Probabilistic segmentation and intensity estimation for microarray images

We describe a probabilistic approach to simultaneous image segmentation and intensity estimation for complementary DNA microarray experiments. The approach overcomes several limitations of existing methods. In particular, it (a) uses a flexible Markov random field approach to segmentation that allows for a wider range of spot shapes than existing methods, including relatively common 'doughnut-shaped' spots; (b) models the image directly as background plus hybridization intensity, and estimates the two quantities simultaneously, avoiding the common logical error that estimates of foreground may be less than those of the corresponding background if the two are estimated separately; and (c) uses a probabilistic modeling approach to simultaneously perform segmentation and intensity estimation, and to compute spot quality measures. We describe two approaches to parameter estimation: a fast algorithm, based on the expectation-maximization and the iterated conditional modes algorithms, and a fully Bayesian framework. These approaches produce comparable results, and both appear to offer some advantages over other methods. We use an HIV experiment to compare our approach to two commercial software products: Spot and Arrayvision.     

Quantitative Analysis of Centromeric FISH Spots during the Cell Cycle by Image Cytometry

Two-color fluorescence in situ hybridization (FISH) with chromosome enumeration DNA probes specific to chromosomes 7, 11, 17, and 18 was applied to CAL-51 breast cancer cells to examine whether the fluorescence intensity of FISH spots was associated with cell cycle progression. The fluorescence intensity of each FISH spot was quantitatively analyzed based on the cell cycle stage determined by image cytometry at the single-cell level. The spot intensity of cells in the G₂ phase was larger than that in the G_0/1 phase. This increased intensity was not seen during the early and mid S phases, whereas the cells in the late S phase showed significant increases in spot intensity, reaching the same level as that observed in the G₂ phase, indicating that alpha satellite DNA in the centromeric region was replicated in the late S phase. Thus, image cytometry can successfully detect small differences in the fluorescence intensities of centromeric spots of homologous chromosomes. This combinational image analysis of FISH spots and the cell cycle with cell image cytometry provides insights into new aspects of the cell cycle. This is the first report demonstrating that image cytometry can be used to analyze the fluorescence intensity of FISH signals during the cell cycle.     

Smart 3D-FISH: automation of distance analysis in nuclei of interphase cells by image processing.

BACKGROUND: Detection of fluorescent probes by fluorescence in situ hybridization in cells with preserved three-dimensional nuclear structures (3D-FISH) is useful for studying the organization of chromatin and localization of genes in interphase nuclei. Fast and reliable measurements of the relative positioning of fluorescent spots specific to subchromosomal regions and genes would improve understanding of cell structure and function. METHODS: 3D-FISH protocol, confocal microscopy, and digital image analysis were used. RESULTS: New software (Smart 3D-FISH) has been developed to automate the process of spot segmentation and distance measurements in images from 3D-FISH experiments. It can handle any number of fluorescent spots and incorporate images of 4',6-diamino-2-phenylindole counterstained nuclei to measure the relative positioning of spot loci in the nucleus and inter-spot distance. Results from a pilot experiment using Smart 3D-FISH on ENL, MLL, and AF4 genes in two lymphoblastic cell lines were satisfactory and consistent with data published in the literature. CONCLUSION: Smart 3D-FISH should greatly facilitate image processing and analysis of 3D-FISH images by providing a useful tool to overcome the laborious task of image segmentation based on user-defined parameters and decrease subjectivity in data analysis. It is available as a set of plugins for ImageJ software.     

Study of fluorescence in situ hybridization for detection of chromosome aberration

The authors applied fluorescence in situ hybridization (FISH) technique for the detection of chromosome aberration in interphase nuclei using the probe specific to alphoid repeats on chromosome 11 and X. Chromosome 11 specific probe showed two major spots in lymphocyte nuclei, while X specific probe showed single spot in male and double spots in female respectively. On the other hand three spots were detected in most of the nuclei from HeLa cells with 11 and X specific probes. We concluded that FISH with the use of chromosome specific probe may become a useful and reliable tool for the detection of chromosome aberration in interphase nuclei.     

