Markov random field based automatic alignment for low SNR imagesfor cryo electron tomography
We present a method for automatic full precision alignmentof the images in a tomographic tilt series. Full-precision automaticalignment of cryo electron microscopy images has remained a difficultchallenge to date, due to the limited electron dose and low imagecontrast. These facts lead to poor signal to noise ratio (SNR) in theimages, which causes automatic feature trackers to generate errors, evenwith high contrast gold particles as fiducial features. To enable fullyautomatic alignment for full-precision reconstructions, we frame theproblem probabilistically as finding the most likely particle tracksgiven a set of noisy images, using contextual information to make thesolution more robust to the noise in each image. To solve this maximumlikelihood problem, we use Markov Random Fields (MRF) to establish thecorrespondence of features in alignment and robust optimization forprojection model estimation. The resultingalgorithm, called RobustAlignment and Projection Estimation for Tomographic Reconstruction, orRAPTOR, has not needed any manual intervention for the difficult datasets we have tried, and has provided sub-pixel alignment that is as goodas the manual approach by an expert user. We are able to automaticallymap complete and partial marker trajectories and thus obtain highlyaccurate image alignment. Our method has been applied to challenging cryoelectron tomographic datasets with low SNR from intact bacterial cells,as well as several plastic section and x-ray datasets.
- Research Organization:
- COLLABORATION - StanfordU.
- Sponsoring Organization:
- USDOE Director. Office of Science. Biological andEnvironmental Research
- DOE Contract Number:
- DE-AC02-05CH11231
- OSTI ID:
- 929767
- Report Number(s):
- LBNL-63633-JArt; JSBIEM; R&D Project: 443E01; BnR: KP1501021; TRN: US200812%%657
- Journal Information:
- Journal of Structural Biology, Vol. 161, Issue 3; Related Information: Journal Publication Date: 03/2008; ISSN 1047-8477
- Country of Publication:
- United States
- Language:
- English
Similar Records
Image registration of low signal-to-noise cryo-STEM data
Cryo-STEM Tomography with Inpainting