Automated noninvasive epithelial cell counting in phase contrast microscopy images with automated parameter selection

Cell counting is commonly used to determine proliferation rates in cell cultures and for adherent cells it is often a ‘destructive’ process requiring disruption of the cell monolayer resulting in the inability to follow cell growth longitudinally. This process is time consuming and utilises significant resource. In this study a relatively inexpensive, rapid and widely applicable phase contrast microscopy‐based technique has been developed that emulates the contrast changes taking place when bright field microscope images of epithelial cell cultures are defocused. Processing of the resulting images produces an image that can be segmented using a global threshold; the number of cells is then deduced from the number of segmented regions and these cell counts can be used to generate growth curves. The parameters of this method were tuned using the discrete mereotopological relations between ground truth and processed images. Cell count accuracy was improved using linear discriminant analysis to identify spurious noise regions for removal. The proposed cell counting technique was validated by comparing the results with a manual count of cells in images, and subsequently applied to generate growth curves for oral keratinocyte cultures supplemented with a range of concentrations of foetal calf serum. The approach developed has broad applicability and utility for researchers with standard laboratory imaging equipment.     

Automated recognition and mapping of immunolabelled neurons in the developing brain

The cerebral cortex is distinguished by layers of neurons of different morphologies and densities. The layers are formed by the migration of newly generated neurons from the ventricular zone to the cortical plate near the outer (pial) boundary of the cortex, along radial paths approximately perpendicular to the cortical surface. Immunochemical labelling makes these cells' patterns visible in brightfield microscopy so that layer formation can be studied. We developed a suite of programs that automatically digitize the entire cortex, identify the labelled cells and compute cell densities along local radial paths. Cell identification used supervised classification on all the significantly stained objects corresponding to maxima in lowpass filtered versions of the digital micrographs. Classification of all the stained objects as cells or noncell objects was made by a decision rule based on morphometric and grey-level texture features, including features based on Gabor functions. Detection sensitivity and classification accuracy were jointly maximized on training data consisting of about 3000 expert-identified neurons in micrographs. Total program performance was tested on a separate (test) set of labelled neurons the same size as the training data set. The program detected 85% of the cells in the test set with a total error of 019. The identified cells' locations were used to compute population densities along normals to the cortical layers, and these densities served as a measure of neuronal migration. Transcortical density profiles obtained by computation and by manual cell counting were very similar. The cell identification program was built on well-established methods in statistical pattern recognition and image analysis and should generalize readily to other histological preparations.     

Cell tracking using phase‐adaptive shape prior

Automated tracking of cell population is very crucial for quantitative measurements of dynamic cell‐cycle behaviour of individual cells. This problem involves several subproblems and a high accuracy of each step is essential to avoid error propagation. In this paper, we propose a holistic three‐component system to tackle this problem. For each phase, we first learn a mean shape as well as a model of the temporal dynamics of transformation, which are used for estimating a shape prior for the cell in the current frame. We then segment the cell using a level set‐based shape prior model. Finally, we identify its phase based on the goodness‐of‐fit of the data to the segmentation model. This phase information is also used for fine‐tuning the segmentation result. We evaluate the performance of our method empirically in various aspects and in tracking individual cells from HeLa H2B‐GFP cell population. Highly accurate validation results confirm the robustness of our method in many realistic scenarios and the essentiality of each component of our integrating system.     

A counting method for density packed cells based on sliding band filter image enhancement

Cell loss and addition is an important biological event in pathology, and it usually provides central information to the changes of biological activity in the histological sections. To develop a reliable and accurate cell counting tools in tissue section, in this paper, we proposed a novel cell nuclei detecting method based on the sliding band filter which is a member of convergence index family. We evaluated the accuracy and performance of our method on density packed retinal outer nuclear layer cell confocal multivariate fluorescence microscopy image datasets. The results show our proposed method exhibited an excellent performance with its accuracy compared with human manual counting. It is worth noting that the proposed cell counting method can clearly benefit for retinal detachment and reattachment visual diagnostics close related to cell loss and addition.     

