CytoPy: An autonomous cytometry analysis framework.

Cytometry analysis has seen a considerable expansion in recent years in the maximum number of parameters that can be acquired in a single experiment. In response to this technological advance there has been an increased effort to develop new computational methodologies for handling high-dimensional single cell data acquired by flow or mass cytometry. Despite the success of numerous algorithms and published packages to replicate and outperform traditional manual analysis, widespread adoption of these techniques has yet to be realised in the field of immunology. Here we present CytoPy, a Python framework for automated analysis of cytometry data that integrates a document-based database for a data-centric and iterative analytical environment. In addition, our algorithm-agnostic design provides a platform for open-source cytometry bioinformatics in the Python ecosystem. We demonstrate the ability of CytoPy to phenotype T cell subsets in whole blood samples even in the presence of significant batch effects due to technical and user variation. The complete analytical pipeline was then used to immunophenotype the local inflammatory infiltrate in individuals with and without acute bacterial infection. CytoPy is open-source and licensed under the MIT license. CytoPy is available at https://github.com/burtonrj/CytoPy, with notebooks accompanying this manuscript (https://github.com/burtonrj/CytoPyManuscript) and software documentation at https://cytopy.readthedocs.io/.

HiPPO and PANDA: Two Bioinformatics Tools to Support Analysis of Mass Cytometry Data.

High-dimensional mass cytometry (Cytometry by Time-Of-Flight; CyTOF) is a multiparametric single-cell approach that allows for more than 40 parameters to be evaluated simultaneously, opening the possibility to dissect cellular heterogeneity and elucidate functional interactions between different cell types. However, the complexity of these data makes analysis and interpretation daunting. We created High-throughput Population Profiler (HiPPO), a tool that reduces the complexity of the CyTOF data and allows homogeneous clusters of cells to be visualized in an intuitive manner. Each subpopulation is mapped to the Population Analysis Database (PANDA), an open-source, manually curated database containing protein expression profiles for selected markers of primary cells, allowing for cell type abundance in the analyzed samples to be monitored. Custom cell definitions can be submitted for targeted identifications. All cell clusters, regardless of their annotation status, are available for further analyses. HiPPO also conducts nonparametric tests to determine whether differences in protein expression levels between conditions are significant. HiPPO strikes a balance between diagnostic power and computational burden. Its minimal computational footprint allows for subpopulations in a heterogeneous sample to be identified and quantified quickly.

New Antifungal Susceptibility Test Based on Chitin Detection by Image Cytometry.

The antifungal susceptibility tests used in clinical laboratories have several limitations. We developed a new test, SensiFONG, based on the detection of chitin levels after exposure to antifungal drugs. The optimal culture conditions were 30°C for 6 h for yeast strains and 26°C for 16 h for molds. The strains were exposed to a range of echinocandin or azole concentrations. Chitin was stained with calcofluor white. The percentage of fungal cells with high chitin levels was determined with an automatic epifluorescence microscope. The SensiFONG results were compared to those with the EUCAST method. Image acquisition and analysis were performed with ScanR software. Fifty-nine strains (28 Candida albicans, 17 Candida glabrata, and 14 Aspergillus fumigatus) were analyzed. Thresholds for the classification of strains as resistant or susceptible were determined for each fungal species. The strains displaying an increase in chitin content of ≥32% for C. albicans, ≥6% for C. glabrata, and ≥17% for A. fumigatus were considered susceptible. The application of these thresholds to all 59 strains resulted in a sensitivity of 0.87, 0.93, and 1.00 and a specificity of 0.93, 0.84, and 0.82 for C. albicans, C. glabrata, and A. fumigatus, respectively. The correlation between the results obtained in the SensiFONG and EUCAST assays was excellent. We developed a new test, SensiFONG, based on a new concept. While current assays assess growth inhibition, our test detects changes in chitin levels after exposure to antifungal drugs. Here, we present preliminary results and we propose a proof of concept of this methodology.

CytoBinning: Immunological insights from multi-dimensional data.

New cytometric techniques continue to push the boundaries of multi-parameter quantitative data acquisition at the single-cell level particularly in immunology and medicine. Sophisticated analysis methods for such ever higher dimensional datasets are rapidly emerging, with advanced data representations and dimensional reduction approaches. However, these are not yet standardized and clinical scientists and cell biologists are not yet experienced in their interpretation. More fundamentally their range of statistical validity is not yet fully established. We therefore propose a new method for the automated and unbiased analysis of high-dimensional single cell datasets that is simple and robust, with the goal of reducing this complex information into a familiar 2D scatter plot representation that is of immediate utility to a range of biomedical and clinical settings. Using publicly available flow cytometry and mass cytometry datasets we demonstrate that this method (termed CytoBinning), recapitulates the results of traditional manual cytometric analyses and leads to new and testable hypotheses.

