A computational procedure for the analysis of electron images of nucleic acid molecules

A new procedure is described for analyzing electron images of labeled nucleic acid molecules. The method makes use of one-dimensional digital image information and determines the best relative orientation of the linear molecules. The analysis is performed with the computer program CLUSTER, which combines the information from each molecule stepwise in an iterative procedure, so that finally a label distribution is obtained, which is the combination of all information available. The rationale behind the analysis is the calculation of similarity coefficients, which are a measure of the probability for the relative orientation of each molecule pair. The method has been thoroughly tested and compared with other procedures described in the literature in order to indicate its performance and power. A biological application concerning the distribution of the protein RNA-polymerase on DNA of phage Mu is presented.     

Signal to noise enhancement in dark field electron micrographs of vasopressin: filtering of arrays of images in reciprocal space

A technique for filtering arrays of images of dispersed molecules is presented which takes advantage of the crystallographic properties of regular arrays and the rapidity of optical treatment. A filter in reciprocal space consisting of perforations in a square or rectangular lattice, as determined by the image arrangement, reduces noise and averages the images simultaneously by transmitting only regularly recurring image structure. The filter is universal for all image contents and introduces no additional biases. The signal to noise improvement approaches √N for N independent images of the same molecule in the same configuration. The procedure, which is simple, rapid and inexpensive, is demonstrated with the aid of dark field electron micrographs of the protein vasopressin and combined with iodination of that molecule to elucidate its structure.     

Real-space filtering of electron images from optical autocorrelation patterns.

The capability for filtering electron images, in real space, is demonstrated to be inherent to a two plate, incoherent optical procedure (originally described by Meyer-Eppler and Darius) for recording autocorrelation functions from transparencies. If an image contains features that occur with more than random frequency, these features may give rise to a resolvable peak in the optically recorded autocorrelation function. The optical requirements for obtaining an image of the transparency that is filtered to observe only features giving rise to some set of such peaks, or to exclude them, are described. The principle is to form an image of the transparency, with a properly placed plano-convex lens, from the incoherent light transmitted through apertures positioned over the peaks in the autocorrelation plane. The application of the method in defining the position and orientation of specific projections of protein molecules, as observed in negative stain by bright-field, or unstained by dark-field electron microscopy, is also described.     

Atomic force microscopy of freeze-fracture replicas of rat atrial tissue

Atomic force microscopy (AFM) has provided three-dimensional (3-D) surface images of many biological specimens at molecular resolution. In the absence of spectroscopic capability for AFM, it is often difficult to distinguish individual components if the specimen contains a population of mixed structures such as in a cellular membrane. In an effort to understand the AFM images better, a correlative study between AFM and the well-established technique of transmission electron microscopy (TEM) was performed. Freeze-fractured replicas of adult rat atrial tissue were examined by both TEM and AFM. The same replicas were analysed and the same details were identified, which allowed a critical comparison of surface topography by both techniques. AFM images of large-scale subcellular structures (nuclei, mitochondria, granules) correlated well with TEM images. AFM images of smaller features and surface textures appeared somewhat different from the TEM images. This presumably reflects the difference in the surface sensitivity of AFM versus TEM, as well as the nature of images in AFM (3-D surface contour) and TEM (2-D projection). AFM images also provided new information about the replica itself. Unlike TEM, it was possible to examine both sides of the replica with AFM; the resolution on one side was significantly greater compared with the other side. It was also possible to obtain quantitative height information which is not readily available with TEM.     

The fine structure of fenestrated adrenocortical capillaries revealed by in-lens field-emission scanning electron microscopy and scanning transmission electron microscopy

Cell biologists probing the physiologic movement of macromolecules and solutes across the fenestrated microvascular endothelial cell have used electron microscopy to locate the postulated pore within the fenestrae. Prior to the advent of in-lens field-emission high-resolution scanning electron microscopy (HRSEM) and ultrathin m et al coating technology, quick-freeze, platinum-carbon replica and grazing thin-section transmission electron microscopy (TEM) methods provided two-dimensional or indirect imaging methods. Wedge-shaped octagonal channels composed of fibrils interwoven in a central mesh were depicted as the filtering structures of fenestral diaphragms in images of platinum replicas enhanced by photographic augmentation. However, image accuracy was limited to replication of the cell surface. Subsequent to this, HRSEM technology was developed and provided a high-fidelity, three-dimensional topographic image of the fenestral surface directly from a fixed and dried bulk adrenal specimen coated with a 1 nm chromium film. First described from TEM replicas, the “flower-like” structure comprising the fenestral pores was readily visualized by HRSEM. High-resolution images contained particulate ectodomains on the lumenal surface of the endothelial cell membrane. Particles arranged in a rough octagonal shape formed the fenestral rim. Digital acquisition of analog photographic recordings revealed a filamentous meshwork in the diaphragm, thus confirming and extending observations from replica and grazing section TEM preparations. Endothelial cell pockets, first described in murine renal peritubular capillaries, were observed in rhesus and rabbit adrenocortical capillaries. This report features recent observations of fenestral diaphragms and endothelial pockets fitted with multiple diaphragms utilizing a Schottky field-emission electron microscope. In-lens staging of bulk and thin section specimens allowed tandem imaging in HRSEM and scanning TEM modes at 25 kV.     

