Direct Comparison On The Performance Of Low-light Cameras For Super-resolution Localization Microscopy

Super-resolution fluorescence microscopy,the most important breakthrough in the field of modern optical microscopy in 21 st century,brings unprecedented opportunities for many fields in biomedical sciences.Super-resolution localization microscopy(SRLM)is a representative super-resolution fluorescence microscopy techinique.The key in SRLM is to distribute overlapped fluorescent molecules into thousands or even tens of thousands of image frames,so that the single molecules can be temporally isolated.A final superresolution image can be obtained through single molecule localization and reconstruction.The quality of single molecule localization is not only determined by the localization algorithm,but also by the quality of the single molecule images.It is important to select a proper low-light camera for ensuring the image quality.In this thesis,we quantitatively evaluated the imaging performances of different low-light cameras for SRLM.This work provides experimental evidences for helping the researchers choose and optimize an optimal low-light camera for SRLM.(1)Raw image quality control in SRLM: We controlled the raw image quality using two different approaches.For the experimental samples,we optimized the pretreatment processes for cell and tissue samples.And,we developed an improved immunofluorescent labeling method using a combination of nanobody and Alexa Flour 647.For the optical system,to maximize the data quality for direct camera comparison,we not only optimized the key experimental parameters of the system,but also employed singal synchronization and stage stabilization.In this way,we obtained an optial system that provides simutanous raw data acquisition,1:1 splitting ratio,and a stage drift of smaller than 10 nm.We also discuss the localization algorithms and the image quality evalution methodology.(2)Direct camera performance comparison via fluoresecent beads: Using the direct camera comparison systems discussed in Chapter 2 and a classic model sample-fluorescent beads,as well as several camera-specific localization algorithms,we directly compared the imaging performance of three representative low-light cameras,including Andor iXon 897 EMCCD,Hamamatus Flash4.0 V2 sCMOS and Hamamatsu Flash4.0 V3 sCMOS.For experiments using fixed fluorescene beads,we found out that both the camera-specific maximum likelihood estimator(csMLE)and the camera-independent maximum likelihood estimator(ciMLE)provided similar localization precision when the iXon 897 EMCCD was used.When the sCMOS cameras were used,csMLE provided slightly better localization precision than the ciMLE when the peak signal is betweent 50 to 250 photons per pixel.However,the different became smaller or even negligible when the signal was stronger.After considering the computational complexity of the csMLE,we concluded that the ciMLE is sufficient for most applications.When the ciMLE is used for single molecule localization,the relative localization precision is: Flash4.0 V3 > Flash4.0 V2 > iXon 897.On the other hand,we also compared the camera performance using moving fluorescence beads.Here ciMLE is used and the found bead trajectories were compared with the actual curves(from nano-stage).We found out that the localization precision performance is: Flash4.0 V3 > Flash4.0 V2 > iXon 897.Therefore,we concluded that,the representative sCMOS cameras are more suitable detector in SRLM than the representative EMCCD camera,because the sCMOS cameras provides not only better SRLM,but also high localization precision.(3)Direct camera performance comparison via biological samples: Chapter 4 used biological samples to further compares the imaging performance of three representative low-light cameras.From cell experiments,we found out both the iXon 897 and the Flash4.0 V3 provided similar microtubule stractures and FRC resolution.For tissue experiments,the Flash4.0 V2 provided even better imaging performance than the iXon 897 camera.Therefore,we proved that the sCMOS cameras are more suitable for SRLM than the EMCCD camera.