Visual analysis of mass cytometry data by hierarchical stochastic neighbour embedding reveals rare cell types.

Mass cytometry allows high-resolution dissection of the cellular composition of the immune system. However, the high-dimensionality, large size, and non-linear structure of the data poses considerable challenges for the data analysis. In particular, dimensionality reduction-based techniques like t-SNE offer single-cell resolution but are limited in the number of cells that can be analyzed. Here we introduce Hierarchical Stochastic Neighbor Embedding (HSNE) for the analysis of mass cytometry data sets. HSNE constructs a hierarchy of non-linear similarities that can be interactively explored with a stepwise increase in detail up to the single-cell level. We apply HSNE to a study on gastrointestinal disorders and three other available mass cytometry data sets. We find that HSNE efficiently replicates previous observations and identifies rare cell populations that were previously missed due to downsampling. Thus, HSNE removes the scalability limit of conventional t-SNE analysis, a feature that makes it highly suitable for the analysis of massive high-dimensional data sets.

A high-throughput system for segmenting nuclei using multiscale techniques.

Automatic segmentation of cell nuclei is critical in several high-throughput cytometry applications whereas manual segmentation is laborious and irreproducible. One such emerging application is measuring the spatial organization (radial and relative distances) of fluorescence in situ hybridization (FISH) DNA sequences, where recent investigations strongly suggest a correlation between nonrandom arrangement of genes to carcinogenesis. Current automatic segmentation methods have varying performance in the presence of nonuniform illumination and clustering, and boundary accuracy is seldom assessed, which makes them suboptimal for this application. The authors propose a modular and model-based algorithm for extracting individual nuclei. It uses multiscale edge reconstruction for contrast stretching and edge enhancement as well as a multiscale entropy-based thresholding for handling nonuniform intensity variations. Nuclei are initially oversegmented and then merged based on area followed by automatic multistage classification into single nuclei and clustered nuclei. Estimation of input parameters and training of the classifiers is automatic. The algorithm was tested on 4,181 lymphoblast nuclei with varying degree of background nonuniformity and clustering. It extracted 3,515 individual nuclei and identified single nuclei and individual nuclei in clusters with 99.8 +/- 0.3% and 95.5 +/- 5.1% accuracy, respectively. Segmented boundaries of the individual nuclei were accurate when compared with manual segmentation with an average RMS deviation of 0.26 microm (approximately 2 pixels). The proposed segmentation method is efficient, robust, and accurate for segmenting individual nuclei from fluorescence images containing clustered and isolated nuclei. The algorithm allows complete automation and facilitates reproducible and unbiased spatial analysis of DNA sequences.

Implementation and evaluation of a new automated interactive image analysis system.

OBJECTIVE: To compare automated interactive screening using the ThinPrep Imaging System with independent manual primary screening of 12,000 routine ThinPrep slides. STUDY DESIGN: With the first 6,000 cases, the Review Scopes (RS) screening results from the 22 fields of view (FOV) only were compared to independent manual primary screening. In the next 6,000 cases, any abnormality detected in the 22 FOV resulted in full manual screening on the cytotechnologist's own microscope. Sensitivity and specificity together with their 95% CIs were calculatedfor each method. RESULTS: In the first set of 6, 000 cases, diagnostic sensitivity and specificity of the imager were 85.19% and 96.67%, respectively. The diagnostic sensitivity and specificity of manual primary screening were 89.38% and 98.42%. This highersensitivity and specificity of manual primary screening were found to be statistically significant. The second set of 6,000 cases demonstrated no significant statistical difference in sensitivity or specificity between the sets of data. CONCLUSION: The results from our study show that the sensitivity and specificity of the imager technology are equivalent to those of manual primary screening. The system is ideally suited to the rapid screening of negative cases, allowing increased laboratory productivity and greater throughput of cases on a daily basis.

Extracting quantitative information from tissue--an industrial perspective.