Slide Preparation Method to Preserve Three-dimensional Chromatin Architecture of Testicular Germ Cells

During testicular germ cell differentiation, the structure of nuclear chromatin dynamically changes. The following describes a method designed to preserve the three-dimensional chromatin arrangement of testicular germ cells found in mice; this method has been termed as the three-dimensional (3D) slide method. In this method, testicular tubules are directly treated with a permeabilization step that removes cytoplasmic material, followed by a fixation step that fixes nuclear materials. Tubules are then dissociated, the cell suspension is cytospun, and cells adhere to slides. This method improves sensitivity towards detection of subnuclear structures and is applicable for immunofluorescence, DNA, and RNA fluorescence in situ hybridization (FISH) and the combination of these detection methods. As an example of a possible application of the 3D slide method, a Cot-1 RNA FISH is shown to detect nascent RNAs. The 3D slide method will facilitate the detailed examination of spatial relationships between chromatin structure, DNA, and RNA during testicular germ cell differentiation.     

A Combinational Clustering Based Method for cDNA Microarray Image Segmentation

Microarray technology plays an important role in drawing useful biological conclusions by analyzing thousands of gene expressions simultaneously. Especially, image analysis is a key step in microarray analysis and its accuracy strongly depends on segmentation. The pioneering works of clustering based segmentation have shown that k-means clustering algorithm and moving k-means clustering algorithm are two commonly used methods in microarray image processing. However, they usually face unsatisfactory results because the real microarray image contains noise, artifacts and spots that vary in size, shape and contrast. To improve the segmentation accuracy, in this article we present a combination clustering based segmentation approach that may be more reliable and able to segment spots automatically. First, this new method starts with a very simple but effective contrast enhancement operation to improve the image quality. Then, an automatic gridding based on the maximum between-class variance is applied to separate the spots into independent areas. Next, among each spot region, the moving k-means clustering is first conducted to separate the spot from background and then the k-means clustering algorithms are combined for those spots failing to obtain the entire boundary. Finally, a refinement step is used to replace the false segmentation and the inseparable ones of missing spots. In addition, quantitative comparisons between the improved method and the other four segmentation algorithms--edge detection, thresholding, k-means clustering and moving k-means clustering--are carried out on cDNA microarray images from six different data sets. Experiments on six different data sets, 1) Stanford Microarray Database (SMD), 2) Gene Expression Omnibus (GEO), 3) Baylor College of Medicine (BCM), 4) Swiss Institute of Bioinformatics (SIB), 5) Joe DeRisi’s individual tiff files (DeRisi), and 6) University of California, San Francisco (UCSF), indicate that the improved approach is more robust and sensitive to weak spots. More importantly, it can obtain higher segmentation accuracy in the presence of noise, artifacts and weakly expressed spots compared with the other four methods.     

Segmentation of cDNA microarray spots using markov random field modeling

Motivation: Spot segmentation is a critical step in microarraygene expression data analysis. Therefore, the performance ofsegmentation may substantially affect the results of subsequentstages of the analysis, such as the detection of differentiallyexpressed genes. Several methods have been developed to segmentmicroarray spots from the surrounding background. In this study,we have proposed a new approach based on Markov random field(MRF) modeling and tested its performance on simulated and realmicroarray images against a widely used segmentation methodbased on Mann–Whitney test adopted by QuantArray software(Boston, MA). Spot addressing was performed using QuantArray.We have also devised a simulation method to generate microarrayimages with realistic features. Such images can be used as goldstandards for the purposes of testing and comparing differentsegmentation methods, and optimizing segmentation parameters. Results: Experiments on simulated and 14 actual microarray imagesets show that the proposed MRF-based segmentation method candetect spot areas and estimate spot intensities with higheraccuracy. Availability: The algorithms were implemented in Matlab^TM (TheMathworks, Inc., Natick, MA) environment. The codes for MRF-basedsegmentation and image simulation methods are available uponrequest. Contact: demirkaya{at}ieee.org     

