Segmentation and tracking of live cells in phase-contrast images using directional gradient vector flow for snakes

Cell shape is an important characteristic of the physiological state of a cell and is used as a primary read-out of cell behaviour in various assays. Automated accurate segmentation of cells in microscopy images is hence of large practical importance in cell biology. We report a simple algorithm for automated cell segmentation in high-magnification phase-contrast images, which takes advantage of the characteristic directionality of the local image intensity gradient at cellular boundaries due to the 'halo-effect'. We employ a two-step algorithm in which a gradient vector flow (GVF) field is first used to direct active contours to an approximate cell boundary. A directional GVF (DGVF) field is then calculated by considering only edges for which the image intensity gradient is directed outwards with respect to the approximate cell contour. Subsequently, the DGVF field is used to refine the cell contour, by directing active contours to edges with the desired gradient directionality. This method allows us to accurately segment cells in an image series, as well as follow the dynamics of cell shape over time in an automated fashion.     

An image processing system for cell behaviour studies in subconfluent cultures

DRIMAPS (digitally recorded interference microscopy with automatic phase-shifting) is a system that we have developed for quantifying the behaviour of cells in subconfluent cultures. The primary data generated by the system consist of phase-shifted interference (PSI) images which are accurate density maps of the distribution of dry mass (non-aqueous material) inside cells. Time-lapse sequences of PSI images may be viewed as movie sequences or processed in various ways to reveal many different aspects of the dynamics of cell growth and motile behaviour. Here we describe the image processing routines that are an integral part of the system and are required for four main functions: 1 initial calculation of the PSI images; 2 compensation of these images for instrumental distortion and instability; 3 identification and tracking of individual cells in a time-lapse sequence of PSI images; 4 extraction of cell behavioural data from a time-lapse sequence of PSI images. The first function converts standard interference microscope images into an image that accurately represents the optical phase-difference introduced by the specimen. The second function recalibrates a sequence of images by taking the cell-free region in each image as a reference plane of zero phase-difference. This is particularly necessary to compensate for the long-term instability of the Horn type of double-beam interference microscope, which has several advantages over other types of interference microscope for studying cell behaviour. The third function compares consecutive images in a sequence in order to trace the identity of individual cells throughout the sequence. Semi-automatic tracking, which allows close interaction with a human operator, is less prone to error than fully automatic tracking. The fourth function automatically extracts dynamic data from the identified cells. These data may include the true mass centroids of cells for translocation analysis and robust morphometric parameters for cell morphology examination. The integrated intensity of a cell is an accurate measure of cell mass and allows the growth (increase in dry mass) of individual cells to be studied. These data may be entered into a relational database of cell behaviour and a rule-based system allows efficient data access and analysis. Experiments with phase-contrast microscopy have revealed that many of these image processing methods are generally useful for cell behaviour studies using more conventional forms of microscopy.     

Sequential processing of quantitative phase images for the study of cell behaviour in real‐time digital holographic microscopy

Transmitted light holographic microscopy is particularly used for quantitative phase imaging of transparent microscopic objects such as living cells. The study of the cell is based on extraction of the dynamic data on cell behaviour from the time‐lapse sequence of the phase images. However, the phase images are affected by the phase aberrations that make the analysis particularly difficult. This is because the phase deformation is prone to change during long‐term experiments. Here, we present a novel algorithm for sequential processing of living cells phase images in a time‐lapse sequence. The algorithm compensates for the deformation of a phase image using weighted least‐squares surface fitting. Moreover, it identifies and segments the individual cells in the phase image. All these procedures are performed automatically and applied immediately after obtaining every single phase image. This property of the algorithm is important for real‐time cell quantitative phase imaging and instantaneous control of the course of the experiment by playback of the recorded sequence up to actual time. Such operator's intervention is a forerunner of process automation derived from image analysis. The efficiency of the propounded algorithm is demonstrated on images of rat fibrosarcoma cells using an off‐axis holographic microscope.     

