Dazzo, F.B., J. Liu, G. Tang, C. Gross, C. Reddy, C. Monosmith, G. Zhu, J. Wang, M. Li, N. Philips, A. Baruti, I. Leader, S. Zamani, R. Verhelst, C. Radek, K. Klemmer, Zhou Ji, P. Smith, S. Kneeshaw and I. Ganesan, M.
Thanyakarn, R. Chandler, C. Hagen, I. Folland, S. Xia, M. Cavanaugh, A. Rosaen, K. Ogbenna, D.
McGarrell, R. Hollingsworth, A. Smucker, S. Nakano, A. Squartini, P. Mateos, S. Gantner, Y. Yanni.
Presented at the All Scientist Meeting (2012-03-15 to 2012-03-16 )
A major challenge in microbial ecology is to develop reliable methods of computer-assisted microscopy that can process and analyze complex digital images of microbial populations and communities at single cell resolution, and compute useful ecological characteristics of their organization and structure in situ without cultivation. To address this challenge, our team of microbiologists, mathematicians and computer scientists has been developing image analysis software that can extract the full information content in digital images of actively growing microbial populations and communities. The long-range mission of this project is to develop and release a comprehensive, innovative suite of bioimage informatics analysis software applications designed to strengthen quantitative, microscopy-based approaches for understanding microbial ecology, at various spatial scales relevant to the individual microbes and their ecological niches. Our software package, called CMEIAS (Center for Microbial Ecology Image Analysis System) consists of new and improved technology computing tools for various stages of image bioinformatics, including image acquisition, processing and segmentation, object analysis and classification, data processing, statistical analysis and exploratory data mining. When finalized, the various software applications and their documentations (refereed journal publications, thoroughly illustrated user manuals, help topic search files, audio-visual training tutorials with accompanying test images) are released as free downloads at our MSU CMEIAS website http://cme.msu.edu/cmeias.
The first release version of CMEIAS (v. 1.27) operates within UTHSCSA ImageTool v. 1.27 in a PC running Windows 32-bit operating system. It can analyze foreground objects in images, reports their numerical abundance and 24 attributes of size, shape and luminosity, provides a 1-dimensional object classifier using a single, user-specified measurement attribute and ≤ 15 upper bin limits (≤ 16 classes), and also features a unique semi-automatic supervised hierarchical tree classifier of all major (cocci, spirals, curved rods, regular rods, prosthecates, branched and unbranched filaments) and several minor (U-rods, ellipsoids, clubs and rudimentary branched rods) microbial morphotypes, equivalent to 98% of the genera described in Bergey’s Manual 9th ed. The CMEIAS morphotype classifier works in 14¬dimensional space, measures changes in abundance of each microbial morphotype present within dynamically growing communities at frequencies down to 0.1% – 0.01% [depending on the sample size] with 97% accuracy, and can add up to 5 additional (rare) morphotypes if present in the community.
Several major upgrades in CMEIAS software applications are at various stages of development. At the core of the suite is a CMEIAS-IT ver. 3.10 upgrade of the UTHSCSA ImageTool host program with numerous revisions in displayed text that improve its user-friendly operation, 6 new toolbar shortcuts for repetitive tasks, a minimum – maximum object size filter, expanded image zoom features, upgrades of several manual object analysis routines, expansion of the range of upper bin limits for the 1-dimensional object classifier, and an expanded help menu. CMEIAS-IT ver. 3.10 operates in 32-bit and 64-bit operating systems.
