Super-resolution segmentation of imaging mass spectrometry data: Solving the issue of low lateral resolution

In the last decade, imaging mass spectrometry has seen incredible technological advances in its applications to biological samples. One computational method of data mining in this field is the spatial segmentation of a sample, which produces a segmentation map highlighting chemically similar regions. An important issue for any imaging mass spectrometry technology is its relatively low spatial or lateral resolution (i.e. a large size of pixel) as compared with microscopy. Thus, the spatial resolution of a segmentation map is also relatively low, that complicates its visual examination and interpretation when compared with microscopy data, as well as reduces the accuracy of any automated comparison. We address this issue by proposing an approach to improve the spatial resolution of a segmentation map. Given a segmentation map, our method magnifies it up to some factor, producing a super-resolution segmentation map. The super-resolution map can be overlaid and compared with a high-res microscopy image. The proposed method is based on recent advances in image processing and smoothes the "pixilated" region boundaries while preserving fine details. Moreover, it neither eliminates nor splits any region. We evaluated the proposed super-resolution segmentation approach on three MALDI-imaging datasets of human tissue sections and demonstrated the superiority of the super-segmentation maps over standard segmentation maps.     

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 segmentation/clustering model for the analysis of array CGH data

Microarray-CGH (comparative genomic hybridization) experiments are used to detect and map chromosomal imbalances. A CGH profile can be viewed as a succession of segments that represent homogeneous regions in the genome whose representative sequences share the same relative copy number on average. Segmentation methods constitute a natural framework for the analysis, but they do not provide a biological status for the detected segments. We propose a new model for this segmentation/clustering problem, combining a segmentation model with a mixture model. We present a new hybrid algorithm called dynamic programming-expectation maximization (DP-EM) to estimate the parameters of the model by maximum likelihood. This algorithm combines DP and the EM algorithm. We also propose a model selection heuristic to select the number of clusters and the number of segments. An example of our procedure is presented, based on publicly available data sets. We compare our method to segmentation methods and to hidden Markov models, and we show that the new segmentation/clustering model is a promising alternative that can be applied in the more general context of signal processing.     

Visualization and analysis of 3D microscopic images

In a wide range of biological studies, it is highly desirable to visualize and analyze three-dimensional (3D) microscopic images. In this primer, we first introduce several major methods for visualizing typical 3D images and related multi-scale, multi-time-point, multi-color data sets. Then, we discuss three key categories of image analysis tasks, namely segmentation, registration, and annotation. We demonstrate how to pipeline these visualization and analysis modules using examples of profiling the single-cell gene-expression of C. elegans and constructing a map of stereotyped neurite tracts in a fruit fly brain.     

A two-step procedure for automatic and accurate segmentation of volumetric CLSM biofilm images

This paper presents a robust two-step segmentation procedure for the study of biofilm structure. Without user intervention, the procedure segments volumetric biofilm images generated by a confocal laser scanning microscopy (CLSM). This automated procedure implements an anisotropic diffusion filter as a preprocessing step and a 3D extension of the Otsu method for thresholding. Applying the anisotropic diffusion filter to even low-contrast CLSM images significantly improves the segmentation obtained with the 3D Otsu method. A comparison of the results for several CLSM data sets demonstrated that the accuracy of this procedure, unlike that of the objective threshold selection algorithm (OTS), is not affected by biofilm coverage levels and thus fills an important gap in developing a robust and objective segmenting procedure. The effectiveness of the present segmentation procedure is shown for CLSM images containing different bacterial strains. The image saturation handling capability of this procedure relaxes the constraints on user-selected gain and intensity settings of a CLSM. Therefore, this two-step procedure provides an automatic and accurate segmentation of biofilms that is independent of biofilm coverage levels and, in turn, lays a solid foundation for achieving objective analysis of biofilm structural parameters.     

Automated tracing of filaments in 3D electron tomography reconstructions using Sculptor and Situs

The molecular graphics program Sculptor and the command-line suite Situs are software packages for the integration of biophysical data across spatial resolution scales. Herein, we provide an overview of recently developed tools relevant to cryo-electron tomography (cryo-ET), with an emphasis on functionality supported by Situs 2.7.1 and Sculptor 2.1.1. We describe a work flow for automatically segmenting filaments in cryo-ET maps including denoising, local normalization, feature detection, and tracing. Tomograms of cellular actin networks exhibit both cross-linked and bundled filament densities. Such filamentous regions in cryo-ET data sets can then be segmented using a stochastic template-based search, VolTrac. The approach combines a genetic algorithm and a bidirectional expansion with a tabu search strategy to localize and characterize filamentous regions. The automated filament segmentation by VolTrac compares well to a manual one performed by expert users, and it allows an efficient and reproducible analysis of large data sets. The software is free, open source, and can be used on Linux, Macintosh or Windows computers.     

