A novel three-dimensional variant of the watershed transform for segmentation of electron density maps

Electron density maps at moderate resolution are often difficult to interpret due to the lack of recognizable features. This is especially true for electron tomograms that suffer in addition to the resolution limitation from low signal-to-noise ratios. Reliable segmentation of such maps into smaller, manageable units can greatly facilitate interpretation. Here, we present a segmentation approach targeting three-dimensional electron density maps derived by electron microscopy. The approach consists of a novel three-dimensional variant of the immersion-based watershed algorithm. We tested the algorithm on calculated data and applied it to a wide variety of electron density maps ranging from reconstructions of single macromolecules to tomograms of subcellular structures. The results indicate that the algorithm is reliable, efficient, accurate, and applicable to a wide variety of biological problems.     

Segmentation of thin structures in electron micrographs using orientation fields

In this paper, we introduce a new approach for segmenting thin structures in electron micrographs. We introduce two new transforms, the Line Filter Transform (LFT) and the Orientation Filter Transform (OFT). The LFT can be viewed as an alternative to anisotropic diffusion algorithms that is particularly useful for thin structures. The OFT utilizes geometrical information about the structure by measuring correlations of local orientations in the image. By combining these methods with a contour extraction and labeling method we construct a segmentation method for thin structures in 2D images. We discuss how the method can be applied slice-by-slice to electron tomograms and illustrate the process by constructing two models of membrane structures from cellular tomograms. The suggested method has the advantage of being relatively insensitive to non-uniform contrast and high-contrast features such as ribosomes.     

Visualization of ATP Synthase Dimers in Mitochondria by Electron Cryo-tomography

Electron cryo-tomography is a powerful tool in structural biology, capable of visualizing the three-dimensional structure of biological samples, such as cells, organelles, membrane vesicles, or viruses at molecular detail. To achieve this, the aqueous sample is rapidly vitrified in liquid ethane, which preserves it in a close-to-native, frozen-hydrated state. In the electron microscope, tilt series are recorded at liquid nitrogen temperature, from which 3D tomograms are reconstructed. The signal-to-noise ratio of the tomographic volume is inherently low. Recognizable, recurring features are enhanced by subtomogram averaging, by which individual subvolumes are cut out, aligned and averaged to reduce noise. In this way, 3D maps with a resolution of 2 nm or better can be obtained. A fit of available high-resolution structures to the 3D volume then produces atomic models of protein complexes in their native environment. Here we show how we use electron cryo-tomography to study the in situ organization of large membrane protein complexes in mitochondria. We find that ATP synthases are organized in rows of dimers along highly curved apices of the inner membrane cristae, whereas complex I is randomly distributed in the membrane regions on either side of the rows. By subtomogram averaging we obtained a structure of the mitochondrial ATP synthase dimer within the cristae membrane.     

Evaluation of denoising algorithms for biological electron tomography

Tomograms of biological specimens derived using transmission electron microscopy can be intrinsically noisy due to the use of low electron doses, the presence of a "missing wedge" in most data collection schemes, and inaccuracies arising during 3D volume reconstruction. Before tomograms can be interpreted reliably, for example, by 3D segmentation, it is essential that the data be suitably denoised using procedures that can be individually optimized for specific data sets. Here, we implement a systematic procedure to compare various nonlinear denoising techniques on tomograms recorded at room temperature and at cryogenic temperatures, and establish quantitative criteria to select a denoising approach that is most relevant for a given tomogram. We demonstrate that using an appropriate denoising algorithm facilitates robust segmentation of tomograms of HIV-infected macrophages and Bdellovibrio bacteria obtained from specimens at room and cryogenic temperatures, respectively. We validate this strategy of automated segmentation of optimally denoised tomograms by comparing its performance with manual extraction of key features from the same tomograms.     

A multi-model approach to simultaneous segmentation and classification of heterogeneous populations of cell nuclei in 3D confocal microscope images.

