cryo-EM Images Research Articles

SPREAD: A Fully Automated Toolkit for Single-Particle Cryogenic Electron Microscopy Data 3D Reconstruction with Image-Network-Aided Orientation Assignment.

For the past decade, cryogenic electron microscopy (cryo-EM) has become an important technology to determine three-dimensional (3D) structures of biomacromolecules. Many software tools have been developed for cryo-EM image processing and 3D reconstruction, covering various computational tasks in cryo-EM data analysis. Despite the recent progress, most of these software tools focus on a single task, such as automatic particle picking or image clustering, whereas software packages covering the whole pipeline of cryo-EM data processing are still few. In this study, we developed a fully automatic single-particle reconstruction and analysis toolkit for cryo-EM data, named SPREAD, which integrates 2D image classification, 3D initial model generation, model selection, and 3D refinement. In SPREAD, we adopt our previously proposed network-based clustering algorithm for 2D image classification, NCEM, and the reference-free resolution measurement method SRes to realize the automatic model ranking and selection procedure. Projection orientation assignment is one of the key steps in initial model generation and 3D refinement. In SPREAD, we use the network-based image similarity metric and introduce a new probabilistic-based orientation searching method, named peak finding, to enhance assignment of the projection orientations. For dealing with both the particle images and projection images in the 3D refinement using SPREAD, we build a mixture image network containing both of these types of images on the basis of the peak-finding results, and then similarities for node pairs are recomputed by a superposed random walk on the network. SPREAD achieves a fully automatic workflow in which nearly no expert domain knowledge and interactive manual operation are involved. Our software can accessed for free at http://www.csbio.sjtu.edu.cn/bioinf/SPREAD/ for academic use.

Research on Denoising of Cryo-em Images Based on Deep Learning

Cryo-em (Cryogenic electron microscopy) is a technology this can build bio-macromolecule of three-dimensional structure. Under the condition of now, the projection image of the biological macromolecule which is collected by the Cryo-em technology that the contrast is low, the signal to noise is low, image blurring, and not easy to distinguish single particle from background, the corresponding processing technology is lagging behind. Therefore, make Cryoem image denoising useful, and maintaining bio-macromolecule of contour or signal of function-construct improve Cryo-em image quality or resolution of Cryo-em three-dimensional structure have important effect. This paper researched a denoising function base on GANs (generative adversarial networks), purpose an improved discriminant model base on Wasserstein distance and an improved image denoising model by add gray constraint. Our model turn discriminant model’s training process from binary classification’s training process into regression task training process, it make GANs in training process more stable, more reasonable parameter passing. Meantime, we also propose an improved generative model by add gray constraint. The experimental results show that our model can increase the peak signal-to-noise ratio of the Cryo-em simulation image by 10.3 dB and improve SSIM (Structural Similarity Index) of the denoised image results by 0.43. Compared with traditional image denoising algorithms such as BM3D (Block Matching 3D), our model can better save the model structure and the vein signal in the original image and the operation speed is faster.

An Insight into NTP (Neural Theorem Provers) in the Context of Image Processing Algorithms R&D-A Simple Suggestion and Short Communication on Exploring HO List and DeepHOL: Machine Learning (ML) of Higher Order Theorem Proving

DeepHOL Neural Prover/HOLight/HOL-Light interfacing with [QRNG-libqrng/Python+ImageAI-Python/Ising Models/Helmholtz Theory/Equations in Python] in the context of [Medical Image Processing/Electron Microscopy(EM)→SEM/TEM/AFM/Cryo-Electron Microscopy(cryo-EM) Image Processing/IoT/HPC/Minsky Machines/Mongo DB-Python based R&D Algorithms] as a useful insight into the World of Advanced Image Processing Informatics Frameworks.

Automatic identification of crossovers in cryo-EM images of murine amyloid protein A fibrils with machine learning.