Estimating surface area by the isotropic fakir method from thick slices cut in an arbitrary direction

The proposed fakir method for estimating surface area is based on counting the intersections between the surface lying within a thick slice, and an isotropic spatial grid consisting of a combination of linear probes called fakir probes. An unbiased procedure using a directly randomized spatial grid rather than sections with randomized directions is presented. The method is applicable if perfectly registered serial sections of the surface are available in a thick slice while the direction of the slice can be arbitrary. The efficiency of the fakir method using different arrangements of orthogonal triplets of fakir probes is evaluated and it is shown that mutually shifted probes are superior to non-shifted ones. The application software for interactive counting of intersections between computer-generated fakir probes and the surface within the stack of digitized images is described and demonstrated by two examples: estimation of the surface area of individual tobacco cell chains using a confocal microscope, and estimation of the total area of exposed surface of mesophyll cells in a barley leaf using a wide-field transmission microscope.     

Centenary of Gaule and Lewin, pioneers in cell counting methodology

We commemorate the one hundredth anniversary of the publication of a pioneering paper on cell counting, by Gaule and Lewin. Their paper describes a new method for counting cells in tissue sections. First they found the mean number of cell profiles per cell by examining 50 selected cells in serial sections. Then they counted the total number of cell profiles in the ganglion, and finally divided this total profile number by the mean number of profiles per cell. They thought this method more accurate than counting cells by counting only profiles that showed the nucleolus because they noted that a cell's nucleolus sometimes appeared in more than one section and that a single cell could have more than one nucleolus.     

Algorithms for automated characterization of cell populations in thick specimens from 3-D confocal fluorescence microscopy data

Methods are presented for the automated, quantitative and three-dimensional (3-D) analysis of cell populations in thick, essentially intact tissue sections while maintaining intercell spatial relationships. This analysis replaces current manual methods which are tedious and subjective. The thick sample is imaged in three dimensions using a confocal scanning laser microscope. The stack of optical slices is processed by a 3-D segmentation algorithm that separates touching and overlapping structures using localization constraints. Adaptive data reduction is used to achieve computational efficiency. A hierarchical cluster analysis algorithm is used automatically to characterize the cell population by a variety of cell features. It allows automatic detection and characterization of patterns such as the 3-D spatial clustering of cells, and the relative distributions of cells of various sizes. It also permits the detection of structures that are much smaller, larger, brighter, darker, or differently shaped than the rest of the population. The overall method is demonstrated for a set of rat brain tissue sections that were labelled for tyrosine hydroxylase using fluorescein-conjugated antibodies. The automated system was verified by comparison with computer-assisted manual counts from the same image fields.     

Automated detection of fluorescent cells in in‐resin fluorescence sections for integrated light and electron microscopy

Integrated array tomography combines fluorescence and electron imaging of ultrathin sections in one microscope, and enables accurate high‐resolution correlation of fluorescent proteins to cell organelles and membranes. Large numbers of serial sections can be imaged sequentially to produce aligned volumes from both imaging modalities, thus producing enormous amounts of data that must be handled and processed using novel techniques. Here, we present a scheme for automated detection of fluorescent cells within thin resin sections, which could then be used to drive automated electron image acquisition from target regions via ‘smart tracking’. The aim of this work is to aid in optimization of the data acquisition process through automation, freeing the operator to work on other tasks and speeding up the process, while reducing data rates by only acquiring images from regions of interest. This new method is shown to be robust against noise and able to deal with regions of low fluorescence.     

Three-dimensional molecular distribution in single cells analysed using the digital imaging microscope

Cellular changes in molecular distribution are believed to underly a wide range of cell functions. In order to investigate changes in molecular distribution in single cells utilizing fluorescent probes we have developed a digital imaging microscope. The system, consisting of both hardware and software, automatically acquires 3-D data sets consisting of optical sections and then processes such data to facilitate the analysis of molecular distribution in single cells. The first major step in processing reverses distortion introduced principally by the optics of the fluorescent microscope. Various procedures for accomplishing this task are compared and a method based on regularization theory is shown to give superior results for several different 3-D images. Following this step features of interest are automatically extracted from 3-D images utilizing an artificial 3-D visual system. This artificial visual system utilizes a system of spatial filters to identify regional characteristics of images, the information obtained from these filters being used to identify and characterize clusters of molecules within the image. This information is then utilized to construct a 3-D graphical model of molecular distribution in single cells. Such models are displayed in 3-D and may be further analysed utilizing interactive 3-D computer graphics. These methods are illustrated by results obtained regarding alpha-actinin distribution in single smooth muscle cells.     