Automated segmentation and morphometry of fluorescently labeled cell nuclei in batches of 3D confocal stacks is essential for quantitative studies. Model-based segmentation algorithms are attractive due to their robustness. Previous methods incorporated a single nuclear model. This is a limitation for tissues containing multiple cell types with different nuclear features. Improved segmentation for such tissues requires algorithms that permit multiple models to be used simultaneously. This requires a tight integration of classification and segmentation algorithms. Two or more nuclear models are constructed semiautomatically from user-provided training examples. Starting with an initial over-segmentation produced by a gradient-weighted watershed algorithm, a hierarchical fragment merging tree rooted at each object is built. Linear discriminant analysis is used to classify each candidate using multiple object models. On the basis of the selected class, a Bayesian score is computed. Fragment merging decisions are made by comparing the score with that of other candidates, and the scores of constituent fragments of each candidate. The overall segmentation accuracy was 93.7% and classification accuracy was 93.5%, respectively, on a diverse collection of images drawn from five different regions of the rat brain. The multi-model method was found to achieve high accuracy on nuclear segmentation and classification by correctly resolving ambiguities in clustered regions containing heterogeneous cell populations.     

Mapping 70S ribosomes in intact cells by cryoelectron tomography and pattern recognition

Cryoelectron tomography (CET) combines the potential of three-dimensional (3D) imaging with a close-to-life preservation of biological samples. It allows the examination of large and stochastically variable structures, such as organelles or whole cells. At the current resolution it becomes possible to visualize large macromolecular complexes in their functional cellular environments. Pattern recognition methods can be used for a systematic interpretation of the tomograms; target molecules are identified and located based on their structural signature and their correspondence with a template. Here, we demonstrate that such an approach can be used to map 70S ribosomes in an intact prokaryotic cell (Spiroplasma melliferum) with high fidelity, in spite of the low signal-to-noise ratio (SNR) of the tomograms. At a resolution of 4.7 nm the average generated from the 236 ribosomes found in a tomogram is in good agreement with high resolution structures of isolated ribosomes as obtained by X-ray crystallography or cryoelectron microscopy. Under the conditions of the experiment (logarithmic growth phase) the ribosomes are evenly distributed throughout the cytosol, occupying approximately 5% of the cellular volume. A subset of about 15% is found in close proximity to and with a distinct orientation with respect to the plasma membrane. This study represents a first step towards generating a more comprehensive cellular atlas of macromolecular complexes.     

Discovering simple regions in biological sequences associated with scoring schemes.

Let A denote an alphabet consisting of n types of letters. Given a sequence S of length L with v(i) letters of type i on A, to describe the compositional properties and combinatorial structure of S, we propose a new complexity function of S, called the reciprocal complexity of S, as C(S) = (i=1) product operator (n) (L/nv(i))(vi) Based on this complexity measure, an efficient algorithm is developed for classifying and analyzing simple segments of protein and nucleotide sequence databases associated with scoring schemes. The running time of the algorithm is nearly proportional to the sequence length. The program DSR corresponding to the algorithm was written in C++, associated with two parameters (window length and cutoff value) and a scoring matrix. Some examples regarding protein sequences illustrate how the method can be used to find regions. The first application of DSR is the masking of simple sequences for searching databases. Queries masked by DSR returned a manageable set of hits below the E-value cutoff score, which contained all true positive homologues. The second application is to study simple regions detected by the DSR program corresponding to known structural features of proteins. An extensive computational analysis has been made of protein sequences with known, physicochemically defined nonglobular segments. For the SWISS-PROT amino acid sequence database (Release 40.2 of 02-Nov-2001), we determine that the best parameters and the best BLOSUM matrix are, respectively, for automatic segmentation of amino acid sequences into nonglobular and globular regions by the DSR program: Window length k = 35, cutoff value b = 0.46, and the BLOSUM 62.5 matrix. The average "agreement accuracy (sensitivity)" of DSR segmentation for the SWISS-PROT database is 97.3%.     

A watershed approach for improving medical image segmentation

In this paper, a novel watershed approach based on seed region growing and image entropy is presented which could improve the medical image segmentation. The proposed algorithm enables the prior information of seed region growing and image entropy in its calculation. The algorithm starts by partitioning the image into several levels of intensity using watershed multi-degree immersion process. The levels of intensity are the input to a computationally efficient seed region segmentation process which produces the initial partitioning of the image regions. These regions are fed to entropy procedure to carry out a suitable merging which produces the final segmentation. The latter process uses a region-based similarity representation of the image regions to decide whether regions can be merged. The region is isolated from the level and the residual pixels are uploaded to the next level and so on, we recall this process as multi-level process and the watershed is called multi-level watershed. The proposed algorithm is applied to challenging applications: grey matter–white matter segmentation in magnetic resonance images (MRIs). The established methods and the proposed approach are experimented by these applications to a variety of simulating immersion, multi-degree, multi-level seed region growing and multi-level seed region growing with entropy. It is shown that the proposed method achieves more accurate results for medical image oversegmentation.     