Automatic tracking of individual migrating cells using low-magnification dark-field microscopy

Many fundamental biological processes, such as the search for food, immunological responses and wound healing, depend on cell migration. Video microscopy allows the magnitude and direction of cell migration to be documented. Here, we present a simple and inexpensive method for simultaneous tracking of hundreds of migrating cells over periods of several days. Low-magnification dark-field microscopy was used to visualize individual cells whereas time-lapse video images were acquired by computer for future analysis. We employed an automated tracking algorithm to identify individual cells on each video image allowing migration paths to be tracked using a nearest neighbour algorithm. To test the method, we followed the time-course of migration of 3T3 fibroblasts, endothelial cells and individual amoeba in the absence of any chemical stimulus gradient. All cell types showed a 'random walk' behaviour in which mean squared displacement in position increased linearly with time. We defined a 'migration coefficient' (D(mig)), analogous to a diffusion coefficient, which gave an estimate of cell migration rate. D(mig) depended on cell type and temperature. When amoebas were made to undergo chemotaxis, the cells no longer followed a random walk but instead moved at a near constant velocity (V(av)) towards the chemotactic stimulus.     

Automated tracking of migrating cells in phase-contrast video microscopy sequences using image registration

Analysis of in vitro cell motility is a useful tool for assessing cellular response to a range of factors. However, the majority of cell-tracking systems available are designed primarily for use with fluorescently labelled images. In this paper, five commonly used tracking systems are examined for their performance compared with the use of a novel in-house cell-tracking system based on the principles of image registration and optical flow. Image registration is a tool commonly used in medical imaging to correct for the effects of patient motion during imaging procedures and works well on low-contrast images, such as those found in bright-field and phase-contrast microscopy. The five cell-tracking systems examined were Retrac, a manual tracking system used as the gold standard; CellTrack, a recently released freely downloadable software system that uses a combination of tracking methods; ImageJ, which is a freely available piece of software with a plug-in for automated tracking (MTrack2) and Imaris and Volocity, both commercially available automated tracking systems. All systems were used to track migration of human epithelial cells over ten frames of a phase-contrast time-lapse microscopy sequence. This showed that the in-house image-registration system was the most effective of those tested when tracking non-dividing epithelial cells in low-contrast images, with a successful tracking rate of 95%. The performance of the tracking systems was also evaluated by tracking fluorescently labelled epithelial cells imaged with both phase-contrast and confocal microscopy techniques. The results showed that using fluorescence microscopy instead of phase contrast does improve the tracking efficiency for each of the tested systems. For the in-house software, this improvement was relatively small (<5% difference in tracking success rate), whereas much greater improvements in performance were seen when using fluorescence microscopy with Volocity and ImageJ.     

Defocusing microscopy

Transparent objects (phase objects) are not visible in a standard brightfield optical microscope. In order to see such objects the most used technique is phase-contrast microscopy. In phase-contrast microscopy the contrast observed is proportional to the optical path difference introduced by the object. If the index of refraction is uniform, phase-contrast microscopy then yields a measure of the thickness profile of phase objects. We show that by slightly defocusing an optical microscope operating in brightfield, phase objects become visible. We modeled such an effect and show that the image contrast of a phase object is proportional to the amount of defocusing and proportional to the two-dimensional Laplacian of the optical path difference introduced by the object. For uniform index of refraction, defocusing microscopy then yields a measure of the curvature profile of phase objects. We extended our previous model for thin objects to thick objects. To check our theoretical model, we use as phase objects polystyrene spherical caps and compare their curvature radii obtained by defocusing microscopy (DM) to those obtained with atomic force microscopy (AFM). We also show that for thick curved phase objects one can reconstruct their thickness profiles from DM images. We illustrate the utility of defocusing microscopy in biological systems to study cell motility. In particular, we visualize and quantitatively measure real-time cytoskeleton curvature fluctuations of macrophages (a cell of the innate immune system). The study of such fluctuations might be important for a better understanding of the engulfment process of pathogens during phagocytosis.     

Roles of cell adhesion molecules nectin and nectin-like molecule-5 in the regulation of cell movement and proliferation