Confocal microscopy and three-dimensional reconstruction of electrophysiologically identified neurons in thick brain slices

Three-dimensional morphology and electrophysiology were correlated from individual neurons in a thick brain slice preparation. The hippocampal formation from immature and adult rats was cut transverse to the longitudinal axis into 500 μ Um-thick slices which were maintained under physiologic conditions. Individual neurons were impaled and physiologically characterized using microelectrodes. Recordings were made from the soma and in some cases from a dendrite. The impaled neurons were filled through the microelectrode with the fluorescent dye lucifer yellow and imaged by confocal scanning laser microscopy using an analog preprocessor. As many as 180 optical sections were recorded as a function of depth through the slices. Images are presented as a series of optical sections, stereo pairs, or three-dimensional reconstructions. Both stereo contouring and volume rendering methods were employed, and the reconstructions were viewed from any arbitrary perspective. Dendritic and axonal fields were separated from each other and displayed separately or as different pseudocolors. The three-dimensional reconstructions provided perspectives that were difficult or impossible to appreciate by viewing the optical sections or conventionally formed stereo pairs.     

Focused ion beam-scanning electron microscope: exploring large volumes of atherosclerotic tissue

Atherogenesis is a pathological condition in which changes in the ultrastructure and in the localization of proteins occur within the vasculature during all stages of the disease. To gain insight in those changes, high-resolution imaging is necessary. Some of these changes will only be present in a small number of cells, positioned in a 'sea' of non-affected cells. To localize this relatively small number of cells, there is a need to first navigate through a large area of the sample and subsequently zoom in onto the area of interest. This approach enables the study of specific cells within their in vivo environment and enables the study of (possible) interactions of these cells with their surrounding cells/environment. The study of a sample in a correlative way using light and electron microscopy is a promising approach to achieve this; however, it is very laborious and additional ultrastructural techniques might be very valuable to find the places of interest.
In this report we show that the focused ion beam-scanning electron microscope is a powerful tool to study biological specimens in a correlative way. With this microscope one can scan for the area of interest at low magnification, in this case the atherosclerotic plaque, and subsequently zoom in, for further analysis on an ultrastructural level, rendering valuable and detailed two- and three-dimensional information of, in this case, the endothelial cells and the vessel wall. Moreover, in combination with pre-embedment labelling of surface exposed antigens, the method allows insight into the 3D distribution of these markers.     

Comparison of two automatic cell‐counting solutions for fluorescent microscopic images

Cell counting in microscopic images is one of the fundamental analysis tools in life sciences, but is usually tedious, time consuming and prone to human error. Several programs for automatic cell counting have been developed so far, but most of them demand additional training or data input from the user. Most of them do not allow the users to online monitor the counting results, either. Therefore, we designed two straightforward, simple‐to‐use cell‐counting programs that also allow users to correct the detection results. In this paper, we present the Cellcounter and Learn 123 programs for automatic and semiautomatic counting of objects in fluorescent microscopic images (cells or cell nuclei) with a user‐friendly interface. Although Cellcounter is based on predefined and fine‐tuned set of filters optimized on sets of chosen experiments, Learn 123 uses an evolutionary algorithm to determine the adapt filter parameters based on a learning set of images. Cellcounter also includes an extension for analysis of overlaying images. The efficiency of both programs was assessed on images of cells stained with different fluorescent dyes by comparing automatically obtained results with results that were manually annotated by an expert. With both programs, the correlation between automatic and manual counting was very high (R² < 0.9), although Cellcounter had some difficulties processing images with no cells or weakly stained cells, where sometimes the background noise was recognized as an object of interest. Nevertheless, the differences between manual and automatic counting were small compared to variations between experimental repeats. Both programs significantly reduced the time required to process the acquired images from hours to minutes. The programs enable consistent, robust, fast and accurate detection of fluorescent objects and can therefore be applied to a range of different applications in different fields of life sciences where fluorescent labelling is used for quantification of various phenomena. Moreover, Cellcounter overlay extension also enables fast analysis of related images that would otherwise require image merging for accurate analysis, whereas Learn 123's evolutionary algorithm can adapt counting parameters to specific sets of images of different experimental settings.     