Visualization of the cell-selective distribution of PUFA-containing phosphatidylcholines in mouse brain by imaging mass spectrometry

Previous studies have shown that MALDI-imaging mass spectrometry (IMS) can be used to visualize the distribution of various biomolecules, especially lipids, in the cells and tissues. In this study, we report the cell-selective distribution of PUFA-containing glycerophospholipids (GPLs) in the mouse brain. We established a practical experimental procedure for the IMS of GPLs. We demonstrated that optimization of the composition of the matrix solution and spectrum normalization to the total ion current (TIC) is critical. Using our procedure, we simultaneously differentiated and visualized the localizations of specific molecular species of GPLs in mouse brain sections. The results showed that PUFA-containing phosphatidylcholines (PCs) were distributed in a cell-selective manner: arachidonic acid- and docosahexaenoic acid-containing PCs were seen in the hippocampal neurons and cerebellar Purkinje cells, respectively. Furthermore, these characteristic localizations of PUFA-PCs were formed during neuronal maturation. The phenomenon of brain cell-selective production of specific PUFA-GPLs will help elucidate the potential physiological functions of PUFAs in specific brain regions.     

Effects of non-local diffusion on structural MRI preprocessing and default network mapping: statistical comparisons with isotropic/anisotropic diffusion

Neuroimaging community usually employs spatial smoothing to denoise magnetic resonance imaging (MRI) data, e.g., Gaussian smoothing kernels. Such an isotropic diffusion (ISD) based smoothing is widely adopted for denoising purpose due to its easy implementation and efficient computation. Beyond these advantages, Gaussian smoothing kernels tend to blur the edges, curvature and texture of images. Researchers have proposed anisotropic diffusion (ASD) and non-local diffusion (NLD) kernels. We recently demonstrated the effect of these new filtering paradigms on preprocessing real degraded MRI images from three individual subjects. Here, to further systematically investigate the effects at a group level, we collected both structural and functional MRI data from 23 participants. We first evaluated the three smoothing strategies' impact on brain extraction, segmentation and registration. Finally, we investigated how they affect subsequent mapping of default network based on resting-state functional MRI (R-fMRI) data. Our findings suggest that NLD-based spatial smoothing maybe more effective and reliable at improving the quality of both MRI data preprocessing and default network mapping. We thus recommend NLD may become a promising method of smoothing structural MRI images of R-fMRI pipeline.     

CODEX: a normalization and copy number variation detection method for whole exome sequencing

High-throughput sequencing of DNA coding regions has become a common way of assaying genomic variation in the study of human diseases. Copy number variation (CNV) is an important type of genomic variation, but detecting and characterizing CNV from exome sequencing is challenging due to the high level of biases and artifacts. We propose CODEX, a normalization and CNV calling procedure for whole exome sequencing data. The Poisson latent factor model in CODEX includes terms that specifically remove biases due to GC content, exon capture and amplification efficiency, and latent systemic artifacts. CODEX also includes a Poisson likelihood-based recursive segmentation procedure that explicitly models the count-based exome sequencing data. CODEX is compared to existing methods on a population analysis of HapMap samples from the 1000 Genomes Project, and shown to be more accurate on three microarray-based validation data sets. We further evaluate performance on 222 neuroblastoma samples with matched normals and focus on a well-studied rare somatic CNV within the ATRX gene. We show that the cross-sample normalization procedure of CODEX removes more noise than normalizing the tumor against the matched normal and that the segmentation procedure performs well in detecting CNVs with nested structures.     

