A Spot Modeling Evolutionary Algorithm for Segmenting Microarray Images

Microarray image compression: SLOCO and the effect of information loss

We read with interest the recently published pilot study on digital whole slide imaging (WSI) for cervicovaginal cytology by Wright and colleagues.1 Their study employed 7 cytologists to review 11 Papanicolaou test slides using digital and conventional light microscope modalities. We acknowledge that their data indicate greater diagnostic accuracy and shorter mean times to reach diagnoses with Papanicolaou test glass slides. However, we would like to express our concern regarding the design of that study and the subsequent conclusion drawn, based on those data, that WSI is not yet ready for routine cytology use.Although we recognize that this was a pilot study, in our opinion, 11 cases is insufficient to reliably measure discordance between digital and conventional microscopy nor is it enough to ensure that users are familiar with a new technology. Prior validation studies comparing WSI and glass slides have demonstrated that, when too few cases are used, the diagnostic accuracy deteriorates. A meta-analysis of the literature indicates that, when an average of 20 cases was employed, the concordance between WSI and glass slides was 75% versus 95% when 60 cases were used.2 Furthermore, although the authors do not specify whether any WSI training was provided in their study, it is unlikely that those cytologists received equivalent exposure to digitized slides as they have had in handling glass slides during their careers. Indeed, previous published data show that accuracy with WSI was 79% without training pathologists to use that technology, compared with 95% when they did receive training.2 The authors do comment that their disagreement rates may, in part, be explained by their lack of familiarity with this new technology.The authors further remark about the current lack of available commercial technology to scan the edges of a glass slide, to help methodically screen WSI slides, and the inability of systems to handle large digital files. We would like to state that there are WSI scanners that are capable of scanning the entire glass slide (Figure), even including the label if required. Their study does highlight an important point, however, ensuring that all material present on a glass slide to be scanned is included in the digital image. Although we admit that it can be challenging to screen a digitized cytology slide, we also feel that it is important to point out that the task can be made much easier using currently available WSI viewer tools for easy and systematic navigation of virtual slides (such as thumbnails, autopanning, keyboard navigation, and tracker capabilities).3 It is important to train users to make use of these tools when viewing virtual slides. It is true that scanning a glass slide generates large WSI digital files. That is why vendors of those WSI devices specify that users employ adequate computers and monitors to handle and view the digital images. The computer workstation (1 gigabyte, random access memory, 64 video random access memory [1 GB RAM, 64 VRAM]) and display monitor (1600 by 1024 pixel resolution) the authors selected for their study seems to have less than the minimum required specifications for the iScan Coreo (Ventana Medical Systems, Tucson, Arizona) instrument they used (2 GB RAM and 1920 by 1200 pixel resolution). That may explain why their system frequently “froze,” causing delay and frustration for their participants. For digital imaging, the workstation's specifications should to be at least, if not higher, than those recommended by the vendor to achieve proper performance and user satisfaction.In conclusion, although we agree with Wright et al that there are still several obstacles to overcome, we believe that, if used effectively, WSI will soon be adopted for daily cytology diagnostic use. To accomplish that, laboratories will need to first invest in the necessary technology, appropriately validate those systems, and adequately train their cytologists to use them.

The accuracy of gene expression measurements generated using cDNA microarrays is dependent on the quality of the image generated following hybridization of fluorescently labelled cDNA. It is not known how this image is influenced by sample preparation factors which such as RNA quality, cDNA synthesis and labelling efficiency. In this study we used a simple metric based on the ratio of the total feature (F) and background (B) fluorescence, which correlates with the visual assessment of 60 microarray images, to determine the influence of sample preparation on image quality. Results indicate that RNA purity (A260/A280) and integrity (18S:28S ratio) do not strongly influence microarray image quality. cDNA having an nucleotide to dye ratio greater than 100 produced poor microarray images, however, cDNA labelled more efficiently was not a guarantee of a better image. The data also indicate that the array image quality is not improved by loading more cDNA into the hybridization mixture however poor image quality did result from a disproportionate amounts of Cy5 and Cy3 labelled cDNA. This study provides insight into the source of variation in microarray image analysis introduced during sample preparation and will assist in the standardisation of cDNA glass slide microarray protocols.

Commentary: Can pathologists interpret digital images as well as they interpret microscope slides?

Microarray image technology is a powerful tool for monitoring the expression of thousands of genes simultaneously. Each microarray experiment produces immense amounts of image data, and efficient storage and transmission requires compression that utilizes the microarray image's structure and unique analysis goals. We have developed a progressive compression scheme for microarray images which can be either lossy or lossless. Our scheme has a coded data structure that allows fast decoding and reprocessing of image subsets, and includes summary statistics and image segmentation information. Since visual fidelity is not the end goal for microarray images, we introduce a new measure of distortion for lossy compression: the sensitivity of microarray information extraction to compression loss. We find that a lossy compression ratio of 8:1 for cDNA microarrays minimally affects downstream processing. The average lossless compression ratio is 1.83:1 for cDNA images and 2.43:1 for inkjet images, comparable to state-of-the-art lossless schemas, yet with added flexibility and information.

