Segmentation of complex cell clusters in microscopic images: Application to bone marrow samples

User-friendly tools for quantifying the dynamics of cellular morphology and intracellular protein clusters

Understanding the heterogeneous dynamics of cellular processes requires not only tools to visualize molecular behavior but also versatile approaches to extract and analyze the information contained in live-cell movies of many cells. Automated identification and tracking of cellular features enable thorough and consistent comparative analyses in a high-throughput manner. Here, we present tools for two challenging problems in computational image analysis: (1) classification of motion for cells with complex shapes and dynamics and (2) segmentation of clustered cells and quantification of intracellular protein distributions based on a single fluorescence channel. We describe these methods and user-friendly software(1) (MATLAB applications with graphical user interfaces) so these tools can be readily applied without an extensive knowledge of computational techniques.

Comparison of segmentation algorithms for fluorescence microscopy images of cells

The analysis of fluorescence microscopy of cells often requires the determination of cell edges. This is typically done using segmentation techniques that separate the cell objects in an image from the surrounding background. This study compares segmentation results from nine different segmentation techniques applied to two different cell lines and five different sets of imaging conditions. Significant variability in the results of segmentation was observed that was due solely to differences in imaging conditions or applications of different algorithms. We quantified and compared the results with a novel bivariate similarity index metric that evaluates the degree of underestimating or overestimating a cell object. The results show that commonly used threshold-based segmentation techniques are less accurate than k-means clustering with multiple clusters. Segmentation accuracy varies with imaging conditions that determine the sharpness of cell edges and with geometric features of a cell. Based on this observation, we propose a method that quantifies cell edge character to provide an estimate of how accurately an algorithm will perform. The results of this study will assist the development of criteria for evaluating interlaboratory comparability.

An automatic segmentation algorithm for 3D cell cluster splitting using volumetric confocal images

With the rapid advance of three-dimensional (3D) confocal imaging technology, more and more 3D cellular images will be available. Segmentation of intact cells is a critical task in automated image analysis and quantification of cellular microscopic images. One of the major complications in the automatic segmentation of cellular images arises due to the fact that cells are often closely clustered. Several algorithms are proposed for segmenting cell clusters but most of them are 2D based. In other words, these algorithms are designed to segment 2D cell clusters from a single image. Given 2D segmentation methods developed, they can certainly be applied to each image slice with the 3D cellular volume to obtain the segmented cell clusters. Apparently, in such case, the 3D depth information with the volumetric images is not really used. Often, 3D reconstruction is conducted after the individualized segmentation to build the 3D cellular models from segmented 2D cellular contours. Such 2D native process is not appropriate as stacking of individually segmented 2D cells or nuclei do not necessarily form the correct and complete 3D cells or nuclei in 3D. This paper proposes a novel and efficient 3D cluster splitting algorithm based on concavity analysis and interslice spatial coherence. We have taken the advantage of using the 3D boundary points detected using higher order statistics as an input contour for performing the 3D cluster splitting algorithm. The idea is to separate the touching or overlapping cells or nuclei in a 3D native way. Experimental results show the efficiency of our algorithm for 3D microscopic cellular images.

Segmentation of leukocytes and erythrocytes in blood smear images

Differential blood count is a standard method in hematological laboratory diagnosis. In the course of developing a computer-assisted microscopy system for the generation of differential blood counts, the detection and segmentation of white and red blood cells forms an essential step and its exactness is a fundamental prerequisite for the effectiveness of the subsequent classification step. We propose a method for the exact segmentation of leukocytes and erythrocytes in a simultaneous and cooperative way. We combine pixel-wise classification with template matching to locate erythrocytes and use a level-set approach in order to get the exact cell contours of leukocyte nucleus and plasma regions as well as erythrocyte regions. An evaluation comparing the performance of the algorithm to the manual segmentation performed by several persons yielded good results.

Automated three-dimensional detection and counting of neuron somata

We present a novel approach for automated detection of neuron somata. A three-step processing pipeline is described on the example of confocal image stacks of NeuN-stained neurons from rat somato-sensory cortex. It results in a set of position landmarks, representing the midpoints of all neuron somata. In the first step, foreground and background pixels are identified, resulting in a binary image. It is based on local thresholding and compensates for imaging and staining artifacts. Once this pre-processing guarantees a standard image quality, clusters of touching neurons are separated in the second step, using a marker-based watershed approach. A model-based algorithm completes the pipeline. It assumes a dominant neuron population with Gaussian distributed volumes within one microscopic field of view. Remaining larger objects are hence split or treated as a second neuron type. A variation of the processing pipeline is presented, showing that our method can also be used for co-localization of neurons in multi-channel images. As an example, we process 2-channel stacks of NeuN-stained somata, labeling all neurons, counterstained with GAD67, labeling GABAergic interneurons, using an adapted pre-processing step for the second channel. The automatically generated landmark sets are compared to manually placed counterparts. A comparison yields that the deviation in landmark position is negligible and that the difference between the numbers of manually and automatically counted neurons is less than 4%. In consequence, this novel approach for neuron counting is a reliable and objective alternative to manual detection.