Image-based axon model highlights heterogeneity in initiation of damage

Head injury simulations predict the occurrence of traumatic brain injury by placing a threshold on the calculated strains for axon tracts within the brain. However, a current roadblock to accurate injury prediction is the selection of an appropriate axon damage threshold. While several computational studies have used models of the axon cytoskeleton to investigate damage initiation, these models all employ an idealized, homogeneous axonal geometry. This homogeneous geometry with regularly spaced microtubules, evenly distributed throughout the model, overestimates axon strength because, in reality, the axon cytoskeleton is heterogeneous. In the heterogeneous cytoskeleton, the weakest cross section determines the initiation of failure, but these weak spots are not present in a homogeneous model. Addressing one source of heterogeneity in the axon cytoskeleton, we present a new semiautomated image analysis pipeline for using serial-section transmission electron micrographs to reconstruct the microtubule geometry of an axon. The image analysis procedure locates microtubules within the images, traces them throughout the image stack, and reconstructs the microtubule structure as a finite element mesh. We demonstrate the image analysis approach using a C. elegans touch receptor neuron due to the availability of high-quality serial-section transmission electron micrograph data sets. The results of the analysis highlight the heterogeneity of the microtubule structure in the spatial variation of both microtubule number and length. Simulations comparing this image-based geometry with homogeneous geometries show that structural heterogeneity in the image-based model creates significant spatial variation in deformation. The homogeneous geometries, on the other hand, deform more uniformly. Since no single homogeneous model can replicate the mechanical behavior of the image-based model, our results argue that heterogeneity in axon microtubule geometry should be considered in determining accurate axon failure thresholds.     

Automated segmentation of molecular subunits in electron cryomicroscopy density maps

Electron cryomicroscopy (cryoEM) is capable of imaging large macromolecular machines composed of multiple components. However, it is currently only possible to achieve moderate resolution at which it may be possible to computationally extract the individual components in the machine. In this work, we present application details of an automated method for detecting and segmenting the components of a large machine in an experimentally determined density map. This method is applicable to object with and without symmetry and takes advantage of global and local symmetry axes if present. We have applied this segmentation algorithm to several cryoEM data sets already deposited in EMDB with various complexities, symmetries and resolutions and validated the results using manually segmented density and available structures of the components in the PDB. As such, automated segmentation could become a useful tool for the analysis of the ever-increasing number of structures of macromolecular machines derived from cryoEM.     

The impact of spatial structure on the accuracy of contour maps of small data sets

Spatial analysis of insect counts provides important information about how insect species respond to the heterogeneity of a given sampling space. Contour mapping is widely used to visualize spatial pest distribution patterns in anthropogenic environments, and in this study we outlined recommendations regarding semivariogram analysis of small data sets (N < 50). Second, we examined how contour maps based upon linear kriging were affected by the spatial structure of the given data set, as error estimation of contour maps appears to have received little attention in the entomological domain. We used weekly trap catches of the warehouse beetle, Trogoderma variabile, and the accuracy assessment was based upon data sets that had either a random spatial structure or were characterized by asymptotic spatial dependence. Asymptotic spatial dependence (typically described with a semivariogram analysis) means that trap catches at locations close to each other are more similar than trap catches at locations further apart. Trap catches were poorly predicted for data sets with a random spatial structure, while there was a significant correlation between observed and predicted trap catches for the spatially rearranged data sets. Therefore, for data sets with a random spatial structure we recommend visualization of the insect counts as scale-sized dots rather than as contour maps.     

A Probabilistic Atlas of Diffuse WHO Grade II Glioma Locations in the Brain

Diffuse WHO grade II gliomas are diffusively infiltrative brain tumors characterized by an unavoidable anaplastic transformation. Their management is strongly dependent on their location in the brain due to interactions with functional regions and potential differences in molecular biology. In this paper, we present the construction of a probabilistic atlas mapping the preferential locations of diffuse WHO grade II gliomas in the brain. This is carried out through a sparse graph whose nodes correspond to clusters of tumors clustered together based on their spatial proximity. The interest of such an atlas is illustrated via two applications. The first one correlates tumor location with the patient’s age via a statistical analysis, highlighting the interest of the atlas for studying the origins and behavior of the tumors. The second exploits the fact that the tumors have preferential locations for automatic segmentation. Through a coupled decomposed Markov Random Field model, the atlas guides the segmentation process, and characterizes which preferential location the tumor belongs to and consequently which behavior it could be associated to. Leave-one-out cross validation experiments on a large database highlight the robustness of the graph, and yield promising segmentation results.     

Robust smooth segmentation approach for array CGH data analysis

MOTIVATION: Array comparative genomic hybridization (aCGH) provides a genome-wide technique to screen for copy number alteration. The existing segmentation approaches for analyzing aCGH data are based on modeling data as a series of discrete segments with unknown boundaries and unknown heights. Although the biological process of copy number alteration is discrete, in reality a variety of biological and experimental factors can cause the signal to deviate from a stepwise function. To take this into account, we propose a smooth segmentation (smoothseg) approach. METHODS: To achieve a robust segmentation, we use a doubly heavy-tailed random-effect model. The first heavy-tailed structure on the errors deals with outliers in the observations, and the second deals with possible jumps in the underlying pattern associated with different segments. We develop a fast and reliable computational procedure based on the iterative weighted least-squares algorithm with band-limited matrix inversion. RESULTS: Using simulated and real data sets, we demonstrate how smoothseg can aid in identification of regions with genomic alteration and in classification of samples. For the real data sets, smoothseg leads to smaller false discovery rate and classification error rate than the circular binary segmentation (CBS) algorithm. In a realistic simulation setting, smoothseg is better than wavelet smoothing and CBS in identification of regions with genomic alterations and better than CBS in classification of samples. For comparative analyses, we demonstrate that segmenting the t-statistics performs better than segmenting the data. AVAILABILITY: The R package smoothseg to perform smooth segmentation is available from http://www.meb.ki.se/~yudpaw.     