A resolution criterion for electron tomography based on cross-validation

Despite much progress in electron tomography, quantitative assessment of resolution has remained a problematic issue. The criteria that are used in single particle analysis, based on gauging the consistency between density maps calculated from half data sets, are not directly applicable because of the uniqueness of a tomographic volume. Here, we propose two criteria based on a cross-validation approach. One, called FSC(e/o), is based on a Fourier shell correlation comparison between tomograms calculated from the even and odd members of a tilt series. The other, called noise-compensated leave-one-out (NLOO), is based on Fourier ring correlation comparisons between an original projection and the corresponding reprojection of the tomogram calculated from all the other projections, taking into account the differing noise statistics. Plotted as a function of tilt angle, they allow assessment of the angular dependence of resolution and quality control over the series of projections. Integrated over all projections, the results give a global figure for resolution. Tests on simulated tomograms established consistency between these criteria and the FSC(ref), a correlation coefficient calculated between a known reference structure and the corresponding portion of a tomogram containing that structure. The two criteria-FSC(e/o) and NLOO-are mutually consistent when residual noise is the major resolution-limiting factor. When the size of the tilt increment becomes a significant factor, NLOO provides a more reliable criterion, as expected, although it is computationally intensive. Applicable to entire tomograms or selected structures, NLOO has also been tested on experimental tomographic data.     

Morphometrical and immunocytochemical characterization of peri-insular and tele-insular acinar cells in the rat pancreas

Morphometrical and immunocytochemical techniques have been applied in order to characterize the pancreatic acinar cells located in peri-insular and tele-insular regions of the pancreas. The results obtained, have shown that the acinar cells of the peri-insular regions are twice as large as those of the tele-insular. On the other hand, the volume density of all organelles, except that of the zymogen granules, remains constant implying that the larger the cell, the larger are its organelles. For the zymogen granules however, their volume density was found to be higher in peri-insular acinar cells. The immunofluorescence technique applied for the demonstration of amylase and chymotrypsinogen has confirmed the presence of an inhomogeneity in the staining. Acinar cells in peri-insular regions show a brighter fluorescent staining. At the electron microscope level, both amylase and chymotrypsinogen were demonstrated in all organelles of acinar cells involved in protein secretion. Quantitative evaluations demonstrate no major differences in the intensity of labeling per micron2 between organelles of peri-insular and tele-insular cells. These results put together demonstrate that peri-insular acinar cells contain higher amounts of secretory proteins because their organelles are larger and their zymogen granules are more numerous. The partition of the exocrine pancreas into peri- and tele-insular regions, confirmed herein through morphometrical and cytochemical techniques, is discussed in relation to the possible influence of the endocrine secretion arising from the islets of Langerhans on the surrounding acinar cells.     

A procedure to deposit fiducial markers on vitreous cryo-sections for cellular tomography

We describe a novel approach for the accurate alignment of images in electron tomography of vitreous cryo-sections. Quantum dots, suspended in organic solvents at cryo-temperatures, are applied directly onto the sections and are subsequently used as fiducial markers to align the tilt series. Data collection can be performed from different regions of the vitreous sections, even when the sections touch the grid only at a few places. We present high-resolution tomograms of some organelles in cryo-sections of human skin cells using this method. The average error in image alignment was about 1nm and the resolution was estimated to be 5-7nm. Thus, the use of section-attached quantum dots as fiducial markers in electron tomography of vitreous cryo-sections facilitates high-resolution in situ 3D imaging of organelles and macromolecular complexes in their native hydrated state.     