Detecting crossovers in cryo-electron microscopy images of protein fibrils is an important step towards determining the morphological composition of a sample. Currently, the crossover locations are picked by hand, which introduces errors and is a time-consuming procedure. With the rise of deep learning in computer vision tasks, the automation of such problems has become more and more applicable. However, because of insufficient quality of raw data and missing labels, neural networks alone cannot be applied successfully to target the given problem. Thus, we propose an approach combining conventional computer vision techniques and deep learning to automatically detect fibril crossovers in two-dimensional cryo-electron microscopy image data and apply it to murine amyloid protein A fibrils, where we first use direct image processing methods to simplify the image data such that a convolutional neural network can be applied to the remaining segmentation problem. LAY DESCRIPTION: The ability of protein to form fibrillary structures underlies important cellular functions but can also give rise to disease, such as in a group of disorders, termed amyloid diseases. These diseases are characterised by the formation of abnormal protein filaments, so-called amyloid fibrils, that deposit inside the tissue. Many amyloid fibrils are helically twisted, which leads to periodic variations in the apparent width of the fibril, when observing amyloid fibrils using microscopy techniques like cryogenic electron microscopy (cryo-EM). Due to the two-dimensional projection, parts of the fibril orthogonal to the projection plane appear narrower than parts parallel to the plane. The parts of small width are called crossovers. The distance between two adjacent crossovers is an important characteristic for the analysis of amyloid fibrils, because it is informative about the fibril morphology and because it can be determined from raw data by eye. A given protein can typically form different fibril morphologies. The morphology can vary depending on the chemical and physical conditions of fibril formation, but even when fibrils are formed under identical solution conditions, different morphologies may be present in a sample. As the crossovers allow to define fibril morphologies in a heterogeneous sample, detecting crossovers is an important first step in the sample analysis. In the present paper, we introduce a method for the automated detection of fibril crossovers in cryo-EM image data. The data consists of greyscale images, each showing an unknown number of potentially overlapping fibrils. In a first step, techniques from image analysis and pattern detection are employed to detect single fibrils in the raw data. Then, a convolutional neural network is used to find the locations of crossovers on each single fibril. As these predictions may contain errors, further postprocessing steps assess the quality and may slightly alter or reject the predictedcrossovers.

Macromolecule Particle Picking and Segmentation of a KLH Database by Unsupervised Cryo-EM Image Processing.

Segmentation is one of the most important stages in the 3D reconstruction of macromolecule structures in cryo-electron microscopy. Due to the variability of macromolecules and the low signal-to-noise ratio of the structures present, there is no generally satisfactory solution to this process. This work proposes a new unsupervised particle picking and segmentation algorithm based on the composition of two well-known image filters: Anisotropic (Perona–Malik) diffusion and non-negative matrix factorization. This study focused on keyhole limpet hemocyanin (KLH) macromolecules which offer both a top view and a side view. Our proposal was able to detect both types of views and separate them automatically. In our experiments, we used 30 images from the KLH dataset of 680 positive classified regions. The true positive rate was 95.1% for top views and 77.8% for side views. The false negative rate was 14.3%. Although the false positive rate was high at 21.8%, it can be lowered with a supervised classification technique.

Filling Adeno-Associated Virus Capsids: Estimating Success by Cryo-Electron Microscopy.

Adeno-associated viruses (AAVs) have been employed successfully as gene therapy vectors in treating various genetic diseases for almost two decades. However, transgene packaging is usually imperfect, and developing a rapid and accurate method for measuring the proportion of DNA encapsidation is an important step for improving the downstream process of large scale vector production. In this study, we used two-dimensional class averages and three-dimensional classes, intermediate outputs in the single particle cryo-electron microscopy (cryo-EM) image reconstruction pipeline, to determine the proportion of DNA-packaged and empty capsid populations. Two different preparations of AAV3 were analyzed to estimate the minimum number of particles required to be sampled by cryo-EM in order for robust calculation of the proportion of the full versus empty capsids in any given sample. Cost analysis applied to the minimum amount of data required for a valid ratio suggests that cryo-EM is an effective approach to analyze vector preparations.

A comparative study of single-particle cryo-EM with liquid-nitrogen and liquid-helium cooling.

Radiation damage is the most fundamental limitation for achieving high resolution in electron cryo-microscopy (cryo-EM) of biological samples. The effects of radiation damage are reduced by liquid-helium cooling, although the use of liquid helium is more challenging than that of liquid nitrogen. To date, the benefits of liquid-nitrogen and liquid-helium cooling for single-particle cryo-EM have not been compared quantitatively. With recent technical and computational advances in cryo-EM image recording and processing, such a comparison now seems timely. This study aims to evaluate the relative merits of liquid-helium cooling in present-day single-particle analysis, taking advantage of direct electron detectors. Two data sets for recombinant mouse heavy-chain apoferritin cooled with liquid-nitrogen or liquid-helium to 85 or 17 K were collected, processed and compared. No improvement in terms of resolution or Coulomb potential map quality was found for liquid-helium cooling. Interestingly, beam-induced motion was found to be significantly higher with liquid-helium cooling, especially within the most valuable first few frames of an exposure, thus counteracting any potential benefit of better cryoprotection that liquid-helium cooling may offer for single-particle cryo-EM.