Image processing routines featured in ImageTool include contrast / brightness adjustment, sharpening, background subtraction, median smoothing, color-to-grayscale conversion, stack averaging, histogram stretch, neighborhood convolution, and a manual pixel editor. The CMEIAS-IT 3.10 upgrade includes 3 image-editing tools to facilitate the segmentation of objects in images prior to analysis. First, a script of CMEIAS Actions operating in Adobe Photoshop / Image Processing Tool Kit has been prepared to semi-automate the editing of foreground and background pixels in brightfield, phase contrast, fluorescence and SEM images of microbial communities in various habitats. Second, a stand-alone CMEIAS Color Segmentation software application has been developed with improved technologies of image processing to facilitate the accurate segmentation of foreground microbial objects within complex RGB digital images where color differentiation really matters most. This application is designed to assist studies that use color classifications to reveal important quantitative information on the ecology of microorganisms at single-cell resolution, e.g., understanding bacterial individuality to explore the mechanisms through which ecological systems work, how individual cells interact with each other and their environment, and tests of the emerging theory of individual-based modeling and ecology which predict that individual cell variation is a major driver of population structure and function. It features a flexible, user-defined color tolerance setting to interactively sample a training set of local color pixels representing the microbes of interest, finds each cell’s boundary, and finally creates a new RGB segmented output image containing these foreground colored microbes in a noise-free background. Its performance has been extensively tested on many environmentally complex color images analyzed pixel-by-pixel one at a time, and operates with ~99% accuracy. This software application also features other image processing routines to fill vacant holes within objects, remove pixel noise by dilate/erode, split images into their RGB/HSI/YUV color model channels, pseudocolor object features differing in brightness to assist in defining their contour (especially important in fluorescence microscopy), and set thresholds for object size and brightness filtering. Third, a CMEIAS Object Separation plugin has been developed to automatically split touching objects (microbial cells) in thresholded images. It performs with 95.8% accuracy and has a unique design that avoids the deletion of pixels resulting in erroneous size reduction of the segmented objects, unlike other automated object separation tools (e.g., watershed) featured in other image processing software.
The digital image analysis cores of the CMEIAS-IT 3.10 upgrade are designed to extract 5 categories of information important to microbial ecology. These include features to analyze: (i) microbial abundance, reporting cell counts, spatial density, % cover, dilution adjusted concentration, filamentous length, biovolume, biomass C, and biosurface area of each individual cell, each individual population of morphotypes or the entire community; (ii) microbial eco-physiology differentiated by size / shape / color / microdensitometry, e.g., allometric scaling / metabolic rate, viability, resource ecology, productivity, cell communication and other cell-cell interactions; (iii) microbial autecology / phylogeny using fluorescent molecular probes (e.g., immunofluorescence, FISH) or autofluorescent cell components (e.g., GFP, F420) and color image segmentation especially useful in studies of gene expression in situ; (iv) microbial in situ spatial pattern analysis, reporting information on local spatial density, nearest neighbor relationships, colonization behavior, biogeography, and spatial autocorrelation; and (v) microbial morphology, reporting information on community diversity, morphological adaptations, bacteriovory, and local community dynamics of succession and stability in response to environmental perturbations.
The object analysis core of the CMEIAS-IT v3.10 upgrade can currently extract 77 different measurement attributes of cell size, shape, luminance, and spatial distribution from each individual foreground object and 39 attributes of cumulative object analysis reported per quadrat of image landscape.
Notable among the measurement attributes of CMEIAS Object Analysis are 8 biovolume formulas to measure this cell abundance parameter accurately for each microbial morphotype classified by CMEIAS, attributes that employ user-specified variables in a universal allometric scaling formula to compute biomass carbon from biovolume data, attributes of luminosity for in situ microdensitometry studies of local biofilm texture, gene expression and metabolic activity, and attributes for spatial distribution analysis.
A major research activity has been to rank the ability of CMEIAS measurement attributes to discriminate biofilm architecture at multiple spatial scales. CMEIAS metrics that exhibit high discriminating power in 2¬dimension biofilm analysis include % Substratum Coverage, Areal Porosity, Mean Diffusion Radius, Maximum Diffusion Radius, Cumulative Length, Fiber length, and Perimeter/Area patch shape. Also, CMEIAS luminosity metrics can provide discriminating information on 3-dimensional biofilm architecture, including information to explore discriminating features provided in surface texture analysis.