In response to chemoattractants, migrating cells form protrusions, such as lamellipodia and filopodia, and structures, such as ruffles over lamellipodia, focal complexes and focal adhesions at leading edges. The formation of these leading edge structures is essential for directional cell movement. Nectin-like molecule-5 (Necl-5) interacts in cis with PDGF receptor and integrin alpha(v)beta(3), and enhances the activation of signalling molecules associated with these transmembrane proteins, which results in the formation of leading edge structures and enhancement of directional cell movement. When migrating cells come into contact with each other, cell-cell adhesion is initiated, resulting in reduced cell velocity. Necl-5 first interacts in trans with nectin-3. This interaction is transient and induces down-regulation of Necl-5 expression at the cell surface, resulting in reduced cell movement. Cell proliferation is also suppressed by the down-regulation of Necl-5, because the inhibitory effect of Necl-5 on Sprouty2, a negative regulator of the Ras signalling, is diminished. PDGF receptor and integrin alpha(v)beta(3), which have interacted with Necl-5, then form a complex with nectin, which initiates cell-cell adhesion and recruits cadherin to the nectin-based cell-cell adhesion sites to form stable adherens junctions. The formation of adherens junctions stops cell movement, in part through inactivation of integrin alpha(v)beta(3) caused by the trans-interaction of nectin. Thus, nectin and Necl-5 play key roles in the regulation of cell movement and proliferation.     

Analysis of integrin (CD11b/CD18) movement during neutrophil adhesion and migration on endothelial cells

Little is known of the distribution of cell surface molecules during the adhesion and migration of leucocytes on endothelial cells. We have used confocal microscopy and a Fab fragment of a non-inhibitory monoclonal antibody recognizing the integrin CD11b/CD18 (Mac-1) to study the movement of this adhesion molecule over time. We found that during the initial stage of neutrophil contact with TNF-α activated human umbilical vein endothelial cells (HUVEC), there is a rapid accumulation of Mac-1 at the contact area between the two cell types. As the neutrophil spreads, Mac-1 redistributes away from this initial contact area. During neutrophil migration on HUVEC, Mac-1 was redistributed to the leading edge of the migrating cell, suggesting that the existing cell surface pool of adhesion molecules is dynamic and can be recruited to the leading front as the cell changes direction. As neutrophils migrate on HUVEC, Mac-1-dense macroaggregates are rapidly formed and broken down at the contact plane between the two cells. The confocal microscope, coupled with the use of non-inhibitory antibodies labelled with photostable fluorophores, is a useful tool for the study of the movement of cell surface molecules over time.     

A method for quantifying cell size from differential interference contrast images: validation and application to osmotically stressed chondrocytes

An automatic image analysis method was developed to determine the shape and size of spheroidal cells from a time series of differential interference contrast (DIC) images. The program incorporates an edge detection algorithm and dynamic programming for edge linking. To assess the accuracy and working range of the method, results from DIC images of different focal planes and resolutions were compared to confocal images in which the cell membrane was fluorescently labelled. The results indicate that a 1‐µm focal drift from the in‐focus plane can lead to an overestimation of cell volume up to 14.1%, mostly due to shadowing effects of DIC microscopy. DIC images allow for accurate measurements when the focal plane lies in a zone slightly above the centre of a spherical cell. In this range the method performs with 1.9% overall volume error without taking into account the error introduced by the representation of the cell as a sphere. As a test case, the method was applied to quantify volume changes due to acute changes of osmotic stress.     

The Large-Scale Digital Cell Analysis System: an open system for nonperturbing live cell imaging

The Large-Scale Digital Cell Analysis System (LSDCAS) was designed to provide a highly extensible open source live cell imaging system. Analysis of cell growth data has demonstrated a lack of perturbation in cells imaged using LSDCAS, through reference to cell growth data from cells growing in CO₂ incubators. LSDCAS consists of data acquisition, data management and data analysis software, and is currently a Core research facility at the Holden Comprehensive Cancer Center at the University of Iowa. Using LSDCAS analysis software, this report and others show that although phase-contrast imaging has no apparent effect on cell growth kinetics and viability, fluorescent image acquisition in the cell lines tested caused a measurable level of growth perturbation using LSDCAS. This report describes the current design of the system, reasons for the implemented design, and details its basic functionality. The LSDCAS software runs on the GNU/Linux operating system, and provides easy to use, graphical programs for data acquisition and quantitative analysis of cells imaged with phase-contrast or fluorescence microscopy (alone or in combination), and complete source code is freely available under the terms of the GNU Public Software License at the project website ( http://lsdcas.engineering.uiowa.edu ).     