Segmentation of touching cell nuclei using gradient flow tracking

Reliable cell nuclei segmentation is an important yet unresolved problem in biological imaging studies. This paper presents a novel computerized method for robust cell nuclei segmentation based on gradient flow tracking. This method is composed of three key steps: (1) generate a diffused gradient vector flow field; (2) perform a gradient flow tracking procedure to attract points to the basin of a sink; and (3) separate the image into small regions, each containing one nucleus and nearby peripheral background, and perform local adaptive thresholding in each small region to extract the cell nucleus from the background. To show the generality of the proposed method, we report the validation and experimental results using microscopic image data sets from three research labs, with both over-segmentation and under-segmentation rates below 3%. In particular, this method is able to segment closely juxtaposed or clustered cell nuclei, with high sensitivity and specificity in different situations.

Three-dimensional tissue cytometer based on high-speed multiphoton microscopy

Image cytometry technology has been extended to 3D based on high-speed multiphoton microscopy. This technique allows in situ study of tissue specimens preserving important cell-cell and cell-extracellular matrix interactions. The imaging system was based on high-speed multiphoton microscopy (HSMPM) for 3D deep tissue imaging with minimal photodamage. Using appropriate fluorescent labels and a specimen translation stage, we could quantify cellular and biochemical states of tissues in a high throughput manner. This approach could assay tissue structures with subcellular resolution down to a few hundred micrometers deep. Its throughput could be quantified by the rate of volume imaging: 1.45 mm(3)/h with high resolution. For a tissue containing tightly packed, stratified cellular layers, this rate corresponded to sampling about 200 cells/s. We characterized the performance of 3D tissue cytometer by quantifying rare cell populations in 2D and 3D specimens in vitro. The measured population ratios, which were obtained by image analysis, agreed well with the expected ratios down to the ratio of 1/10(5). This technology was also applied to the detection of rare skin structures based on endogenous fluorophores. Sebaceous glands and a cell cluster at the base of a hair follicle were identified. Finally, the 3D tissue cytometer was applied to detect rare cells that had undergone homologous mitotic recombination in a novel transgenic mouse model, where recombination events could result in the expression of enhanced yellow fluorescent protein in the cells. 3D tissue cytometry based on HSMPM demonstrated its screening capability with high sensitivity and showed the possibility of studying cellular and biochemical states in tissues in situ. This technique will significantly expand the scope of cytometric studies to the biomedical problems where spatial and chemical relationships between cells and their tissue environments are important.

An automated method for cell detection in zebrafish

Quantification of cells is a critical step towards the assessment of cell fate in neurological disease or developmental models. Here, we present a novel cell detection method for the automatic quantification of zebrafish neuronal cells, including primary motor neurons, Rohon-Beard neurons, and retinal cells. Our method consists of four steps. First, a diffused gradient vector field is produced. Subsequently, the orientations and magnitude information of diffused gradients are accumulated, and a response image is computed. In the third step, we perform non-maximum suppression on the response image and identify the detection candidates. In the fourth and final step the detected objects are grouped into clusters based on their color information. Using five different datasets depicting zebrafish cells, we show that our method consistently displays high sensitivity and specificity of over 95%. Our results demonstrate the general applicability of this method to different data samples, including nuclear staining, immunohistochemistry, and cell death detection.

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

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

High throughput quantification of cells with complex morphology in mixed cultures

Automated image-based and biochemical assays have greatly increased throughput for quantifying cell numbers in in vitro studies. However, it has been more difficult to automate the counting of specific cell types with complex morphologies in mixed cell cultures. We have developed a fully automated, fast, accurate and objective method for the quantification of primary human GFAP-positive astrocytes and CD45-positive microglia from images of mixed cell populations. This method, called the complex cell count (CCC) assay, utilizes a combination of image processing and analysis operations from MetaMorph (Version 6.2.6, Molecular Devices). The CCC assay consists of four main aspects: image processing with a unique combination of morphology filters; digital thresholding; integrated morphometry analysis; and a configuration of object standards. The time needed to analyze each image is 1.82s. Significant correlations have been consistently achieved between the data obtained from CCC analysis and manual cell counts. This assay can quickly and accurately quantify the number of human astrocytes and microglia in mixed cell culture and can be applied to quantifying a range of other cells/objects with complex morphology in neuroscience research.