A microtubule RELION-based pipeline for cryo-EM image processing

Microtubules are polar filaments built from αβ-tubulin heterodimers that exhibit a range of architectures in vitro and in vivo. Tubulin heterodimers are arranged helically in the microtubule wall but many physiologically relevant architectures exhibit a break in helical symmetry known as the seam. Noisy 2D cryo-electron microscopy projection images of pseudo-helical microtubules therefore depict distinct but highly similar views owing to the high structural similarity of α- and β-tubulin. The determination of the αβ-tubulin register and seam location during image processing is essential for alignment accuracy that enables determination of biologically relevant structures. Here we present a pipeline designed for image processing and high-resolution reconstruction of cryo-electron microscopy microtubule datasets, based in the popular and user-friendly RELION image-processing package, Microtubule RELION-based Pipeline (MiRP). The pipeline uses a combination of supervised classification and prior knowledge about geometric lattice constraints in microtubules to accurately determine microtubule architecture and seam location. The presented method is fast and semi-automated, producing near-atomic resolution reconstructions with test datasets that contain a range of microtubule architectures and binding proteins.

P5025Structural insights into catecholaminergic polymorphic ventricular tachycardia-associated RyR2 mutant channels using a three-dimensional in silico model

Abstract Background The cardiac ryanodine receptors (RyR2) are large tetrameric calcium-permeant ion channels found in cardiac muscle sarcoplasmic reticulum, which play an important role in the control of intracellular Ca2+ release and cardiac contraction. Mutations in the RYR2 gene are associated with lethal arrhythmia diseases including catecholaminergic polymorphic ventricular tachycardia (CPVT) resulting from increased diastolic Ca2+ leak from mutant channels. RyR2 is a huge protein that each subunit of tetramer is comprised of 4967 amino acids, which hampers the detailed in vitro analysis of RyR2 mutant channels. Purpose We aimed to analyze the structural features of RyR2 mutant channels identified in our cohort with inherited arrhythmias using RyR2 three-dimensional (3D) in silico model to reveal the arrhythmogenic mechanisms. Methods A targeted next-generation sequencing panel for inherited arrhythmias was employed for genetic diagnosis of the patients. Then, we mapped the identified mutations on RyR2 3D structural model developed by cryo-EM images (PDB: 5go9, 5goa, Peng Science 2016) and investigated the relationship between the location of the mutations and specific functional sites. Results As a result of genetic analysis, we identified 93 RYR2 mutations from 112 probands with CPVT (n=93) or long-QT syndrome (LQTS) (n=19).64 of 93 (69%) RYR2 mutations are located in three “hot-spot” area (N-terminal (residues 77–466), central (2246–2534), and channel (3778–4959) hotspot. RyR2 3D in silico modeling revealed that the mutations are regionally distributed mainly in three parts: N-terminal, periphery, and channel part (Figure A). In N-terminal part (1–642 amino acid), 9 of 13 mutations alter the charges of the amino acids (Figure B). Especially, R169L, R169Q, and G172E are close to the interface between two neighboring subunits (∼20Å). These mutations which change the amino acid charge may cause a complete disruption of the ionic pair network and result in largest structural changes, which facilitates RyR2 channel opening. In periphery part (643–3528aa), 22 of 33 mutations are close to the two predicted binding sites of FKBP12.6, a stabilizer of RyR2 (∼5–40Å, Figure C). The mutations are supposed to disturb the binding affinity to the FKBP12.6 resulting in RyR2 channel instability. In channel part (3613–4968aa), 16 of 40 mutations are located near two interface. (FigureD) 12 mutations are close to the Ca2+ sensor and the other 4 mutations are adjacent to the pore-forming segment. Especially, V4821I is just located on this segment and strongly expected to impair the channel function. Above all, RyR2 3D in silico modeling revealed that 63 of all 93 (68%) identified mutations are supposed to be pathogenic. Location of RYR2 mutations in 3D model Conclusion 3D structural model of RyR2 is useful for the investigation of the pathogenic mechanisms of CPVT-related mutations. Further studies are needed to elucidate the relationship between the location of the mutations and clinical phenotypes.

Diverse Morphology and Structural Features of Old and New World Hantaviruses.

