Abstract:The recent introduction of Direct Detector Devices (DDDs) in cryo-EM represents a crucial step forward for this structural technique.. As expected, the quality of these DDD images is much better than the one obtained from Charged Coupled Devices (CCDs) but, additionally, their fast image acquisition rate makes possible the collection of “movies” composed of individual “frames”, opening the possibility to the study of how frozen hydrated specimens temporally behave as a function of electron dose rate. Indeed, biological specimens in a solid matrix of amorphous ice behave as if they were moving when being imaged, resulting in Beam Induced Movement (BIM). It turns out that BIM is a very serious experimental “resolution barrier” in cryo-electron microscopy. However, BIM “correction” is not an easy task, and several approaches have been already proposed. In this work, we present a method to correct for BIM at the image level, resulting in an integrated image where much of the BIM blurring is compensated. The methodology is based on a robust optical flow approach that can deal both with local and global movements in a very fast manner thanks to its implementation in a Graphic Process Unit (GPU). Additionally, the spatial analysis of the optical flow in between frames allow for a detailed, objective and quantitative analysis of the BIM pattern itself, providing with a new tool to evaluate this crucial effect in cryo-EM. The new approach is publically available as part of XMIPP 3.1.
Keywords: Direct Detector Devices; Single Particle Analysis; Electron Microscopy
The authors would like to acknowledge economical support from the Spanish Ministry of Economy and Competitiveness through grants AIC–A–2011–0638 and BIO2010-16566, the Comunidad de Madrid through grant CAM(S2010/BMD-2305) and the NSF through grant 1114901, as well as postdoctoral “Juan de la Cierva” grant with reference JCI-2011-10185. C.O.S. Sorzano is recipient of a Ramón y Cajal fellow.