Mean curvature motion facilitates the segmentation and surface visualization of electron tomograms

Three-dimensional visualization of biological samples is essential for understanding their architecture and function. However, it is often challenging due to the macromolecular crowdedness of the samples and low signal-to-noise ratio of the cryo-electron tomograms. Denoising and segmentation techniques address this challenge by increasing the signal-to-noise ratio and by simplifying the data in images. Here, mean curvature motion is presented as a method that can be applied to segmentation results, created either manually or automatically, to significantly improve both the visual quality and downstream computational handling. Mean curvature motion is a process based on nonlinear anisotropic diffusion that smooths along edges and causes high-curvature features, such as noise, to disappear. In combination with level-set methods for image erosion and dilation, the application of mean curvature motion to electron tomograms and segmentations removes sharp edges or spikes in the visualized surfaces, produces an improved surface quality, and improves overall visualization and interpretation of the three-dimensional images.     

LigBase: a database of families of aligned ligand binding sites in known protein sequences and structures

A database comprising all ligand-binding sites of known structure aligned with all related protein sequences and structures is described. Currently, the database contains approximately 50000 ligand-binding sites for small molecules found in the Protein Data Bank (PDB). The structure-structure alignments are obtained by the Combinatorial Extension (CE) program (Shindyalov and Bourne, Protein Eng., 11, 739-747, 1998) and sequence-structure alignments are extracted from the ModBase database of comparative protein structure models for all known protein sequences (Sanchez et al., Nucleic Acids Res., 28, 250-253, 2000). It is possible to search for binding sites in LigBase by a variety of criteria. LigBase reports summarize ligand data including relevant structural information from the PDB file, such as ligand type and size, and contain links to all related protein sequences in the TrEMBL database. Residues in the binding sites are graphically depicted for comparison with other structurally defined family members. LigBase provides a resource for the analysis of families of related binding sites.     

WatershedCounting3D: a new method for segmenting and counting punctate structures from confocal image data

Current research in cell biology frequently uses light microscopy to study intracellular organelles. To segment and count organelles, most investigators have used a global thresholding method, which relies on homogeneous background intensity values within a cell. Because this is not always the case, we developed WatershedCounting3D, a program that uses a modified watershed algorithm to more accurately identify intracellular structures from confocal image data, even in the presence of an inhomogeneous background. We give examples of segmenting and counting endoplasmic reticulum exit sites and the Golgi apparatus.     

Automatic liver segmentation in MRI images using an iterative watershed algorithm and artificial neural network

Precise liver segmentation in abdominal MRI images is one of the most important steps for the computer-aided diagnosis of liver pathology. The first and essential step for diagnosis is automatic liver segmentation, and this process remains challenging. Extensive research has examined liver segmentation; however, it is challenging to distinguish which algorithm produces more precise segmentation results that are applicable to various medical imaging techniques. In this paper, we present a new automatic system for liver segmentation in abdominal MRI images. The system includes several successive steps. Preprocessing is applied to enhance the image (edge-preserved noise reduction) by using mathematical morphology. The proposed algorithm for liver region extraction is a combined algorithm that utilizes MLP neural networks and watershed algorithm. The traditional watershed transformation generally results in oversegmentation when directly applied to medical image segmentation. Therefore, we use trained neural networks to extract features of the liver region. The extracted features are used to monitor the quality of the segmentation using the watershed transform and adjust the required parameters automatically. The process of adjusting parameters is performed sequentially in several iterations. The proposed algorithm extracts liver region in one slice of the MRI images and the boundary tracking algorithm is suggested to extract the liver region in other slices, which is left as our future work. This system was applied to a series of test images to extract the liver region. Experimental results showed positive results for the proposed algorithm.     

Development of a visualization platform oriented to Lake water quality targets management - A case study of Lake Taihu

Lake Taihu is well known for its severe environmental degradation. In previous studies of lake quality target management, the water quality targets were poorly correlated with watershed pollutant reduction, and most studies lacked visualized management platform that covered all elements including lakes, in-lake estuaries, rivers and watershed regions. In this study, a browser/server-based visualization platform for lake quality target management was developed. Five models that covered both the watershed and lake scales were integrated based on two critical functions. First, the proposed method can be used to determine watershed pollutant reduction amounts based on certain lake quality target parameters, such as those for TN, TP, NH₃N and COD. Second, the method can simulate the lake quality trends associated with different watershed adjustment plans. The platform was deployed by the Taihu Basin Authority (TBA) of the Ministry of Water Resources. Overall, this platform is a useful tool for watershed-lake environmental management.     

