FIRT : filtered iterative reconstruction technique with information restoration
(In collaboration with Prof. Fa Zhang, Institute of Computation Technology, CAS)
Electron tomography (ET) combining subsequent sub-volume averaging has been becoming a unique way to study the in-situ 3D structures of macromolecular complexes. However, information missing in electron tomography due to limited angular sampling is still the bottleneck in high-resolution electron tomography application. Based on the understanding of smooth nature of biological specimen, we developed a new iterative image reconstruction algorithm, FIRT (filtered iterative reconstruction technique) for electron tomography by combining the algebra reconstruction technique (ART) and the nonlinear diffusion (ND) filter technique. In comparison to ART and weight back projection method, FIRT could generate a better reconstruction with reduced ray artifacts and significant improved correlation with ground truth (for simulated data). Furthermore, FIRT could partially restore the information at the non-sampled angular region, which will be subsequently useful for both cellular and molecular ET with better quality and improved structural details.
Reference:
Chen Y., Zhang Y., Zhang K.*, Deng Y., Zhang F.* and Sun F.* (2016), FIRT: filtered iterative reconstruction technique with information restoration. Journal of Structural Biology 195(1) : 49-61. doi: 10.1016/j.jsb.2016.04.015.
(Changes of this version : 1. The cross-validation procedure was further improved with an automatic determination of the verification resolution. 2. A preprocessing procedure containing mode subtraction and thickness-related normalization was added. 3. A program for verification filtering in Fourier domain was added.)
(Changes of this version : 1. The help information was updated for better description. 2. Fix the errors when reading some MRC files with non-consistent head.)