Robust 3D DNA FISH Using Directly Labeled Probes

3D DNA FISH has become a major tool for analyzing three-dimensional organization of the nucleus, and several variations of the technique have been published. In this article we describe a protocol which has been optimized for robustness, reproducibility, and ease of use. Brightly fluorescent directly labeled probes are generated by nick-translation with amino-allyldUTP followed by chemical coupling of the dye. 3D DNA FISH is performed using a freeze-thaw step for cell permeabilization and a heating step for simultaneous denaturation of probe and nuclear DNA. The protocol is applicable to a range of cell types and a variety of probes (BACs, plasmids, fosmids, or Whole Chromosome Paints) and allows for high-throughput automated imaging. With this method we routinely investigate nuclear localization of up to three chromosomal regions.     

FISH Finder: a high-throughput tool for analyzing FISH images

MOTIVATION: Fluorescence in situ hybridization (FISH) is used to study the organization and the positioning of specific DNA sequences within the cell nucleus. Analyzing the data from FISH images is a tedious process that invokes an element of subjectivity. Automated FISH image analysis offers savings in time as well as gaining the benefit of objective data analysis. While several FISH image analysis software tools have been developed, they often use a threshold-based segmentation algorithm for nucleus segmentation. As fluorescence signal intensities can vary significantly from experiment to experiment, from cell to cell, and within a cell, threshold-based segmentation is inflexible and often insufficient for automatic image analysis, leading to additional manual segmentation and potential subjective bias. To overcome these problems, we developed a graphical software tool called FISH Finder to automatically analyze FISH images that vary significantly. By posing the nucleus segmentation as a classification problem, compound Bayesian classifier is employed so that contextual information is utilized, resulting in reliable classification and boundary extraction. This makes it possible to analyze FISH images efficiently and objectively without adjustment of input parameters. Additionally, FISH Finder was designed to analyze the distances between differentially stained FISH probes. AVAILABILITY: FISH Finder is a standalone MATLAB application and platform independent software. The program is freely available from: http://code.google.com/p/fishfinder/downloads/list.     

植物rRNA基因组织模式的荧光原位杂交比较分析

采用荧光原位杂交技术,对分属5个科的10种植物的分生细胞的18S-25S rRNA基因(45S rDNA)的组织模式进行了比较分析.45S rDNA探针在所有供试植物的间期核都产生了两种杂交信号:荧光强、位于核仁周边的纽和荧光较弱分布于核仁内的点,表明不同植物间期核的rDNA染色质的组织模式相似.在每种植物的部分间期细胞都观察到点与纽相连或从纽发出的情况,而且点的数目越多纽就变得越小,点的有无和数目的多少与细胞的活性呈正相关.这些事实表明,纽代表了处于凝缩状态的非活性的rDNA染色质,点是由纽解凝缩而来,rDNA异染色质解凝缩形成点是植物rRNA基因活跃转录的细胞学表现,在同一物种中点的多少代表了间期核rDNA转录活性的强弱.我们的结果支持点是核仁内活性rRNA基因组织的结构单位及rRNA合成发生地点的推论.我们的结果还显示,不同植物间期核的rDNA染色质的组织也存在一些差异,其中核仁内点的最大数目有较大的不同.在所有供试植物的有丝分裂前中期细胞,45S rDNA探针在rDNA位点都产生了松散的信号块和许多点,表明植物的rDNA位点在有丝分裂前中期还有较活跃的转录.     

