Data Acquisition And Processing Methods For Super-resolution Localization Microscopy With Large Field Of View

Super-resolution localization microscopy(SRLM)is able to visualize nanoscale fine structures beyond diffraction limit,and thus becomes a promising optical imaging tool for studying the ultrastructure,distribution,and function of cellular molecules,organelles and cells.The development of SRLM with large field-of-view(FOV)is made possible by combing SRLM with new low-light detectors with large pixel array,and this combination brings new chances to applications like high-throughput imaging,high-content screening.This paper aims to push the maturation of SRLM with large FOV by developing a series of data acquisition and processing tools,including SRLM with back-illuminated sCMOS(Scientific Complementary Metal Oxide Semiconductor)detector,on-line sparse molecule localization algorithm,and on-line ending method for image acquisition.(1)SRLM with back-illuminated sCMOS detector.Using a photon transfer curve measurement system and a direct detector evaluation system,this paper estimated the imaging performance of a back-illuminated sCMOS detector on three levels(single pixel,single molecule and super-resolution imaging).The experiments based on uniform incident light,fluorescent beads,and cell sample with two systems found that: compared with two typical low-light detectors,this back-illuminated sCMOS detector presents advantageous imaging performance on these three levels.This paper also provided a pilot study on applying this back-illuminated sCMOS detector in SRLM with large FOV.(2)On-line sparse molecule localization method.This paper analyzed the signal distribution model of sparse molecules and developed an on-line sparse molecule localization algorithm called FFTLocalization(Fast Fourier Transform Localization).FFTLocalization simplifies the estimation of the precise localization of sparse molecules and avoids massive iterations in traditional fitting-based methods.This paper further accelerated the data analysis speed of FFTLocalization using GPU(Graphics Processing Unit)parallel computation.Simulated and experimental data show that FFTLolcaization can realize on-line sparse molecule localization for SRLM with large FOV.(3)On-line ending method for image acquisition.Since the ultimate goal of SRLM is to image fine structures instead of obtaining a full location list of molecules,this paper proposed structure resolving index(SRI)concept based on the correlation of localizations and provided a SRI-based method for on-line ending image acquisition of SRLM.SRI can theoretically reduce excess data volume brought by redundant localizations from the same molecules with multiblinking property or adjacent molecules that are too close to be resolved.Simulated and experimental data show that SRI can be used as an effective ending method for image acquisition and reduces 26.0% ~ 54.9% raw data volume while assuring the resolving power of the final super-resolution image.In summary,this paper developed a series of data acquisition and processing tools for SRLM with large FOV,and further improved the capability of usability of SRLM.This study will enhance the application of SRLM in high-throughput imaging,high-content screening,etc.