Single Molecule Identification And Localizationalgorithm Andanti-drifting Method For Super-resolution Fluorescence Microscopy

In1873, German physicist Ernst Abbe found that the resolution limit (Abbe limit)of the far-field optical microscopy is about250nm due to the diffraction of light. Sincethe1990s, many new optical imaging technologies and methods have been developed,and break the Abbe limit. Among these technologies and methods, single moleculelocalization microscopy (SMLM), such as photoactivation localization microscopy(PALM) and stochastic optical reconstruction microscopy (STORM), they are based ondetecting and localizing of single molecule fluorescence blinks, can archieve20to30nm lateral spatial resolution, and60to70nm axial spatial resolution.In SMLM, identification of single molecule fluorescence blink remains a greatchallenge due to the faintness of single molecule fluorescence and the existence ofbackground and noise. In this thesis, a robust algorithm about identification andlocalization of single molecule fluorescence blinks (SNSMIL) is presented, thealgorithm is based on intrinsic property of photons detecting (Possion noise), it improvethe identification accuracy greatly even with the cases of high or inhomogeneousbackground. The algorithm is implemented on graphic processing unit (GPU), makingreal time analysis possible by using a large number of threads.Other side, a SMLM measurement typically needs to record about1000-100000frames (typically with a speed of10-1000frames/s), that continues few minutes orlonger, the sample drift (typically1-10nm/s) has become a serious and unnegligibleproblem due to mechanical relaxations, temperature changes and other reasons, it canreduce the resolution of images, even distort images. In this thesis, a sub-nanometerprecision, low cost sample anti-drifting method is presented. with this method, bothfluorescence images and bright field images are recorded, then minimize the normalizedroot-mean-square error (NRMSE) between bright field images to get sample drifts, thismethod not only can be used for post drift correction, but also for real time driftcompensation.