Live Imaging of Mitosis in the Developing Mouse Embryonic Cortex

Although of short duration, mitosis is a complex and dynamic multi-step process fundamental for development of organs including the brain. In the developing cerebral cortex, abnormal mitosis of neural progenitors can cause defects in brain size and function. Hence, there is a critical need for tools to understand the mechanisms of neural progenitor mitosis. Cortical development in rodents is an outstanding model for studying this process. Neural progenitor mitosis is commonly examined in fixed brain sections. This protocol will describe in detail an approach for live imaging of mitosis in ex vivo embryonic brain slices. We will describe the critical steps for this procedure, which include: brain extraction, brain embedding, vibratome sectioning of brain slices, staining and culturing of slices, and time-lapse imaging. We will then demonstrate and describe in detail how to perform post-acquisition analysis of mitosis. We include representative results from this assay using the vital dye Syto11, transgenic mice (histone H2B-EGFP and centrin-EGFP), and in utero electroporation (mCherry-α-tubulin). We will discuss how this procedure can be best optimized and how it can be modified for study of genetic regulation of mitosis. Live imaging of mitosis in brain slices is a flexible approach to assess the impact of age, anatomy, and genetic perturbation in a controlled environment, and to generate a large amount of data with high temporal and spatial resolution. Hence this protocol will complement existing tools for analysis of neural progenitor mitosis.     

High resolution spatial mapping of brominated pyrrole-2-aminoimidazole alkaloids distributions in the marine sponge Stylissa flabellata via MALDI-mass spectrometry imaging

A number of pharmacologically active brominated pyrrole-2-aminoimidazole (B-P-2-AI) alkaloids have been isolated from several families of marine sponges, including those belonging to the genus Stylissa. In the present study, MALDI mass spectrometry imaging (MALDI-imaging) was applied to determine the spatial distribution of B-P-2-AIs within 20 μm cross sections of S. flabellata. A number of previously characterised B-P-2-AIs were readily identified by MALDI-imaging and confirmed by MS-MS and NMR profiling. Unknown B-P-2-AIs were also observed. Discrete microchemical environments were revealed for several B-P-2-AIs including dibromophakellin which was localised within the external pinacoderm and internal network of choanoderm chambers. Additionally, dibromopalau'amine and konbu'acidin B were also found to be confined to the choanoderm, while sceptrin was found to be highly abundant within the mesohyl. Further brominated compounds of unknown structure were also observed to have distinct localisation in both choanoderm chambers and the pinacoderm. These findings provide insights into the chemical ecology of S. flabellata, as most B-P-2-AIs were found on highly exposed surfaces, where they may act to prevent pathogens, predation and/or biofouling. Moreover this study demonstrates the power of MALDI-imaging to visualise the location of a range of metabolites in situ and to characterise compounds by MS-MS directly from intact specimens without the need for extraction. These methodologies facilitate selective targeting of micro-regions of sponge to screen for symbiotic microbial candidates or genes that may be involved in the production of the correlated compounds, and may represent a change in paradigm for natural product drug development.     

Transferability of object-based rule sets for mapping coastal high marsh habitat among different regions in Georgian Bay,Canada

Coastal wetlands of eastern and northern Georgian Bay, Canada provide critical habitat for a variety of biota yet few have been delineated and mapped because of their widespread distribution and remoteness. This is an impediment to conservation efforts aimed at identifying significant habitat in the Laurentian Great Lakes. We propose to address this deficiency by developing an approach that relies on use of high-resolution remote sensing imagery to map wetland habitat. In this study, we use IKONOS satellite imagery to classify coastal high marsh vegetation (seasonally inundated) and assess the transferability of object-based rule sets among different regions in eastern Georgian Bay. We classified 24 wetlands in three separate satellite scenes and developed an object-based approach to map four habitat classes: emergent, meadow/shrub, senescent vegetation and rock. Independent rule sets were created for each scene and applied to the other images to empirically examine transferability at broad spatial scales. For a given habitat feature, the internally derived rule sets based on field data collected from the same scene provided significantly greater accuracy than those derived from a different scene (80.0 and 74.3%, respectively). Although we present a significant effect of ruleset origin on accuracy, the difference in accuracy is minimal at 5.7%. We argue that this should not detract from its transferability on a regional scale. We conclude that locally derived and object-based rule sets developed from IKONOS imagery can successfully classify complex vegetation classes and be applied to different regions without much loss of accuracy. This indicates that large–scale mapping automation may be feasible with images with similar spectral, spatial, contextual, and textural properties.     