To further understanding of the structure and morphology of the Orthohantavirus, family Hantaviridae, we have employed cryo-electron microscopy (cryo-EM) for three New World hantaviruses: Andes (ANDV), Sin Nombre (SNV), and Black Creek Canal (BCCV). Building upon our prior cryo-EM and cryo-tomography study of the Old World hantavirus, Hantaan virus (HTNV), we have expanded our studies to examine the entire virion population present in cell culture supernatant. Hence, in contrast to the prior cryo-EM/ET studies in which we used a polyethylene precipitation, a sucrose gradient, and a sucrose cushion, we used two sucrose cushions. We inactivated the material after the first cushion. We tested the method using HTNV which has a known cryo-EM structure and observed equivalent results. Therefore, we used this method to assess the particle distribution of the New World hantaviruses by cryo-EM. Cryo-EM images showed a diverse range of sizes and morphologies for the New World viruses that we classified as round, tubular, and irregular. Strikingly, BCCV virions were mostly tubular. These first cryo-EM images of the New World Orthohantavirus confirm prior EM observations that noted tubular projections of SNV at the plasma membrane during virion morphogenesis but were not confirmed. These findings underscore the need for further investigation of virion morphogenesis of the Orthohantavirus.

Electron cryo-microscopy for elucidating the dynamic nature of live-protein complexes

BackgroundRecent innovations in the field of electron cryo-microscopy (cryo-EM) have achieved structure determination of target macromolecules at “near-atomic resolution”. In addition, cryo-EM has the potential to deal with proteins in multiple conformations in close to physiological conditions. Scope of reviewThis is an overview of the latest technical and methodological developments in cryo-EM, especially key features for elucidating the dynamic nature of specimens. Major conclusionsIt is now possible to elucidate “near-atomic resolution” structures by selecting “well-behaved” particles in Single Particle Analysis (SPA) of cryo-EM images. General significanceCurrently, cryo-EM is utilized actively as an alternative to X-ray crystallography to obtain high-resolution structures of macromolecules, especially those which are hard to crystalize. One possible reason for difficulty in crystallization could be the dynamic nature of the specimen. New developments in the field of cryo-EM will make it possible to deal with specimen heterogeneity and reveal the dynamic nature of specimens.

A Super-Clustering Approach for Fully Automated Single Particle Picking in Cryo-EM.

Structure determination of proteins and macromolecular complexes by single-particle cryo-electron microscopy (cryo-EM) is poised to revolutionize structural biology. An early challenging step in the cryo-EM pipeline is the detection and selection of particles from two-dimensional micrographs (particle picking). Most existing particle-picking methods require human intervention to deal with complex (irregular) particle shapes and extremely low signal-to-noise ratio (SNR) in cryo-EM images. Here, we design a fully automated super-clustering approach for single particle picking (SuperCryoEMPicker) in cryo-EM micrographs, which focuses on identifying, detecting, and picking particles of the complex and irregular shapes in micrographs with extremely low signal-to-noise ratio (SNR). Our method first applies advanced image processing procedures to improve the quality of the cryo-EM images. The binary mask image-highlighting protein particles are then generated from each individual cryo-EM image using the super-clustering (SP) method, which improves upon base clustering methods (i.e., k-means, fuzzy c-means (FCM), and intensity-based cluster (IBC) algorithm) via a super-pixel algorithm. SuperCryoEMPicker is tested and evaluated on micrographs of β-galactosidase and 80S ribosomes, which are examples of cryo-EM data exhibiting complex and irregular particle shapes. The results show that the super-particle clustering method provides a more robust detection of particles than the base clustering methods, such as k-means, FCM, and IBC. SuperCryoEMPicker automatically and effectively identifies very complex particles from cryo-EM images of extremely low SNR. As a fully automated particle detection method, it has the potential to relieve researchers from laborious, manual particle-labeling work and therefore is a useful tool for cryo-EM protein structure determination.