The CMEIAS Spatial Analysis module is designed to analyze microbial biogeography across multiple spatial scales, including their in situ spatial patterns of colonization on surfaces. These analyses include plot-less point pattern, plot-based quadrat-lattice and geostatistical analyses of the null hypothesis of spatial randomness, from which their colonization behavior can be deduced. Object analysis attributes can report the location of each individual microbial object (Cartesian X, Y coordinates relative to a landmark origin), shortest linear distances to their 1st and 2nd nearest neighbors, cumulative empirical distribution of the 1st nearest neighbor distance, and a Cluster Index that reports the intensity to which each cell is clustered with close neighbors. Cumulative Object Analysis attributes provide quadrat-based metrics of local spatial density using area-weighted abundance of cell concentration / cumulative length / biovolume / biomass C / biosurface area, random point to nearest neighbor for spatial pattern analysis, and postings of the quadrat X, Y centroid and its geometric centroid weighted by local object density. CMEIAS georeferenced data can be analyzed geostatistically to test for and mathematically model spatially autocorrelated intensities of the selected Z-variate for each cell, test for anisotropy in their spatial autocorrelation, measure the range of effective separation distance that defines the spatial scale of influence in cell-cell interactions, and produce statistically defendable 2-D and 3-D kriging maps of the heterogeneity in Z-variate intensity interpolated over the entire georeferenced spatial domain, even in areas not sampled.
The CMEIAS Spatial Analysis module is supported by an AreaPolygonAOI script for plot-based spatial analysis of foreground objects enclosed within a user-defined polygon area of interest instead of the entire image. Also, a stand-alone CMEIAS Quadrat Maker software application has been developed to facilitate the optimization of grid lattice dimensions that divide an image into smaller, constant size quadrats for high-resolution spatial resolution of local microbial density, produces an optimized / indexed image with identification labels of each quadrat in the spatial domain, and then cuts a copy of the original image into size-optimized individual quadrat images with column-row numbered annotation defined by the optimized grid that are ready for image stack construction followed by plot-based and geostatistical spatial distribution analyses.
A powerful CMEIAS JFrad software program is being developed to compute 11 discriminating fractal dimensions along coastlines of microbiofilm colonies in single or batch image inputs. The CMEIAS outputs of biofilm fractal geometry provides quantitative insights into aggregated patterns resulting from the scale-dependent heterogeneous fractal variability in limiting resource partitioning, and reflects the high efficiency at which cells position themselves when faced with the interactive forces of microbial coexistence to optimize their allocation of nutrient resources on a local scale.
The CMEIAS Object Classification module features 5 different object classifiers: an IT-Cmeias 1¬D(imensional) classifier, a Cmeias Morphotype Classifier, a Cmeias OMU classifier, an Individual Morphotype Classifier with Object Analysis, and an Individual OMU Classifier with Object Analysis. The 1¬dimensional classifier sort objects based on division of a scale defined by a single object analysis feature of size, shape, luminosity or spatial proximity. Up to 20 upper class limits can be used to define the decision boundaries separating each bin class. The CMEIAS Morphotype classifier works in 14¬dimensional space and retains the same design as described above for ver. 1.27. The CMEIAS Operational Morphological Unit (OMU) Classifier sub-classifies each population of morphotypes in the community based on their morphological signature defined at 0.1 um resolution by a multilinear matrix of upper bin limits optimized as the least-overlapping borders that separate clusters for each measurement attribute used in the classification scheme. These OMU classification assignments are defined by 2 size border files; the default size border file is derived from hi-low cluster analyses of microbial size data in Bergey’s knowledgebase of taxonomically described microorganisms, and the user-defined size border file is optimized for classification of all OMU’s present in the specific community under investigation. A powerful CMEIAS Size Border Cluster Analysis tool has been developed to optimize these OMU sub¬classifications. It uses a Monte Carlo simulation to perform an exploratory, multilinear unsupervised cluster analysis of each size/shape data array, and ranks the statistically significant schemes with least overlapping upper class limits that separate clusters within each array. Each upper class limit in highly ranked cluster schemes is then tested and validated using the 1-D object classifier, and the resultant cluster model of optimized “best cut top solution” upper bin limits is then coded into the user-defined size border file to operate the OMU classifier. This flexible design expands the ability of CMEIAS to recognize and classify the full range of OMU diversity optimized for the specific community under investigation. Also supporting the classifiers are an improved set of easily recognized pseudocolor assignments for each morphotype in the classified output image, and a CMEIAS Object Label edit tool that interactively reassigns objects to the best fit class when they are located at the continuum between classification borders, creates new morphotype classes if necessary, and reconstructs new images of morphotype-specific populations derived from classified community images containing multiple morphotypes.