A method to compensate for light attenuation with depth in three-dimensional DNA image cytometry using a confocal scanning laser microscope

A method to compensate for attenuation of detected light with increased depth of the collected optical section, and its application in three-dimensional (3-D) DNA image cytometry is described. The method is based on studying the stack of 2-D histograms that can be formed from each consecutive pair of sections in a stack of optical serial sections. An attenuation factor is calculated interactively and a new compensated section series is computed. Formalin-fixed paraffin-embedded rat tissue was stained with propidium iodide. Each cell nucleus is extracted by thresholding and its total intensity is calculated. The coefficient of variation (CV) of the total intensity of all cells in each stack is computed. For comparison the CV of the same cells is computed in the uncompensated stacks. This study shows a significantly lower CV for the compensated data, thus contributing to the accuracy of DNA quantification in 3-D DNA image cytometry.     

Three-dimensional electron microscopy of entire cells

The digital processing of serial electron-microscope sections containing laser-induced topographical references allows a three-dimensional (3-D) reconstruction of entire cells at a depth resolution of 40–60 nm by the use of novel image analysis methods. The images are directly processed by a video-camera placed under the electron microscope in TEM mode or by the electron counting device in STEM mode. The deformations associated with the cutting of embedded cells are back-calculated by new computer algorithms developed for image analysis and treatment. They correct the artefacts caused by serial sectioning and automatically reconstruct the third dimension of the cells. Used in such a way, our data provide definitive information on the 3-D architecture of cells. This computer-assisted 3-D analysis represents a new tool for the documentation and analysis of cell ultrastructure and for morphometric studies. Furthermore, it is now possible for the observer to view the contents of the reconstructed tissue volume in a variety of different ways using computer-aided display techniques.     

Counting cells with stereology: Random versus serial sectioning

Counts of cells and nuclei from sections provide information central to studying structural changes in cells, tissues, and organs. This study considers some of the practical problems associated with counting cells with the newer random and serial sectioning methods of stereology and tests the hypothesis that similar cell counts can be obtained with both random and serial sectioning methods. Using irregularly shaped nuclei from alveolar cells of the goat lung, we compared cell counts derived from random (electron microscopic) and serial sectioning (light microscopic) methods. The results showed that both sectioning methods gave similar cell counts (10⁷/cm³ of parenchyma) for type 1 epithelial cells (5.0 vs. 5.0; P=1.0), type 2 epithelial cells (8.6 vs. 9.8; P=0.42) and interstitial cells (34.6 vs. 33.4; P=0.64), provided that corrections were introduced for sectionrelated biases and that the nuclei of the random sectioning method were corrected for shape. We found counting biases of 5%–7% for nuclear shape and 16% for section compression. These observations support the hypothesis that similar cell counts can be obtained with random and serial sectioning, even when nuclei have irregular shapes.     

The saucor, a new stereological tool for analysing the spatial distributions of cells, exemplified by human neocortical neurons and glial cells

The 3D spatial arrangement of particles or cells, for example glial cells, with respect to other particles or cells, for example neurons, can be characterized by the radial number density function, which expresses the number density of so-called 'secondary' particles as a function of their distance to a 'primary' particle. The present paper introduces a new stereological method, the saucor, for estimating the radial number density using thick isotropic uniform random or vertical uniform random sections. In the first estimation step, primary particles are registered in a disector. Subsequently, smaller counting windows are drawn with random orientation around every primary particle, and the positions of all secondary particles within the windows are recorded. The shape of the counting windows is designed such that a large portion of the volume close to the primary particle is examined and a smaller portion of the volume as the distance to the primary object increases. The experimenter can determine the relation between these volumina as a function of the distance by adjusting the parameters of the window graph, and thus reach a good balance between workload and obtained information. Estimation formulae based on the Horvitz-Thompson theorem are derived for both isotropic uniform random and vertical uniform random designs. The method is illustrated with an example where the radial number density of neurons and glial cells around neurons in the human neocortex is estimated using thick vertical sections for light microscopy. The results indicate that the glial cells are clustered around the neurons and the neurons have a tendency towards repulsion from each other.     