Correlated Cryo Super‐Resolution Light and Cryo‐Electron Microscopy on Mammalian Cells Expressing the Fluorescent Protein rsEGFP2

AbstractSuper‐resolution light microscopy (SRM) enables imaging of biomolecules within cells with nanometer precision. Cryo‐fixation by vitrification offers optimal structure preservation of biological specimens and permits sequential cryo electron microscopy (cryoEM) on the same sample, but is rarely used for SRM due to various technical challenges and the lack of fluorophores developed for vitrified conditions. Here, a protocol to perform correlated cryoSRM and cryoEM on intact mammalian cells using fluorescent proteins and commercially available equipment is described. After cell culture and sample preparation by plunge‐freezing, cryoSRM is performed using the reversibly photoswitchable fluorescent protein rsEGFP2. Next, a super‐resolved image is reconstructed to guide cryoEM imaging to the feature of interest. Finally, the cryoSRM and cryoEM images are correlated to combine information from both imaging modalities. Using this protocol, a localization precision of 30 nm for cryoSRM is routinely achieved. No impediments to successive cryoEM imaging are detected, and the protocol is compatible with a variety of cryoEM techniques. When the optical set‐up and analysis pipeline is established, the total duration of the protocol for experienced cryoEM users is 3 days, not including cell culture.

Negative‐Stain Transmission Electron Microscopy of Molecular Complexes for Image Analysis by 2D Class Averaging

Negative‐stain transmission electron microscopy (EM) is a technique that has provided nanometer resolution images of macromolecules for about 60 years. Developments in cryo‐EM image processing have maximized the information gained from averaging large numbers of particles. These developments can now be applied back to negative‐stain image analysis to ascertain domain level molecular structure (10 to 20 Å) more quickly and efficiently than possible by atomic resolution cryo‐EM. Using uranyl acetate stained molecular complexes of influenza hemagglutinin bound to Fab 441D6, we describe a simple and efficient means to collect several hundred micrographs with SerialEM. Using RELION, we illustrate how tens of thousands of complexes can be auto‐picked and classified to accurately describe the domain level topology of this unconventional hemagglutinin head‐domain epitope. By comparing to the cryo‐EM density map of the same complex, we show that questions about epitope mapping and conformational heterogeneity can readily be answered by this negative‐stain method. © 2019 The Authors.

Cryo-EM workshop: lectures on cryo-EM image formation and 3D reconstruction

Cryo-EM workshop: lectures on cryo-EM image formation and 3D reconstruction

AutoCryoPicker: an unsupervised learning approach for fully automated single particle picking in Cryo-EM images

BackgroundAn important task of macromolecular structure determination by cryo-electron microscopy (cryo-EM) is the identification of single particles in micrographs (particle picking). Due to the necessity of human involvement in the process, current particle picking techniques are time consuming and often result in many false positives and negatives. Adjusting the parameters to eliminate false positives often excludes true particles in certain orientations. The supervised machine learning (e.g. deep learning) methods for particle picking often need a large training dataset, which requires extensive manual annotation. Other reference-dependent methods rely on low-resolution templates for particle detection, matching and picking, and therefore, are not fully automated. These issues motivate us to develop a fully automated, unbiased framework for particle picking.ResultsWe design a fully automated, unsupervised approach for single particle picking in cryo-EM micrographs. Our approach consists of three stages: image preprocessing, particle clustering, and particle picking. The image preprocessing is based on multiple techniques including: image averaging, normalization, cryo-EM image contrast enhancement correction (CEC), histogram equalization, restoration, adaptive histogram equalization, guided image filtering, and morphological operations. Image preprocessing significantly improves the quality of original cryo-EM images. Our particle clustering method is based on an intensity distribution model which is much faster and more accurate than traditional K-means and Fuzzy C-Means (FCM) algorithms for single particle clustering. Our particle picking method, based on image cleaning and shape detection with a modified Circular Hough Transform algorithm, effectively detects the shape and the center of each particle and creates a bounding box encapsulating the particles.ConclusionsAutoCryoPicker can automatically and effectively recognize particle-like objects from noisy cryo-EM micrographs without the need of labeled training data or human intervention making it a useful tool for cryo-EM protein structure determination.

A Representation Theory Perspective on Simultaneous Alignment and Classification.

Single particle cryo-electron microscopy (EM) is a method for determining the 3-D structure of macromolecules from many noisy 2-D projection images of individual macromolecules whose orientations and positions are random and unknown. The problem of orientation assignment for the images motivated work on general multireference alignment. The recently introduced non-unique games framework provides a representation theoretic approach to alignment over compact groups, and offers a convex relaxation which is formulated as semidefinite programs with certificates of global optimality under certain circumstances. One of the great opportunities in cryo-EM is studying heterogeneous samples, containing two or more distinct classes or conformations of molecules. Taking advantage of this opportunity presents an algorithmic challenge: determining both the class and orientation of each particle. We generalize multireference alignment to a problem of alignment and classification, and we propose to extend non-unique games to the problem of simultaneous alignment and classification with the goal of simultaneously classifying cryo-EM images and aligning them within their respective classes.