<p>Microarray technology allows the simultaneous monitoring of thousands of genes. Based on the gene expression measurements, microarray technology have proven powerful in gene expression profiling for discovering new types of diseases and for predicting the type of a disease. Gridding, segmentation and intensity extraction are the three important steps in microarray image analysis. Clustering algorithms have been used for microarray image segmentation with an advantage that they are not restricted to a particular shape and size for the spots. Instead of using single feature clustering algorithm, this paper presents multiple feature clustering algorithm with three features for each pixel such as pixel intensity, distance from the center of the spot and median of surrounding pixels. In all the traditional clustering algorithms, number of clusters and initial centroids are randomly selected and often specified by the user. In this paper, a new algorithm based on empirical mode decomposition algorithm for the histogram of the input image will generate the number of clusters and initial centroids required for clustering. It overcomes the shortage of random initialization in traditional clustering and achieves high computational speed by reducing the number of iterations. The experimental results show that multiple feature Fuzzy C-means has segmented the microarray image more accurately than other algorithms.</p>

Advances in cereal genomics and applications in crop breeding

The study evaluated the reproducibility of telecytology diagnosis of cervical smears on a randomly selected 50 cases under the conditions of Georgia. Fifty cervical smears (benign, 14; atypical squamous cells of undetermined significance [ASCUS], 14; low-grade squamous intraepithelial lesion [LSIL], 10; high-grade squamous intraepithelial lesion [HSIL], 12) were selected. The digital images were captured at a maximum resolution of 2048 × 1536 pixels and transmitted by electronic mail. Diagnosis of glass slides and digital images was done independently in a double-blind manner by three cytologists, versus the diagnosis of digital images followed by diagnosis of glass slides 3 months later. The procedure was repeated after 3 months. Diagnoses were recorded as benign, ASCUS, LSIL, and HSIL. Diagnostic accuracy and interobserver reproducibility were analyzed using an interclass correlation coefficient, which revealed good interobserver agreement for the first (0.82) and second (0.68) glass slide diagnoses and the first (0.80) and second (0.66) digital image diagnoses. The kappa values for interobserver variation between first and second glass slide diagnoses and first and second digital image diagnoses showed good to excellent agreement. Digital images are suitable substitutes for glass slides; telecytology can be used as an alternative method for the cytologic diagnosis of cervical smears, particularly in quality assurance programs.

We developed methods for characterizing a set of microarray images and for subsequently simulating microarray images with statistical properties similar to those in the original set. Characterization involved measuring properties of individual spots and performing analysis of variance to determine the relative contributions of individual pins used for printing and individual slides to the variation observed in spot physical properties, slide background properties, and intensity of individual genes. Slide backgrounds and individual spot nonuniformities were modeled as 2D causal Markov random fields, and parameters for these were derived from the set of real images. The results of the characterization were then used to generate realistic replicates of the original dataset that can be used for evaluating microarray data processing and analysis techniques. We demonstrated the process on a set of microarray images derived from a mouse kidney experiment. The characterization of these images showed that slides from two of the six mice have significantly different spot properties from the rest. Simulated images from the set are shown to realistically model most properties of the slides, save for large handling defects. We concluded that characterization should be an important part of any microarray experiment to maintain quality control, and that realistic simulations of microarray images can be produced using these methods.

DNA microarrays are now widely used to measure expression levels and DNA copy number in biological samples. Ratios of relative abundance of nucleic acids are derived from images of regular arrays of spots containing target genetic material to which fluorescently labeled samples are hybridized. Whereas there are a number of methods in use for the quantification of images, many of the software systems in wide use either encourage or require extensive human interaction at the level of individual spots on arrays. We present a fully automatic system for microarray image quantification. The system automatically locates both subarray grids and individual spots, requiring no user identification of any image coordinates. Ratios are computed based on explicit segmentation of each spot. On a typical image of 6000 spots, the entire process takes less than 20 sec. We present a quantitative assessment of performance on multiple replicates of genome-wide array-based comparative genomic hybridization experiments. By explicitly identifying the pixels in each spot, the system yields more accurate estimates of ratios than systems assuming spot circularity. The software, called, runs on Windows platforms and is available free of charge for academic use.

The gold standard for enrollment and endpoint assessment in metabolic dysfunction‐associated steatosis clinical trials is histologic assessment of a liver biopsy performed on glass slides. However, obtaining the evaluations from several expert pathologists on glass is challenging, as shipping the slides around the country or around the world is time‐consuming and comes with the hazards of slide breakage. This study demonstrated that pathologic assessment of disease activity in steatohepatitis, performed using digital images on the AISight whole slide image management system, yields results that are comparable to those obtained using glass slides. The accuracy of scoring for steatohepatitis (nonalcoholic fatty liver disease activity score ≥4 with ≥1 for each feature and absence of atypical features suggestive of other liver disease) performed on the system was evaluated against scoring conducted on glass slides. Both methods were assessed for overall percent agreement with a consensus “ground truth” score (defined as the median score of a panel of three pathologists’ glass slides). Each case was also read by three different pathologists, once on glass and once digitally with a minimum 2‐week washout period between the modalities. It was demonstrated that the average agreement across three pathologists of digital scoring with ground truth was noninferior to the average agreement of glass scoring with ground truth [noninferiority margin: −0.05; difference: −0.001; 95% CI: (−0.027, 0.026); and p < 0.0001]. For each pathologist, there was a similar average agreement of digital and glass reads with glass ground truth (pathologist A, 0.843 and 0.849; pathologist B, 0.633 and 0.605; and pathologist C, 0.755 and 0.780). Here, we demonstrate that the accuracy of digital reads for steatohepatitis using digital images is equivalent to glass reads in the context of a clinical trial for scoring using the Clinical Research Network scoring system.