Improved and Robust Detection of Cell Nuclei from Four Dimensional Fluorescence Images

Segmentation-free direct methods are quite efficient for automated nuclei extraction from high dimensional images. A few such methods do exist but most of them do not ensure algorithmic robustness to parameter and noise variations. In this research, we propose a method based on multiscale adaptive filtering for efficient and robust detection of nuclei centroids from four dimensional (4D) fluorescence images. A temporal feedback mechanism is employed between the enhancement and the initial detection steps of a typical direct method. We estimate the minimum and maximum nuclei diameters from the previous frame and feed back them as filter lengths for multiscale enhancement of the current frame. A radial intensity-gradient function is optimized at positions of initial centroids to estimate all nuclei diameters. This procedure continues for processing subsequent images in the sequence. Above mechanism thus ensures proper enhancement by automated estimation of major parameters. This brings robustness and safeguards the system against additive noises and effects from wrong parameters. Later, the method and its single-scale variant are simplified for further reduction of parameters. The proposed method is then extended for nuclei volume segmentation. The same optimization technique is applied to final centroid positions of the enhanced image and the estimated diameters are projected onto the binary candidate regions to segment nuclei volumes.Our method is finally integrated with a simple sequential tracking approach to establish nuclear trajectories in the 4D space. Experimental evaluations with five image-sequences (each having 271 3D sequential images) corresponding to five different mouse embryos show promising performances of our methods in terms of nuclear detection, segmentation, and tracking. A detail analysis with a sub-sequence of 101 3D images from an embryo reveals that the proposed method can improve the nuclei detection accuracy by 9 over the previous methods, which used inappropriate large valued parameters. Results also confirm that the proposed method and its variants achieve high detection accuracies ( 98 mean F-measure) irrespective of the large variations of filter parameters and noise levels.     

The potential of 3D-FISH and super-resolution structured illumination microscopy for studies of 3D nuclear architecture: 3D structured illumination microscopy of defined chromosomal structures visualized by 3D (immuno)-FISH opens new perspectives for studies of nuclear architecture

Three-dimensional structured illumination microscopy (3D-SIM) has opened up new possibilities to study nuclear architecture at the ultrastructural level down to the ~100 nm range. We present first results and assess the potential using 3D-SIM in combination with 3D fluorescence in situ hybridization (3D-FISH) for the topographical analysis of defined nuclear targets. Our study also deals with the concern that artifacts produced by FISH may counteract the gain in resolution. We address the topography of DAPI-stained DNA in nuclei before and after 3D-FISH, nuclear pores and the lamina, chromosome territories, chromatin domains, and individual gene loci. We also look at the replication patterns of chromocenters and the topographical relationship of Xist-RNA within the inactive X-territory. These examples demonstrate that an appropriately adapted 3D-FISH/3D-SIM approach preserves key characteristics of the nuclear ultrastructure and that the gain in information obtained by 3D-SIM yields new insights into the functional nuclear organization.     

Laser scanning confocal microscopy and quantitative microscopy with a charge coupled device camera improve detection of human papillomavirus DNA revealed by fluorescence in situ hybridization

Epithelial cervical CaSki, SiHa and HeLa cells containing respectively 600 copies of human papillomavirus (HPV) DNA type 16, 1–2 copies of HPV DNA type 16 and 10–50 copies of HPV DNA type 18 were used as model to detect different quantities of integrated HPV genome. The HPV DNA was identified on cell deposits with specific biotinylated DNA probes either by enzymatic in situ hybridization (EISH) or fluorescence in situ hybridization (FISH) involving successively a rabbit anti-biotin antibody, a biotinylated goat anti-rabbit antibody and streptavidin-alkaline phosphatase complex or streptavidin-fluorescein isothiocyanate complex. With brightfield microscopy and EISH, hybridization spots were observed in CaSki and HeLa cells but hardly any in SiHa cells. With fluorescence microscopy and FISH, hybridization spots were clearly seen only on CaSki cell nuclei. In an attempt to improve the detection of low quantities of HPV DNA signals revealed by FISH, laser scanning confocal microscopy (LSCM) and quantitative microscopy with an intensified charge coupled device (CCD) camera were used. With both LSCM and quantitative microscopy, as few as 1–2 copies of HPV DNA were detected and found to be confined to cell nuclei counterstained with propidium iodide. Under Nomarski phase contrast, a good preservation of the cell structure was observed. With quantitative microscopy, differences in the number, size, total area and integrated fluorescence intensity of hybridization spots per nucleus were revealed between CaSki, SiHa and HeLa cells. Considered altogether our results shows that in situ hybridization is a powerful technique to detect small amounts of nucleic acid sequences but the choice of the technique for cell examination is important. Single genes of HPV were visualized most efficiently by association of FISH with LSCM or quantitative microscopy with an intensified CCD camera.     