基于多尺度标记优化分水岭方法的油茶冠幅提取

针对尺度对地物空间结构的限制以及传统分水岭分割易产生树冠过分割等问题,选择长沙县明月村油茶基地为研究区,提出一种基于多尺度标记优化分水岭分割油茶树冠的方法。首先使用高分辨率无人机影像采集图像,分析影像特征,构建油茶分类体系,提取油茶林分布区域。其次,运用多尺度区域迭代增长方法提取树冠标记,将标记应用于多阈值尺度的分水岭变换,并结合Johnson指数选取树冠标记增长和分水岭阈值的最优尺度,实现油茶单木的准确识别。结果表明:多尺度标记优化分水岭方法在分离油茶单木时,树冠面积提取值与目视解译参考值的相对误差为9.4%;单木总体识别精度为89.4%,相对于传统的分水岭分割方法精度提高了34.8%;通过Johnson指数确定的最优迭代增长尺度为20,分水岭分割阈值尺度为85,对比不同尺度组合下的油茶冠幅提取结果,最优尺度下的油茶冠幅提取精度最高(R²=0.75)。多尺度标记优化分水岭方法能较准确地分离油茶树冠,将该方法应用于无人机影像树冠分割,可有效提高经济林调查的效率。     

Sheets, ribbons and tubules - how organelles get their shape

Most membrane-bound organelles have elaborate, dynamic shapes and often include regions with distinct morphologies. These complex structures are relatively conserved throughout evolution, which indicates that they are important for optimal organelle function. Various mechanisms of determining organelle shape have been proposed - proteins that stabilize highly curved membranes, the tethering of organelles to other cellular components and the regulation of membrane fission and fusion might all contribute.     

Does sulphide detoxication occur in the gills of the hydrothermal vent shrimp, Rimicaris exoculata?Y a-t-il détoxication des sulfures dans les branchies de la crevette hydrothermale, Rimicaris exoculata ?

Ultrastructural observations of the gills of the hydrothermal vent shrimp Rimicaris exoculata reveal that the epithelial cells contain numerous mitochondria clustered around unusual organelles (diameter of 0.7 to 2.5 microns) containing membrane stacks. These organelles were termed sulphide-oxidising bodies (SOBs) by structural analogy with organelles observed in the tissues of species adapted to sulphide-rich environments. Moreover, in the gills of R. exoculata, mitochondria display numerous electron-dense granules in their stroma. Such ultrastructural features suggest that sulphide detoxication may probably occur in the gills of R. exoculata. Comparable structures were also described in the gills of other hydrothermal vent species, as the alvinellid Pompeii worms that, as R. exoculata, are housing ectosymbiotic bacteria.     

Electron microscopy of high pressure frozen samples: bridging the gap between cellular ultrastructure and atomic resolution

Transmission electron microscopy has provided most of what is known about the ultrastructural organization of tissues, cells, and organelles. Due to tremendous advances in crystallography and magnetic resonance imaging, almost any protein can now be modeled at atomic resolution. To fully understand the workings of biological “nanomachines” it is necessary to obtain images of intact macromolecular assemblies in situ. Although the resolution power of electron microscopes is on the atomic scale, in biological samples artifacts introduced by aldehyde fixation, dehydration and staining, but also section thickness reduces it to some nanometers. Cryofixation by high pressure freezing circumvents many of the artifacts since it allows vitrifying biological samples of about 200 μm in thickness and immobilizes complex macromolecular assemblies in their native state in situ. To exploit the perfect structural preservation of frozen hydrated sections, sophisticated instruments are needed, e.g., high voltage electron microscopes equipped with precise goniometers that work at low temperature and digital cameras of high sensitivity and pixel number. With them, it is possible to generate high resolution tomograms, i.e., 3D views of subcellular structures. This review describes theory and applications of the high pressure cryofixation methodology and compares its results with those of conventional procedures. Moreover, recent findings will be discussed showing that molecular models of proteins can be fitted into depicted organellar ultrastructure of images of frozen hydrated sections. High pressure freezing of tissue is the base which may lead to precise models of macromolecular assemblies in situ, and thus to a better understanding of the function of complex cellular structures.