applied to STORM data processing.This algorithm can sparse the overlapped spot data with too high fluorescent molecule density and too small spacing(100 nm),which is later used in multi-molecule localization algorithms for super-resolution image reconstruction.Compare with the traditional pre-processing algorithm K-factor,the VSPI algorithm is more effective using the same reconstruction algorithms.Furthermore,the VSPI algorithm is used to pre-process the blinking data of a new STORM fluorescent probe.Combined with the existing super-resolution image reconstruction algorithm,the research of live cell mitochondrial super-resolution image reconstruction is carried out.Under the premise of keeping the spatial resolution unchanged,increasing the labeling density can reduce the number of image frames,so as to achieve the goal of improving the imaging temporal resolution,providing an effective tool for the research of living cell super-resolution imaging.The main work of this thesis includes:1.The VSPI algorithm is applied to the sparse pre-processing of high-density STORM super-resolution data.Firstly,the simulation data and experimental data on fluorescent beads are processed by VSPI to verify the sparse processing effect,and the result shows that its pre-processing ability is better than K-factor algorithm.Secondly,the VSPI algorithm is combined with the multi-molecular localization algorithm Thunder STORM to analyze simulation data and fixed cell imaging data,the super-resolution reconstruction results before and after data pre-processing using the VSPI algorithm are compared,and the differences among the super-resolution results of different densities are quantitatively studied,preparing for the further application of the VSPI algorithm to live cell data processing.2.The photophysical properties(including absorption and emission spectra),photoblinking properties(including saturated excitation power,average fluorescence on time,duty cycle,anti-photobleaching,etc.)and intercellular specific labeling of a newly developed fluorescent probe for STORM are analyzed.Then the probe is used to label mitochondria in live cells for STORM imaging,so as to collect live cell STORM experimental data,providing the live cell data support for VSPI pre-processing algorithm.3.The VSPI algorithm is used to pre-process the above-mentioned live cell experimental data,which are further used in multi-molecular localization algorithm FALCON to reconstruct super-resolution images,and the results are compared with those from reconstruction without pre-processing,so as to study the application of VSPI algorithm to live cell STORM imaging. |
PDF Full Text Request