BackgroundThe quality of cDNA microarray data is crucial for expanding its application to other research areas, such as the study of gene regulatory networks. Despite the fact that a number of algorithms have been suggested to increase the accuracy of microarray gene expression data, it is necessary to obtain reliable microarray images by improving wet-lab experiments. As the first step of a cDNA microarray experiment, spotting cDNA probes is critical to determining the quality of spot images.ResultsWe developed a governing equation of cDNA deposition during evaporation of a drop in the microarray spotting process. The governing equation included four parameters: the surface site density on the support, the extrapolated equilibrium constant for the binding of cDNA molecules with surface sites on glass slides, the macromolecular interaction factor, and the volume constant of a drop of cDNA solution. We simulated cDNA deposition from the single model equation by varying the value of the parameters. The morphology of the resulting cDNA deposit can be classified into three types: a doughnut shape, a peak shape, and a volcano shape. The spot morphology can be changed into a flat shape by varying the experimental conditions while considering the parameters of the governing equation of cDNA deposition. The four parameters were estimated by fitting the governing equation to the real microarray images. With the results of the simulation and the parameter estimation, the phenomenon of the formation of cDNA deposits in each type was investigated.ConclusionThis study explains how various spot shapes can exist and suggests which parameters are to be adjusted for obtaining a good spot. This system is able to explore the cDNA microarray spotting process in a predictable, manageable and descriptive manner. We hope it can provide a way to predict the incidents that can occur during a real cDNA microarray experiment, and produce useful data for several research applications involving cDNA microarrays.

Immunohistochemical (IHC) α-synuclein (Asyn) pathology in peripheral biopsies may be a biomarker of Parkinson disease (PD). The multi-center Systemic Synuclein Sampling Study (S4) is evaluating IHC Asyn pathology within skin, colon and submandibular gland biopsies from 60 PD and 20 control subjects. Asyn pathology is being evaluated by a blinded panel of specially trained neuropathologists. Preliminary work assessed 2 candidate immunoperoxidase methods using a set of PD and control autopsy-derived sections from formalin-fixed, paraffin-embedded blocks of the 3 tissues. Both methods had 100% specificity; one, utilizing the 5C12 monoclonal antibody, was more sensitive in skin (67% vs 33%), and was chosen for further use in S4. Four trainee neuropathologists were trained to perform S4 histopathology readings; in subsequent testing, their scoring was compared to that of the trainer neuropathologist on both glass slides and digital images. Specificity and sensitivity were both close to 100% with all readers in all tissue types on both glass slides and digital images except for skin, where sensitivity averaged 75% with digital images and 83.5% with glass slides. Semiquantitative (0-3) density score agreement between trainees and trainer averaged 67% for glass slides and 62% for digital images.

Trichrome, Congo red, and thioflavin S stains are comparable for the diagnosis of amyloid deposits in endomyocardial biopsies.

BackgroundMicroarray image analysis processes scanned digital images of hybridized arrays to produce the input spot-level data for downstream analysis, so it can have a potentially large impact on those and subsequent analysis. Signal saturation is an optical effect that occurs when some pixel values for highly expressed genes or peptides exceed the upper detection threshold of the scanner software (216 - 1 = 65, 535 for 16-bit images). In practice, spots with a sizable number of saturated pixels are often flagged and discarded. Alternatively, the saturated values are used without adjustments for estimating spot intensities. The resulting expression data tend to be biased downwards and can distort high-level analysis that relies on these data. Hence, it is crucial to effectively correct for signal saturation.ResultsWe developed a flexible mixture model-based segmentation and spot intensity estimation procedure that accounts for saturated pixels by incorporating a censored component in the mixture model. As demonstrated with biological data and simulation, our method extends the dynamic range of expression data beyond the saturation threshold and is effective in correcting saturation-induced bias when the lost information is not tremendous. We further illustrate the impact of image processing on downstream classification, showing that the proposed method can increase diagnostic accuracy using data from a lymphoma cancer diagnosis study.ConclusionsThe presented method adjusts for signal saturation at the segmentation stage that identifies a pixel as part of the foreground, background or other. The cluster membership of a pixel can be altered versus treating saturated values as truly observed. Thus, the resulting spot intensity estimates may be more accurate than those obtained from existing methods that correct for saturation based on already segmented data. As a model-based segmentation method, our procedure is able to identify inner holes, fuzzy edges and blank spots that are common in microarray images. The approach is independent of microarray platform and applicable to both single- and dual-channel microarrays.