Cryo-CARE: Content-Aware Image Restoration for Cryo-Transmission Electron Microscopy Data

Buchholz, Tim-Oliver; Jordan, Mareike A; Pigino, Gaia; Jug, Florian

doi:10.1109/isbi.2019.8759519

Cited by 132 publications

(118 citation statements)

References 9 publications

Supporting

Mentioning

116

Contrasting

Order By: Relevance

“…In cryo-TEM, the acquisition of high-SNR images is not possible due to beam induced damage [10]. Buchholz et al show in [3] how NOISE2NOISE training can be applied to data acquired with a direct electron detector. To enable a qualitative assessment, we applied N2V to the same data as in [3].…”

Section: Cryo-tem Datamentioning

confidence: 99%

“…In cases where ground truth data is physically unobtainable, N2N can still enable the training of denoising networks. However, this requires that two images capturing the same content (s) with independent noises (n, n ) can be acquired [3].…”

Section: Introductionmentioning

confidence: 99%

“…Additionally, we apply N2V training and prediction to three biomedical datasets: cryo-TEM images from [3], and two datasets from the Cell Tracking Challenge 2 [20]. For all these examples, the traditional training scheme cannot be applied due to the lack of ground truth data and N2N training is only applicable on the cryo-TEM data.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Noise2Void - Learning Denoising From Single Noisy Images

Krull

Buchholz

Jug

2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

Self Cite

966

827

View full text Add to dashboard Cite

The field of image denoising is currently dominated by discriminative deep learning methods that are trained on pairs of noisy input and clean target images. Recently it has been shown that such methods can also be trained without clean targets. Instead, independent pairs of noisy images can be used, in an approach known as NOISE2NOISE (N2N). Here, we introduce NOISE2VOID (N2V), a training scheme that takes this idea one step further. It does not require noisy image pairs, nor clean target images. Consequently, N2V allows us to train directly on the body of data to be denoised and can therefore be applied when other methods cannot. Especially interesting is the application to biomedical image data, where the acquisition of training targets, clean or noisy, is frequently not possible. We compare the performance of N2V to approaches that have either clean target images and/or noisy image pairs available. Intuitively, N2V cannot be expected to outperform methods that have more information available during training. Still, we observe that the denoising performance of NOISE2VOID drops in moderation and compares favorably to training-free denoising methods. Ground Truth BSD68Input BM3D

show abstract

Section: Cryo-tem Datamentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Noise2Void - Learning Denoising From Single Noisy Images

Krull

Buchholz

Jug

2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

Self Cite

966

827

View full text Add to dashboard Cite

show abstract

“…These methods offer new approaches for training deep neural network models for denoising in challenging domains. In cryo-EM, neural network denoising software has only just started to emerge for dataset-by-dataset tomogram denoising 15,16 and single particle micrograph denoising 17 . However, there have not been any systematic evaluation of these methods to date nor general denoising models developed.…”

Section: Introductionmentioning

confidence: 99%

Topaz-Denoise: general deep denoising models for cryoEM and cryoET

Bepler¹,

Kelley

Noble

et al. 2019

Preprint

View full text Add to dashboard Cite

Cryo-electron microscopy (cryoEM) is becoming the preferred method for resolving protein structure. Low signal-to-noise (SNR) in cryoEM images reduces the confidence and throughput of structure determination during several steps of data processing, resulting in impediments such as missing particle orientations. Denoising cryoEM images can not only improve downstream analysis but also accelerate the time-consuming data collection process by allowing lower electron dose micrographs to be used for analysis without compromising structural interpretability. Here, we present Topaz-Denoise, a deep learning method for reliably increasing the SNR of cryoEM images in seconds. By training on a dataset composed of thousands of micrographs collected across a wide range of imaging conditions, we are able to learn models capturing the complexity of the cryoEM image formation process. While this general idea has been deployed successfully in natural imaging, protein threading, and proteomics, it has yet to be applied systematically in cryoEM where the lack of ground truth signal has been a long-standing limitation. To address this, we make the key insight that forming paired, independent micrographs from even and odd camera movie frames enables us to train denoising models without observing ground truth signal. We demonstrate that our denoising model improves SNR by roughly 100x over raw micrographs and 1.8x over other methods. Notably, we show that denoising with our general model enables solving the first 3D single particle structure of clustered protocadherin, an elongated particle with previously-elusive views. Topaz-Denoise and pre-trained general models are now included in Topaz ( https://github.com/tbepler/topaz ), a free and open-source software package that focuses on particle picking, and has been integrated into the Appion cryoEM suite. We expect that Topaz-Denoise will be of broad utility to the cryoEM community for improving micrograph interpretability and accelerating analysis.