A simple method for AFM tip characterization by polystyrene spheres

An AFM image would not be the true topography of a surface because of the limitation of a finite size of the tip. The true topography of the surface can be deduced if we can know the tip shape. In this paper a simple method has been established to determine the profile of an AFM tip. A geometrical model for the tip and a spherical object has been proposed to show the procedure for deducing the tip shape from AFM images. Isolated spheres and closely packed spheres with different diameters have been observed to confirm the tip shape by this method. It is a non-destructive method to determine the tip shape and the results can be used for future reconstruction of an AFM image.     

Structural features of cell walls from tomato cells adapted to grow on the herbicide 2,6-dichlorobenzonitrile

Evidence from high-resolution images of primary cell walls suggests that the cell wall is constructed from at least two independent yet coextensive fibrous networks, one based on cellulose/hemicellulose and the other on pectin. The ability to analyse the structure of each of these networks in isolation has been hampered by a lack of suitable biological material such as mutants. However, the recent use of the cellulose-synthesis inhibitor 2,6-dichlorobenzonitrile (DCB) that prevents the formation of the cellulose-xyloglucan network while allowing the pectin network to form a functional wall offers the unique opportunity of studying at least the pectin network independently. A range of electron microscopy techniques and a novel spectroscopy method are used to study the walls from tomato suspension cells adapted to growth on DCB. Measurements of the minimum cell wall thickness derived from thin sections of dehydrated walls show that the marked reduction in level of the cellulose/hemicellulose network affects neither the thickness of the wall formed, nor the apparent spacing of pectin molecules. However, images obtained by the fast-freeze, deep-etch, rotary-shadowed (FDR) replica technique show that the three-dimensional architecture of these pectin-rich walls is very different from that of nonadapted walls. Fourier transform infrared (FTIR) microspectroscopy data and immunogold-labelling studies provide additional evidence that supports the previous biochemical data.     

Mathematical classification of tight junction protein images

We present the rationale for the development of mathematical features used for classification of images stained for selected tight junction proteins. The project examined localization of zonula occludens‐1, claudin‐1 and F‐actin in a model epithelium, Madin–Darby canine kidney II cells. Cytochalasin D exposure was used to perturb junctional localization by actin cytoskeleton disruption. Mathematical features were extracted from images to reliably reveal characteristic information of the pattern of protein localization. Features, such as neighbourhood standard deviation, gradient of pixel intensity measurement and conditional probability, provided meaningful information to classify complex image sets. The newly developed mathematical features were used as input to train a neural network that provided a robust method of individual image classification. The ability for researchers to make determinations concerning image classification while minimizing human bias is an important advancement for the field of tight junction cellular biology.     

Nanometre localization of single ReAsH molecules

ReAsH is a red‐emitting dye that binds to the unique sequence Cys‐Cys‐Xaa‐Xaa‐Cys‐Cys (where Xaa is a noncysteine amino acid) in the protein. We attached a single ReAsH to a calmodulin with an inserted tetracysteine motif and immobilized individual calmodulins to a glass surface at low density. Total internal reflection fluorescence microscopy was used to image individual ReAsH molecules. We determined the centre of the distribution of photons in the image of a single molecule in order to determine the position of the dye within 5 nm precision and with an image integration time of 0.5 s. The photostability of ReAsH was also characterized and observation times ranging from several seconds to over a minute were observed. We found that 2‐mercaptoethanesulphonic acid increased the number of collected photons from ReAsH molecules by a factor of two. Individual ReAsH molecules were then moved via a nanometric stage in 25 or 40 nm steps, either at a constant rate or at a Poisson‐distributed rate. Individual steps were clearly seen, indicating that the observation of translational motion on this scale, which is relevant for many biomolecular motors, is possible with ReAsH.     