Segmentation of ascidian notochord cells in DIC timelapse images

We have developed a method to automatically segment notochord cell boundaries from differential interference contrast (DIC) timelapse images of the elongating ascidian tail. The method is based on a specialized parametric active contour, the network snake, which can be initialized as a network of arbitrary but fixed topology and provides an effective framework for simultaneously segmenting multiple touching cells. Several modifications to the original network snake were necessary for high-quality segmentation, including linear Gaussian derivative filtering to reconstruct edge maps from DIC images and a new energy function to improve the segmentation of critical cell-cell vertices. We find that post-intercalation ascidian notochord cells exhibit two distinct cell behaviors: lateral cell edges expand along the AP axis while showing a rapid pulsatile behavior, whereas anterior and posterior cell edges contract smoothly.     

A pattern recognition method for lattice distortion measurement from HRTEM images

The idea of the method is to analyse a crystal lattice by creating a grid of quadrilaterals corresponding to repeated cells that are visible in the image. This approach combines image processing elements with a continuum field theory, to create a distortion-independent similarity measure that is used to select the most appropriate among possible lattice configurations. Subsequently, displacement and distortion fields are computed from individual cell positions. The method allows one to obtain these fields even for images where a periodic cell does not necessarily appear as a single dot of intensity in a high-resolution transmission electron microscopy (HRTEM) image, which results in a lower accuracy of commonly used approaches, namely geometric phase and peak finding. The results obtained from this method are verified quantitatively by comparison with known distortion tensor distributions and Burgers vector values on both simulated and real images.     

Real‐time three‐dimensional imaging of cell division by differential interference contrast microscopy

Differential interference contrast (DIC) microscopy can provide information about subcellular components and organelles inside living cells. Applicability to date, however, has been limited to 2D imaging. Unfortunately, understanding of cellular dynamics is difficult to extract from these single optical sections. We demonstrate here that 3D differential interference contrast microscopy has sub‐diffraction limit resolution both laterally and vertically, and can be used for following Madin Darby canine kidney cell division process in real time. This is made possible by optimization of the microscope optics and by incorporating computer‐controlled vertical scanning of the microscope stage.     

An efficient semi-automatic algorithm for cell contour extraction

An interactive semi-automatic procedure for extraction of cell contours from light microscope images is presented. The user is required to specify four contour points and the algorithm determines the rest of the contour automatically. The algorithm exploits the fact that cell contours have lower grey level than their immediate surrounding and are usually very similar in shape to some piece-wise ellipses. A cost function is defined to detect the cell contours incorporating both the elliptical shape and local image intensities. The procedure is fast, reliable and well suited for routine interactive applications.     

Sample preparation for electron microscopy of internal cell structure.

Methods are reviewed for examination of internal cell structure by high-resolution scanning electron microscopy and compared with the rapid-freeze deep-etch replica technique used in transmission electron microscopy. Rapid freezing of fresh material, followed by freeze-fracture, provides a theoretically attractive approach in ultrastructure studies, but the high protein and solute content of most cells prevents a deep three-dimensional view for material frozen without some form of extraction. After discussion of other methods it is concluded that the most useful general approach, at least for cultured cells, is to first permeabilize or break open the cells in a medium which preserves the structure under study in a functional state as, for example, the movement of chromosomes along the division spindle, or transport of proteins within the Golgi region. After permeabilization, with attendant partial extraction, the preparation can be fixed, then viewed by either deep-etch replication, or by high-resolution scanning electron microscopy, with structure of interest revealed in deep view.     

Automated classification and visualization of fluorescent live cell microscopy images

Robotic, high‐throughput microscopy is a powerful tool for small molecule screening and classifying cell phenotype, proteomic and genomic data. An important hurdle in the field is the automated classification and visualization of results collected from a data set of tens of thousands of images. We present a method that approaches these problems from the perspective of flow cytometry with supporting open‐source code. Image analysis software was created that allowed high‐throughput microscopy data to be analysed in a similar manner as flow cytometry. Each cell on an image is considered an object and a series of gates similar to flow cytometry is used to classify and quantify the properties of cells including size and level of fluorescent intensity. This method is released with open‐source software and code that demonstrates the method's implementation. Accuracy of the software was determined by measuring the levels of apoptosis in a primary murine myoblast cell line after exposure to staurosporine and comparing these results to flow cytometry.     

