Deep-learning methods for contrast enhancement and artifact reduction in cryo-electron tomography: a systematic analysis of the state of the art and proposed improvements

Jones, H.N.; Deshmukh, A.; Pande, K.

doi:10.1107/S2059798326001166

Acta Crystallographica Section D
Acta Crystallographica
Section D STRUCTURAL BIOLOGY

IUCr IT WDC

home archive editors for authors for readers submit subscribe open access

view article

Figure 1
Visual comparison of deep-learning methods. Orthogonal slices through the tomogram and the log-magnitude Fourier space are shown for tomogram 5 of the EMPIAR-10045 data set at 4× binning for (a) the raw tomogram reconstructed with filtered back-projection and 3D CTF correction as implemented in IMOD and after denoising with (b) CryoCare, (c) DeepDeWedge, (d) Topaz-denoise, (e) CryoSamba and (f) IsoNet. CryoCare, Topaz-denoise and CryoSamba are denoising methods, while DeepDeWedge and IsoNet are methods for joint denoising and missing-wedge compensation. IsoNet is better at missing-wedge compensation, as can be seen by the appearance of filled-in frequencies in Fourier space.

STRUCTURAL
BIOLOGY

ISSN: 2059-7983

Volume 82| Part 3| March 2026| Pages 168-186

https://doi.org/10.1107/S2059798326001166

Open

access

Follow Acta Cryst. D

Search IUCr Journals		doi		Advanced search
Author		volume	page