A segmentation algorithm for automated tracking of fast swimming unlabelled cells in three dimensions

Recent advances in microscopy and cytolabelling methods enable the real time imaging of cells as they move and interact in their real physiological environment. Scenarios in which multiple cells move autonomously in all directions are not uncommon in biology. A remarkable example is the swimming of marine spermatozoa in search of the conspecific oocyte. Imaging cells in these scenarios, particularly when they move fast and are poorly labelled or even unlabelled requires very fast three-dimensional time-lapse (3D+t) imaging. This 3D+t imaging poses challenges not only to the acquisition systems but also to the image analysis algorithms. It is in this context that this work describes an original automated multiparticle segmentation method to analyse motile translucent cells in 3D microscopical volumes. The proposed segmentation technique takes advantage of the way the cell appearance changes with the distance to the focal plane position. The cells translucent properties and their interaction with light produce a specific pattern: when the cell is within or close to the focal plane, its two-dimensional (2D) appearance matches a bright spot surrounded by a dark ring, whereas when it is farther from the focal plane the cell contrast is inverted looking like a dark spot surrounded by a bright ring. The proposed method analyses the acquired video sequence frame-by-frame taking advantage of 2D image segmentation algorithms to identify and select candidate cellular sections. The crux of the method is in the sequential filtering of the candidate sections, first by template matching of the in-focus and out-of-focus templates and second by considering adjacent candidates sections in 3D. These sequential filters effectively narrow down the number of segmented candidate sections making the automatic tracking of cells in three dimensions a straightforward operation.     

Recognition,presence, and survival of locust central nervous glia in situ and in vitro

Insect glial cells serve functions for the formation, maintenance, and performance of the central nervous system in ways similar to their vertebrate counterparts. Characterization of physiological mechanisms that underlie the roles of glia in invertebrates is largely incomplete, partly due to the lack of markers that universally label all types of glia throughout all developmental stages in various species. Studies on primary cell cultures from brains of Locusta migratoria demonstrated that the absence of anti-HRP immunoreactivity, which has previously been used to identify glial cells in undissociated brains, can also serve as a reliable glial marker in vitro, but only in combination with a viability test. As cytoplasmic membranes of cultured cells are prone to degradation when they lose viability, only cells that are both anti-HRP immunonegative and viable should be regarded as glial cells, whereas the lack of anti-HRP immunoreactivity alone is not sufficient. Cell viability can be assessed by the pattern of nuclear staining with DAPI (4′,6-diamidino-2-phenylindole), a convenient, sensitive labeling method that can be used in combination with other immunocytochemical cellular markers. We determined the glia-to-neuron ratio in central brains of fourth nymphal stage of Locusta migratoria to be 1:2 both in situ and in dissociated primary cell cultures. Analysis of primary cell cultures revealed a progressive reduction of glial cells and indicated that dead cells detach from the substrate and vanish from the analysis. Such changes in the composition of cell cultures should be considered in future physiological studies on cell cultures from insect nervous systems. Microsc. Res. Tech. 2009. © 2008 Wiley-Liss, Inc.     

Cs corrected STEM EELS: Analysing beam sensitive carbon nanomaterials in cellular structures

Identification of individual single wall nanotubes (SWNTs) within a cellular structure can provide vital information towards understanding the potential mechanisms of uptake, their localisation and whether their structure is transformed within a cell. To be able to image an individual SWNT in such an environment a resolution is required that is not usually appropriate for biological sections. Standard transmission electron microscopy (TEM) techniques such as bright field imaging of these cellular structures result in very weak contrast. Traditionally, researchers have stained the cells with heavy metal stains to enhance the cellular structure, however this can lead to confusion when analysing the samples at high resolution. Subsequently, alternative methods have been investigated to allow high resolution imaging and spectroscopy to identify SWNTs within the cell; here we will concentrate on the sample preparation and experimental methods used to achieve such resolution.     

A method of rational particle counting (author's transl)]

A method is described, which simplifies the counting of particles in a known volume. The counting of neurons of a brain region is used as an example for this procedure. The total volume is processed by histological serial sections. The total area of all sections of this volume is divided into counting squares by an ocular grid, from which we get a sample. This sample is described by the frequency distribution. The real number of particles in the volume is estimated by the values of the frequency distribution. The way of dividing, the size and the way of selection of these counting squares for a certain exactness of the estimation is determined. The method can be used for non-systematic distribution of particles in morphological structures.