Analysis of menisci formed on cones for single field of view parasite egg microscopy

Parasite ova caused to accumulate in a single microscopic field simplifies monitoring soil‐transmitted helminthiasis by optical microscopy. Here we demonstrate new egg‐accumulating geometries based on annular menisci formed on the surface of a wetted cone. Fluidic features extracted from profile images of the system provided mathematical representations of the meniscus gradient that were compared quantitatively to numerical solutions of an axisymmetric Young–Laplace equation. Our results show that the governing dynamics of these systems is dominated by the surface tension of the fluid. These image analysis and mathematical tools provide simple quantitative methods for system analysis and optimization.     

Automated image analysis of atomic force microscopy images of rotavirus particles

A variety of biological samples can be imaged by the atomic force microscope (AFM) under environments that range from vacuum to ambient to liquid. Generally imaging is pursued to evaluate structural features of the sample or perhaps identify some structural changes in the sample that are induced by the investigator. In many cases, AFM images of sample features and induced structural changes are interpreted in general qualitative terms such as markedly smaller or larger, rougher, highly irregular, or smooth. Various manual tools can be used to analyze images and extract more quantitative data, but this is usually a cumbersome process. To facilitate quantitative AFM imaging, automated image analysis routines are being developed. Viral particles imaged in water were used as a test case to develop an algorithm that automatically extracts average dimensional information from a large set of individual particles. The extracted information allows statistical analyses of the dimensional characteristics of the particles and facilitates interpretation related to the binding of the particles to the surface. This algorithm is being extended for analysis of other biological samples and physical objects that are imaged by AFM.     

Analysis of high resolution transmission electron microscope images of crystalline-amorphous interfaces

For the analysis of images of homogeneous crystalline–amorphous interfaces we propose to average them along the interface obtaining the averaged interface image or the averaged intensity profile. Due to averaging, contrast components with the periodicity of the crystalline area of the image are extracted. Thus, the contrast features originating from the random overlap of the projected potentials of atoms in the amorphous layer are suppressed. It is shown that averaged images can be simulated by the multi-slice method using the novel approach to model the near interfacial amorphous structure by its mean atomic density distribution in front of the crystalline boundary. The crystalline structure is represented by its known atomic positions. We apply the proposed method to the investigation of the near interfacial short-range order in the c-Si/a-Ge crystalline–amorphous interface.     

Automated magnification calibration in transmission electron microscopy using Fourier analysis of replica images

The magnification factor in transmission electron microscopy is not very precise, hampering for instance quantitative analysis of specimens. Calibration of the magnification is usually performed interactively using replica specimens, containing line or grating patterns with known spacing. In the present study, a procedure is described for automated magnification calibration using digital images of a line replica. This procedure is based on analysis of the power spectrum of Fourier transformed replica images, and is compared to interactive measurement in the same images. Images were used with magnification ranging from 1,000 x to 200,000 x. The automated procedure deviated on average 0.10% from interactive measurements. Especially for catalase replicas, the coefficient of variation of automated measurement was considerably smaller (average 0.28%) compared to that of interactive measurement (average 3.5%). In conclusion, calibration of the magnification in digital images from transmission electron microscopy may be performed automatically, using the procedure presented here, with high precision and accuracy.     

Computer-assisted morphometry: point, intersection, and profile counting and three-dimensional reconstruction.

The use of computers in morphometry can involve 1) automated image analysis, semiautomated image analysis and point, intersection, intercept and profile counts of two-dimensional images on tissue sections with mathematical extrapolation to the third dimension, 2) direct measurement of volumes, surfaces, lengths, and curvature using x,y,z coordinates of serial sectioned images, or 3) stereologic techniques and serial sections which is a combination of 1 and 2 above. Automated and semiautomated image analysis are generally restricted to specimens that are characterized by differential contrast such as interalveolar septa in the lung or histochemically stained mucous granules in pulmonary epithelium. Point, intersection, and profile counts using hand-held, notebook PCs, portable PCs, or standard PCs and MS-DOS-based application programs are extremely efficient, precise, affordable, and convenient methods of quantitating average values of a population. When morphometric measurements of individual structures are required, computer-assisted three-dimensional reconstruction using x,y,z coordinates of the surface outline from serial sections is a tedious yet precise method. We describe a computer program that efficiently estimates mean caliper diameter, volume, and surface area with less than five percent error with five sections per structure. We also describe a program that does digital image subtraction on serial sections, superimposes digitally generated test systems on biological images, and accumulates point, intersection, and profile counts using a Macintosh II series computer.     