Cryo-EM structures of herpes simplex virus type 1 portal vertex and packaged genome

Herpesviruses are enveloped viruses prevalent in the human population, responsible for a host of pathologies ranging from cold sores to birth defects and cancers. They are characterized by a highly pressurized, T (triangulation number) = 16 pseudo-icosahedral capsid encapsidating a tightly packed dsDNA genome1–3. A key process in the herpesvirus life cycle involves the recruitment of an ATP-driven terminase to a unique portal vertex to recognize, package, and cleave concatemeric dsDNA, ultimately giving rise to a pressurized, genome-containing virion4,5. Though this process has been studied in dsDNA phages6–9—with which herpesviruses bear some similarities—a lack of high-resolution in situ structures of genome-packaging machinery has prevented the elucidation of how these multi-step reactions, which require close coordination among multiple actors, occur in an integrated environment. Thus, to better define the structural basis of genome packaging and organization in the prototypical herpesvirus, herpes simplex virus type 1 (HSV-1), we developed sequential localized classification and symmetry relaxation methods to process cryoEM images of HSV-1 virions, enabling us to decouple and reconstruct hetero-symmetric and asymmetric elements within the pseudo-icosahedral capsid. Here we show in situ structures of the unique portal vertex, genomic termini, and ordered dsDNA coils in the capsid spooled around a disordered dsDNA core. We identify tentacle-like helices and a globular complex capping the portal vertex not observed in phages, indicative of adaptations in the DNA-packaging process specific to herpesviruses. Finally, our atomic models of portal vertex elements reveal how the five-fold-related capsid accommodates symmetry mismatch imparted by the dodecameric portal—long a mystery in icosahedral viruses—and inform possible DNA sequence-recognition and headful-sensing pathways involved in genome packaging. Our work represents the first fully symmetry-resolved structure of a portal vertex and first atomic model of a portal complex in a eukaryotic virus.

Fluorescence-Based Detection of Membrane Fusion State on a Cryo-EM Grid using Correlated Cryo-Fluorescence and Cryo-Electron Microscopy.

Correlated light and electron microscopy (CLEM) has become a popular technique for combining the protein-specific labeling of fluorescence with electron microscopy, both at room and cryogenic temperatures. Fluorescence applications at cryo-temperatures have typically been limited to localization of tagged protein oligomers due to known issues of extended triplet state duration, spectral shifts, and reduced photon capture through cryo-CLEM objectives. Here, we consider fluorophore characteristics and behaviors that could enable more extended applications. We describe how dialkylcarbocanine DiD, and its autoquenching by resonant energy transfer (RET), can be used to distinguish the fusion state of a lipid bilayer at cryo-temperatures. By adapting an established fusion assay to work under cryo-CLEM conditions, we identified areas of fusion between influenza virus-like particles and fluorescently labeled lipid vesicles on a cryo-EM grid. This result demonstrates that cryo-CLEM can be used to localize functions in addition to tagged proteins, and that fluorescence autoquenching by RET can be incorporated successfully into cryo-CLEM approaches. In the case of membrane fusion applications, this method provides both an orthogonal confirmation of functional state independent of the morphological description from cryo-EM and a way to bridge room-temperature kinetic assays and the cryo-EM images.

Clustering Enhancement of Noisy Cryo-Electron Microscopy Single-Particle Images with a Network Structural Similarity Metric

The reconstruction of a three-dimensional model from cryo-electron microscopy (cryo-EM) two-dimensional images is currently a mainstream technology for revealing the structure of biomacromolecules. In this structure solution protocol, an important step is to identify each particle's projection orientation. Because the obtained single-particle images are often too noisy, clustering is an important step to mitigate noise by averaging images within the same class. The core of clustering is to place similar cryo-EM images into the same class; hence, measurement of similarity between data samples is an essential element in any clustering algorithm. As the cryo-EM images are highly noisy, directly measuring the similarity of two images will be easily biased by the hidden noise. In this study, we propose a new network structural similarity metric-based clustering protocol NCEM for clustering the noisy cryo-EM images. We first construct an image complex network for all of the cryo-EM single-particle images, where each image is represented as a node in the network. Then the similarity between two images is refined from the network structural geometry. By extending the similarity measurement from two independent images to their corresponding neighboring sets in the network, this new NCEM has typical advantages over direct measurement of two images for its noise resistance by using the structural information on the network. Our experimental results for both synthetic and real data sets demonstrate the efficacy of the protocol.