A methodology for registration of a histological slide and in vivo MRI volume based on optimizing mutual information

We present a method for registering histology and in vivo imaging that requires minimal microtoming and is automatic following the user's initialization. In this demonstration, we register a single hematoxylin-and-eosin-stained histological slide of a coronal section of a rat brain harboring a 9L gliosarcoma with an in vivo 7T MR image volume of the same brain. Because the spatial resolution of the in vivo MRI is limited, we add the step of obtaining a high spatial resolution, ex vivo MRI in situ for intermediate registration. The approach taken was to maximize mutual information in order to optimize the registration between all pairings of image data whether the sources are MRI, tissue block photograph, or stained sample photograph. The warping interpolant used was thin plate splines with the appropriate basis function for either 2-D or 3-D applications. All registrations were implemented by user initialization of the approximate pose between the two data sets, followed by automatic optimization based on maximizing mutual information. Only the higher quality anatomical images were used in the registration process; however, the spatial transformation was directly applied to a quantitative diffusion image. Quantitative diffusion maps from the registered location appeared highly correlated with the H&E slide. Overall, this approach provides a robust method for coregistration of in vivo images with histological sections and will have broad applications in the field of functional and molecular imaging.     

Quantitative Tumor Segmentation for Evaluation of Extent of Glioblastoma Resection to Facilitate Multisite Clinical Trials

Standard-of-care therapy for glioblastomas, the most common and aggressive primary adult brain neoplasm, is maximal safe resection, followed by radiation and chemotherapy. Because maximizing resection may be beneficial for these patients, improving tumor extent of resection (EOR) with methods such as intraoperative 5-aminolevulinic acid fluorescence-guided surgery (FGS) is currently under evaluation. However, it is difficult to reproducibly judge EOR in these studies due to the lack of reliable tumor segmentation methods, especially for postoperative magnetic resonance imaging (MRI) scans. Therefore, a reliable, easily distributable segmentation method is needed to permit valid comparison, especially across multiple sites. We report a segmentation method that combines versatile region-of-interest blob generation with automated clustering methods. We applied this to glioblastoma cases undergoing FGS and matched controls to illustrate the method's reliability and accuracy. Agreement and interrater variability between segmentations were assessed using the concordance correlation coefficient, and spatial accuracy was determined using the Dice similarity index and mean Euclidean distance. Fuzzy C-means clustering with three classes was the best performing method, generating volumes with high agreement with manual contouring and high interrater agreement preoperatively and postoperatively. The proposed segmentation method allows tumor volume measurements of contrast-enhanced T₁-weighted images in the unbiased, reproducible fashion necessary for quantifying EOR in multicenter trials.     

Analysis of genet size and local gene flow in the ectomycorrhizal basidiomycete Suillus spraguei (synonym S. pictus)

Several recent fine-scale genetic structure studies of ectomycorrhizal fungi have reported significant spatial clustering of genets with similar genotypes, supporting locally restricted gene flow. In this study we used genotype data from microsatellite markers and spatial autocorrelation analysis to examine local gene flow in Suillus spraguei at distances up to 2 km. Previously developed microsatellite markers for S. spraguei from Japan were unsuccessful at amplifying DNA isolated from sporocarps found in New York state, and other research suggested that both are disjunct species. Novel microsatellite markers therefore were developed with New York specimens. We identified nine polymorphic microsatellite loci and developed primer sets to amplify these regions. We tested the efficiency of the primer sets on 50 sporocarps collected from a natural Pinus strobus stand. The majority of the markers were in Hardy-Weinberg and linkage equilibrium. The location of all sampled sporocarps was recorded and used along with multilocus genotype data to create a genet map. The distance between sporocarps with the same multilocus genotype was small (≤ 7.65 m) and the majority of sporocarps collected were genetically unique, suggesting frequent spore establishment and sexual recombination on this site. Spatial autocorrelation analysis did not support clustering of similar genotypes, suggesting few restrictions to gene flow within this local population.