Automatic selection of molecular images from dark field electron micrographs

A method is described which can be used objectively to select putative molecular images from dark field electron micrographs of unstained molecules. The only characteristic of the molecule required for automatic selection is an estimate of molecular weight. Structures are selected from micrographs by a series of steps including: low pass filtering, edge detection and mass determination. The procedure is shown to be reliable for images with signal-to-noise ratios of at least 4.0. Moreover, the method is insensitive to both the shape and the number of molecules in the image. Five different molecules with molecular weights between MW 330,000 and MW 4000 are successfully selected from low dose STEM and high dose tilt beam dark field electron micrographs.     

Cytoskeletal architecture of neuromuscular junction: Localization of vinculin

Cytoskeletons underneath the postsynaptic membrane of neuromuscular junctions were studied by using a quick-freeze deep-etched method and immunoelectron microscopy of ultrathin frozen sections. In a quick-freeze deep-etched replica of fresh, unfixed muscles, 8.9 ± 1.5-nm particles were present on the true postsynaptic membrane surface. Underneath this receptor-rich postsynaptic membrane, networks of fine filaments were observed. These cytoskeletal networks were more clearly observed in extracted samples. In these samples, diameters of the filaments which formed networks were measured. In the platinum replica, three kinds of filament were recognized—12 nm, 9 nm, and 7 nm in diameter. The 12-nm filament seemed to correspond to the intermediate filament. The other two filaments formed meshworks between intermediate filaments and plasma membrane. In ultrathin frozen sections vinculin label was localized just beneath the plasma membrane. Thirty-six percent of the label was within 18 nm from the cytoplasmic side of the plasma membrane and 50% was within 30 nm. Taking the size of the vinculin molecule into account, it was concluded that vinculin is localized just beneath the plasma membrane and might play some role in anchoring filaments which formed meshworks underneath the plasma membrane.     

Jumping mode AFM imaging of biomolecules in the repulsive electrical double layer

We present a method to image single biomolecules in aqueous media by atomic force microscope (AFM) without establishing any mechanical contact between the tip and the sample. It works by placing the feedback set point in the repulsive electrical double-layer curve just before the mechanical instability occurs. We use the jumping operation mode, where the set point is controlled at every image point and a stable imaging is achieved for several hours. This is a necessary condition for this method to be operative, otherwise the tip can fall in contact in a short time. The method is applied to image single-avidin protein molecules deposited on cleaved mica. In addition, the dependence of the height of avidin molecules as a function of ion concentration, due to differences in surface charge density of mica and avidin, is tentatively used to deduce relative values of these quantities.     

Feature selection for surface defect classification of extruded aluminum profiles

This research investigates detection and classification of two types of the surface defects in extruded aluminium profiles; blisters and scratches. An experimental system is used to capture images and appropriate statistical features from a novel technique based on gradient-only co-occurrence matrices (GOCM) are proposed to detect and classify three distinct classes; non-defective, blisters and scratches. The developed methodology makes use of the Sobel edge detector to obtain the gradient magnitude of the image (GOCM). A comparison is made between the statistical features extracted from the original image (GLCM) and those extracted from the gradient magnitude (GOCM). This paper describes in detail every step of the image processing with example pictures illustrating the methodology. The features extracted from the image processing are classified by a two-layer feed-forward artificial neural network. The artificial neural network training is tested using different combinations of statistical features with different topologies. Features are compared individually and grouped. Results are discussed, achieving up to 98.6 % total testing accuracy.     

基于图像拼接的表面粗糙度测量方法

针对光切显微镜图像法测量表面粗糙度时取样长度较小，不足以客观表征工件表面质量的问 题，提出一种基于图像拼接的表面粗糙度测量方法。根据表面粗糙度序列图像中相邻帧图像的重叠区域构 建匹配模板，采用归一化相关算法计算相似度量进行相邻帧图像匹配与拼接，以加权融合算法平滑拼接后 图像的重影和接缝。对拼接后的粗糙度图像提取单侧边缘特征，通过最小二乘法拟合最小二乘轮廓中线， 建立轮廓算 数 平 均 偏 差 和 轮 廓 最 大 高 度 评 定 模 型。实 验 结 果 表 明，采 用 拼 接 算 法 后 取 样 长 度 增 加 了 １９８３．５２μｍ，测量精度平均提高了１．１６％。