The two “Individual” object classifiers in the CMEIAS-IT ver. 3.10 upgrade are of the semi-unsupervised type, associating the 14-dimensional morphotype classifier and the OMU classifier with a user-selected set of additional measurement attributes available in the Object Analysis plugin. These classifiers produce an output table of mixed object classification and object analysis data reported for each individual microbial object. These matrices can be evaluated by exploratory morphotype-weighted cluster analyses and data mining techniques to compute their unsupervised classification based on the (dis)similarities between microbial community structures. Various measures of eco-physiology (e.g., allometric scaling, metabolic activity, resource ecology, nutrient uptake systems, predatory bacteriovory, etc.) can be inferred from the output data acquired by the individual object classification-analysis modules.
A CMEIAS Data Toolpack addin application is being built to compile, tabulate, analyze, graph and compute numerous ecological statistics on CMEIAS population and community analysis data within Microsoft Excel. The program concatenates CMEIAS object analysis and classification data extracted from multiple images within the same community dataset and builds the properly formatted input tables of data for further statistical analyses. The addin then computes descriptive statistics, 2-d scatterplot and frequency distribution analyses with user-selected bin parameters on CMEIAS object analysis data, analyzes the tolerance envelope of confidence intervals surrounding CMEIAS object classification data derived from multiple image datasets to determine if the sample size (number of images and cell objects found within them) is sufficient to estimate diversity, constructs ranked abundance plots of community structure, computes optimized bin limits for frequency distribution analyses, and computes / tabulates / plots numerous diversity indices and (dis)similarity coefficients of classification data that compare the richness, dominance, evenness, diversity, relative abundance and composition of the microbial communities. These indices are useful in comparing communities and in ecological succession studies.
Recent collaborative studies using CMEIAS include analysis of the autecological biogeography and rice growth-promoting activity of candidate biofertilizer inoculant strains of rhizobia, microbial community analyses of corn leaf surfaces (normal vs. genetically engineered BT-corn), seasonal succession and food-web dynamics of epilithic microbial biofilm communities on streambed rocks in freshwater streams, analysis of active communities within soil aggregates, discrimination of biofilm architecture / diversity / eco-physiology / spatial ecology of freshwater microbial communities, in situ spatial scale of bacterial cell¬to-cell communication (quorum sensing) at single-cell resolution, and shifts in community structure of human vaginal microflora in health and bacterial vaginosis disease.
In summary, CMEIAS-based applications can help to fill major gaps in our understanding of microbial ecology by providing accurate, robust and user-friendly computing tools that extract ecologically important, quantitative information from digital images of microbes, at multiple spatial scales relevant to their diversity, abundance, physiology and in situ spatial distribution. The awesome computational power of CMEIAS used in conjunction with computer-assisted microscopy at single-cell resolution adds an exciting new dimension to microbial community analysis, and it is especially valuable when combined with molecular-based and other methods of polyphasic analysis. CMEIAS tools are admirably suited for microbial community analyses currently under investigation in this KBS-LTER program focused on row crop agriculture and in other studies of microbial community analysis conducted in other LTER programs. For instance, early studies using computer-assisted microscopy and CMEIAS image analysis are underway to examine diatom diversity and the epiphytic microbial communities that develop on dominant aquatic plants (lily pad, saw grass and peiphyton) in the Florida everglades ecosystem as a link to the Florida Coastal Everglades LTER program hosted at Florida International University. Finally, the CMEIAS website (http://cme.msu.edu/cmeias) includes a page entitled “Publications using CMEIAS” containing hyperlinked entries that describe various aspects of CMEIAS research and its applications as background information for this extended summary.Get poster
Back to meeting | Show |