An automatic block and spot indexing with k-nearest neighbors graph for microarray image analysis

MOTIVATION: In this paper, we propose a fully automatic block and spot indexing algorithm for microarray image analysis. A microarray is a device which enables a parallel experiment of ten to hundreds of thousands of test genes in order to measure gene expression. Due to this huge size of experimental data, automated image analysis is gaining importance in microarray image processing systems. Currently, most of the automated microarray image processing systems require manual block indexing and, in some cases, spot indexing. If the microarray image is large and contains a lot of noise, it is very troublesome work. In this paper, we show it is possible to locate the addresses of blocks and spots by applying the Nearest Neighbors Graph Model. Also, we propose an analytic model for the feasibility of block addressing. Our analytic model is validated by a large body of experimental results. RESULTS: We demonstrate the features of automatic block detection, automatic spot addressing, and correction of the distortion and skewedness of each microarray image.     

Automated segmentation of routinely hematoxylin-eosin-stained microscopic images by combining support vector machine clustering and active contour models

OBJECTIVE: To develop a method for the automated segmentation of images of routinely hematoxylin-eosin (H-E)-stained microscopic sections to guarantee correct results in computer-assisted microscopy. STUDY DESIGN: Clinical material was composed 50 H-E-stained biopsies of astrocytomas and 50 H-E-stained biopsies of urinary bladder cancer. The basic idea was to use a support vector machine clustering (SVMC) algorithm to provide gross segmentation of regions holding nuclei and subsequently to refine nuclear boundary detection with active contours. The initialization coordinates of the active contour model were defined using a SVMC pixel-based classification algorithm that discriminated nuclear regions from the surrounding tissue. Starting from the boundaries of these regions, the snake fired and propagated until converging to nuclear boundaries. RESULTS: The method was validated for 2 different types of H-E-stained images. Results were evaluated by 2 histopathologists. On average, 94% of nuclei were correctly delineated. CONCLUSION: The proposed algorithm could be of value in computer-based systems for automated interpretation of microscopic images.     

Automated detection of tunneling nanotubes in 3D images.

BACKGROUND: This paper presents an automated method for the identification of thin membrane tubes in 3D fluorescence images. These tubes, referred to as tunneling nanotubes (TNTs), are newly discovered intercellular structures that connect living cells through a membrane continuity. TNTs are 50-200 nm in diameter, crossing from one cell to another at their nearest distance. In microscopic images, they are seen as straight lines. It now emerges that the TNTs represent the underlying structure of a new type of cell-to-cell communication. METHODS: Our approach for the identification of TNTs is based on a combination of biological cell markers and known image processing techniques. Watershed segmentation and edge detectors are used to find cell borders, TNTs, and image artifacts. Mathematical morphology is employed at several stages of the processing chain. Two image channels are used for the calculations to improve classification of watershed regions into cells and background. One image channel displays cell borders and TNTs, the second is used for cell classification and displays the cytoplasmic compartments of the cells. The method for cell segmentation is 3D, and the TNT detection incorporates 3D information using various 2D projections. RESULTS: The TNT- and cell-detection were applied to numerous 3D stacks of images. A success rate of 67% was obtained compared with manual identification of the TNTs. The digitalized results were used to achieve statistical information of selected properties of TNTs. CONCLUSION: To further explore these structures, automated detection and quantification is desirable. Consequently, this automated recognition tool will be useful in biological studies on cell-to-cell communication where TNT quantification is essential.