Research On Denoising And Artifact Correction Methods For Super-Resolution Localization Images

Super-resolution localization microscopic imaging has broken through the optical diffraction limit and achieved nanoscale spatial resolution,making it an important tool in life science research.Researchers can observe fine structures such as organelles inside cells using super-resolution localization microscopes.However,life science research has set higher requirements for the spatial resolution of super-resolution localization microscopic imaging,which exceeds the processing capacity of existing methods,necessitating further improvements in super-resolution image spatial resolution.Background fluorescence removal and blinking artifact correction are studied in this paper by analyzing the key data processing steps in super resolution positioning images and focusing on the difficulties and deficiencies encountered by current methods:(1)Adaptive background noise removal method.Although super-resolution location data contains a large number of mixed noises,the traditional super-resolution denoising method can only remove one type of noise,and the parameters must be manually set.To address these issues,a multi-step adaptive denoising framework(MSDenoiser)for super-resolution location data is proposed,which is based on the properties of various types of background noise.The framework fully exploits the benefits of various denoising algorithms,adaptively learns the algorithm parameters,and gradually removes various background noises from super-resolution localization images.We demonstrated the effectiveness of the proposed denoising framework on simulations and various types of experimental data sets,demonstrating that MSDenoiser can filter non-specific polymer noise points that cannot be filtered using other methods.(2)Image denoising evaluation index for super-resolution localization microtubules.The super-resolution image of microtubule structure was reconstructed in this paper to generate the Ground Truth of experimental data on microtubule structure,and the evaluation index was established by extracting the image’s skeleton information.The evaluation index can be used to evaluate different levels of noise filtering,according to the results of simulation and experimental data analysis.Given the lack of a denoising evaluation index in current microtubule experimental data,using microtubule image skeleton information as the evaluation index is an exploration with some practical application value.(3)Blinking artifact correction method based on space-time information.Existing blinking artifact correction methods frequently require additional kinetic experiments with fluorescent molecules to determine the time threshold.Based on the super-resolution localzation table and its prior knowledge,a blinking artifact correction method based on spatio-temporal information is developed to solve this problem.Without any additional experiments,this method can directly obtain the time and space thresholds from the location table.The results of simulation and experimental data analysis show that this method can generate the super-resolution localization image without blinking artifacts and bring it closer to the sample’s underlying structure.This method addresses the issue that the current method requires more experiments to determine the time and space thresholds and improves the spatial resolution of super-resolution positioning images.To summarize,two high precision data processing methods based on the inherent characteristics of positioning table data are developed in this paper,which improves the signal-to-noise ratio of super resolution images and provides higher quality positioning tables for subsequent post-processing steps.It is expected to broaden the research scope of super resolution positioning image research and promote its application in the analysis of fine structure and function.