Signal analysis of total internal reflection fluorescent speckle microscopy (TIR-FSM) and wide-field epi-fluorescence FSM of the actin cytoskeleton and focal adhesions in living cells

Fluorescent speckle microscopy (FSM) uses low levels of fluorescent proteins to create fluorescent speckles on cytoskeletal polymers in high‐resolution fluorescence images of living cells. The dynamics of speckles over time encode subunit turnover and motion of the cytoskeletal polymers. We sought to improve on current FSM technology by first expanding it to study the dynamics of a non‐polymeric macromolecular assembly, using focal adhesions as a test case, and second, to exploit for FSM the high contrast afforded by total internal reflection fluorescence microscopy (TIR‐FM). Here, we first demonstrate that low levels of expression of a green fluorescent protein (GFP) conjugate of the focal adhesion protein, vinculin, results in clusters of fluorescent vinculin speckles on the ventral cell surface, which by immunofluorescence labelling of total vinculin correspond to sparse labelling of dense focal adhesion structures. This demonstrates that the FSM principle can be applied to study focal adhesions. We then use both GFP‐vinculin expression and microinjected fluorescently labelled purified actin to compare quantitatively the speckle signal in FSM images of focal adhesions and the actin cytoskeleton in living cells by TIR‐FM and wide‐field epifluorescence microscopy. We use quantitative FSM image analysis software to define two new parameters for analysing FSM signal features that we can extract automatically: speckle modulation and speckle detectability. Our analysis shows that TIR‐FSM affords major improvements in these parameters compared with wide‐field epifluorescence FSM. Finally, we find that use of a crippled eukaryotic expression promoter for driving low‐level GFP‐fusion protein expression is a useful tool for FSM imaging. When used in time‐lapse mode, TIR‐FSM of actin and GFP‐conjugated focal adhesion proteins will allow quantification of molecular dynamics within interesting macromolecular assemblies at the ventral surface of living cells.     

Electron microscopy localization and characterization of functionalized composite organic-inorganic SERS nanoparticles on leukemia cells

We demonstrate the use of electron microscopy as a powerful characterization tool to identify and locate antibody-conjugated composite organic-inorganic nanoparticle (COINs) surface enhanced Raman scattering (SERS) nanoparticles on cells. U937 leukemia cells labeled with antibody CD54-conjugated COINs were characterized in their native, hydrated state using wet scanning electron microscopy (SEM) and in their dehydrated state using high-resolution SEM. In both cases, the backscattered electron (BSE) detector was used to detect and identify the silver constituents in COINs due to its high sensitivity to atomic number variations within a specimen. The imaging and analytical capabilities in the SEM were further complemented by higher resolution transmission electron microscopy (TEM) images and scanning Auger electron spectroscopy (AES) data to give reliable and high-resolution information about nanoparticles and their binding to cell surface antigens.     

应用Canny算法和灰度等高线的金相组织封闭边缘提取

为了对金相组织的多项组织含量进行定量分析,提出应用Canny算法和灰度等高线相结合的方法来产生封闭金相组织边缘。首先,采用高斯函数对金相组织的灰度图像进行模糊处理,然后,用Canny算法对模糊后的图像获取具有单边缘效应的原始边缘。通过最大类间方差法对原始边缘进行自动计算,获得抑制虚假边缘的阈值,并通过该阈值获取基础边缘。由于基础边缘具有不连续性,因此需根据基础边缘端点的邻域灰度中值计算所需的灰度等高线,同时根据设定的等高线和基础边缘融合的条件,以基础边缘的端点为起点生成封闭边缘。实验结果表明,该算法可有效地生成符合要求的封闭边缘,采用该算法产生的封闭边缘进行金相组织含量测定可使测量误差降低到±1%,可测组织含量范围达到5%～95%。该算法能满足多项金相组织含量测定的定量分析要求且具有普适性。     

Depth enhancement of 3D microscopic living‐cell image using incoherent fluorescent digital holography

Multilayer images of living cells are typically obtained using confocal or multiphoton microscopy. However, limitations on the distance between consecutive scan layers hinder high‐resolution three‐dimensional reconstruction, and scattering strongly degrades images of living cell components. Consequently, when overlapping information from different layers is focused on a specific point in the camera, this causes uncertainty in the depiction of the cell components. We propose a method that combines the Fresnel incoherent correlation holography and a depth‐of‐focus reduction algorithm to enhance the depth information of three‐dimensional cell images. The proposed method eliminates overlap between light elements in the different layers inside living cells and limitations on the interlayer distance, and also enhances the contrast of the reconstructed holograms of living cells.