摘要
由于超分辨率显微镜需要额外的照明来换取更高的空间分辨率,导致了现在的超分辨率显微镜存在时间分辨率低和记录时间短两个问题。虽然结构光照明显微镜在获得超分辨率图像和所需要光子数方面有一个更好的平衡,相比其他超分辨率技术更有优势,但是传统的结构光照明显微镜的重构方式在信噪比比较低的时候会明显存在伪影。在本文中,我们利用生物样本在xyt三维空间的连续性作为先验知识,发展出了基于海森范数的正则化方式,并用来重构出伪影更少的结构光照明显微镜的超分辨率图像。与维纳反卷积这个经典的重构算法相比,海森反卷积只需要在成像时百分之十的光子数就可以达到同样的超分辨图像的重建效果。在激光强度为8—250W/cm2的情况下,海森结构光照明显微镜在活细胞中对快速分泌的囊泡、内质网等进行空间分辨率达到88nm,时间分辨率达到188Hz的超分辨率图像重建。全新的时间分辨率和空间分辨率使得我们可以在囊泡分泌过程中观测到新的中间态,观察到了囊泡靠近细胞膜的过程,以及之后的分泌小孔进一步扩大的过程这两个囊泡分泌过程中中间状态。使用基于海森正则项的结构光照明显微镜,我们可以在活细胞中观测到清晰的线粒体嵴。以及嵴结构在线粒体融合,线粒体分裂过程中的动态活动,甚至是单个没有发生融合或者分裂的线粒体内的嵴的融合的过程。
Current super-resolution（SR）microscopy is limited in both its temporal resolution and its recording length due to the extensive illumination required to obtain a sufficient amount of photons for achieving the desired spatial resolution.Although structured illumination microscopy（SIM）excels in attaining a high spatial resolution with a low amount of photons,the conventional post-processing approaches often produce artifacts during the reconstruction of raw images with a poor signal-to-noise ratio.Here,we developed a Hessian regularization approach that uses the continuity of biological structures in the xyt dimensions as a priori knowledge to constrain the Wiener deconvolution（Hessian-SIM）,which outperforms the current state-of-the-art algorithms in attaining artifact-minimized SR images with a photon dose less than10% of that required in conventional SIM.Under a modest illumination intensity of approximately 8—250 W/cm^2,Hessian-SIM enables 88 nm and 188 Hz live SR imaging of rapidly moving secretory vesicles and endoplasmic reticulum structures without motion artifacts.This approach allows the identification of novel vesicle fusion intermediates,as demonstrated by the first visualization of the collapsing of a vesicle to the plasma membrane followed by constriction of an enlarged pore.Using Hessian-SIM,we have visualized dynamics of mitochondrial cristae structures during mitochondria fusion,fission and inter-cristae remodeling in single mitochondria.
作者
黄小帅
李柳菊
范俊超
刘彦梅
谭山
陈良怡
Huang Xiaoshuai;Li Liuju;Fan Juanchao;Liu Yanmei;Tan Shan;Chen Liangyi(State Key Laboratory of Membrane Biology,Beijing Key Laboratory of Cardiometabolic Molecular Medicine,Institute of Molecular Medicine,Peking University,Beijing 100871,China;Key Laboratory of Image Processing and Intelligent Control of Ministry of Education of China School of Automation,Huazhong University of Science and Technology,Wuhan 430074,China)
出处
《中国科学基金》
CSSCI
CSCD
北大核心
2018年第4期367-375,共9页
Bulletin of National Natural Science Foundation of China
基金
国家自然科学基金委项目(项目批准号:31327901
31521062
31570839
31428004
61375018
61672253
91750203)
国家重点研发计划基金(批准号:2016YFA0500400)
国家重点基础研究发展计划(批准号:2013CB531200)
北京市自然科学基金-海淀原始创新联合基金项目(批准号:L172003)资助