show abstract

“…To enhance the contrast of macromolecular structures and facilitate visualization/interpretation of tomograms and subtomogram picking a nonlinear anisotropic diffusion filter by IMOD 65 or cryo-CARE denoising strategy based on training of a neural network 69,70 .…”

Section: Cilia Preparation For Cryo-emmentioning

confidence: 99%

The molecular structure of primary cilia revealed by cryo-electron tomography

Kiesel

Viar

Tsoy

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Primary cilia are microtubule-based organelles involved in key signaling and sensing processes in eukaryotic cells. Unlike motile cilia, which have been thoroughly studied, the structure and the composition of primary cilia remain largely unexplored despite their fundamental role in development and homeostasis. They have for long been falsely regarded as simplified versions of motile cilia because they lack distinctive elements such as dynein arms, radial spokes, and central pair complex. However, revealing the detailed molecular composition and 3D structure of primary cilia is necessary in order to understand the mechanisms that govern their functions. Such structural investigations are so far being precluded by the challenging preparation of primary cilia for cryo-electron microscopy. Here, we developed an enabling method for investigating the structure of primary cilia at molecular resolution by cryo-electron tomography. We show that the well-known "9+0" arrangement of microtubule doublets is present only at the base of the primary cilium. A few microns away from the base the ciliary architecture changes into an unstructured bundle of EB1-decorated microtubule singlets and some actin filaments. Our results suggest the existence of a previously unobserved crosstalk between actin filaments and microtubules in the primary cilium. Our work provides unprecedented insights into the molecular structure of primary cilia and a general framework for uncovering their molecular composition and function in health and disease. This opens up new possibilities to study aspects of this important organelle that have so far been out of reach.We would like to thank the Electron Microscopy Facility (in particular Tobias Fürstenhaupt, Weihua Leng, Michaela Wilsch-Bräuninger) and the Light Microscope Facility from the Services and Facilities of the MPI-CBG for their support. We are thankful to Helin Rägel and Cécilie Martin-Lemaitre for their tips on MDCKII cell culture, Noreen Walker for the Imaris tutorial, and Tim-Oliver Buchholz for denoising cryo-tomography data. We thank Pavel Tomancak, Florian Jug, Dennis Diener, and Jan Brugues for the fruitful discussions and suggestions to the manuscript. We thank Oscar Gonzales for IT support. . developed the cryo-peel-off method, prepared samples for FM and EM imaging, acquired and reconstructed room temperature and cryo-tomograms, contributed to FM data acquisition, analysed EM and FM data, prepared figures, interpreted results, and contributed to writing and revising the manuscript. G.A.V. prepared samples and contributed to data acquisition of the room temperature tomography, analyzed cryo-EM data with subtomogram averaging and tomogram segmentation, analysed EM and FM data, prepared figures, interpreted results, and contributed to writing and revising the manuscript. N.T. analyzed cryo-ET data to average the microtubule singlets, contributed to supplementary figure preparation, contributed to the interpretation of data, and contributed to writing and revising the manuscript. R.M. pre...

show abstract

Cryo-CARE: Content-Aware Image Restoration for Cryo-Transmission Electron Microscopy Data

Cited by 132 publications

References 9 publications

Noise2Void - Learning Denoising From Single Noisy Images

Noise2Void - Learning Denoising From Single Noisy Images

Topaz-Denoise: general deep denoising models for cryoEM and cryoET

The molecular structure of primary cilia revealed by cryo-electron tomography

Contact Info

Product

Resources

About