The focus of this article is to provide an overview of the current technologies for the pharmaceutical and biotech industry. Disease processes express themselves in the functional and structural disturbance of cellular systems. Cells and their metabolites constitute the building blocks of tissues and entire organisms. Studying the spatial and temporal phenotype of disease processes in tissues at the cellular level reveals a multitude of information about the progress and status of a disease. Detailed exploration of tissues by slide-based cytometry is an important source of information about disease processes. Technological and analytical advances allow us to shed a new light on tissues and to come to a better understanding of the complexity of disease processes. Dealing with complex multidimensional datasets from tissue samples requires an advanced approach to image processing and data management. The increase in computing power and the continuing research into imaging algorithms allow us to improve the exploration of the data content of tissues.

Multispectral/hyperspectral image enhancement for biological cell analysis.

BACKGROUND: Microscopes form projected images from illuminated objects, such as cellular tissue, which are recorded at a distance through the optical system's field of view. A telescope on a satellite or airplane also forms images with a similar optical projection of objects on the ground. Typical visible illuminations form a displayed set of three-color channels (Red Green Blue [RGB]) that are combined from three image sensor arrays (e.g., focal plane arrays) into a single pixel coding for each color present in the image. Analysis of these RGB color images develops a qualitative image representation of the objects. METHODS: Independent component analysis (ICA) is used for analysis and enhancement of multispectral images, and compared with the similar and widely used principal component analysis. RESULTS: The data examples indicate that the ICA enhancement, and the resulting RGB image combination display, can be useful in processing datacubes of cellular data where isolation of unknown subtle image elements representing objects is desired. CONCLUSIONS: ICA image enhancement can aid processing of datacubes of cellular data by clarifying subtle image elements. These parallelizable algorithms can be implemented for real-time, online analysis.

Comparison of a visual to a computer-assisted technique for detecting apoptosis.

In many studies, fluorescent dyes (ethidium bromide [EB] and acridine orange [AO]) are used to stain DNA to determine if nuclei are apoptotic. However, there are numerous visual methods for counting these stained DNA that may lead to inaccuracies Measuring apoptosis by the visual counting method may be imprecise because of the variability of individuals' perception of color. Therefore, the authors compared a visual method of counting chromatin for apoptosis with a method relying on a computer program. They began counting chromatin using the visual method, in which individuals identify the stained DNA using their own visual perception. For comparison, they used a software-based counting method (analySIS software) to determine the color (hue) of the stained DNA. Using the numeric hue values from the software eliminates the variations in human color perception. Intra and interrater reliability of the visual and computer-assisted counting methods were evaluated with Spearman's. The authors found statistical significance in the intrarater reliability (r = 1.0, P = 0.0001 for all chromatin categories) and interrater reliability (r = 0.975, P = 0.005 for both readings) when using the software program. No statistical significance was found for the visual counting method, indicating inaccuracy between and within raters. Thus, the computer-assisted counting method of identifying the damaged DNA is more accurate and precise than the individual's visual perception of color. Based on these data, apoptosis measurements using color staining with EB and AO should be determined using hue values generated by a computer program and not by a researcher's visual assessment.

Flow and image cytometry in thymic neoplasia: correlation with clinical outcome.

OBJECTIVE: To compare nuclear DNA by flow (FCM) and image cytometry (ICM) in thymic neoplasms and to relate results to clinical outcome. STUDY DESIGN: DNA ploidy of 44 thymomas and 6 thymic carcinomas was studied by FCM and ICM of single nuclear suspensions from paraffin blocks. RESULTS: By FCM, 33 thymomas (75%) and one thymic carcinoma (17%) were diploid; 6 thymomas (14%) and 4 thymic carcinomas (67%) were aneuploid. By ICM, 36 thymomas (82%) were diploid; 7 thymomas (16%) and 6 thymic carcinomas (100%) were aneuploid. Mean follow-up in 44 cases was 46.2 months (range, 1-162). Ten patients with persistent/recurrent disease included four with thymic carcinoma, who died of the disease (two aneuploid by both techniques, two aneuploid by ICM with unsatisfactory/diploid FCM). Four had invasive thymoma and recurrence after 13-150 months (two diploid and two aneuploid by both methods), one had diploidy and noninvasive thymoma that recurred at 92 months, and one had an epithelial thymoma that recurred at 144 months (aneuploid by FCM, diploid by ICM). CONCLUSION: The results obtained in this preliminary, retrospective study show a high concordance between FCM and ICM; aneuploidy correlated with poor outcome by both methodologies. While these findings are encouraging, larger numbers of cases will be needed to define the role of FCM and ICM in predicting outcome in thymic tumors.

Correlation between silver-stained nucleolar organizer region area and cell cycle time.

BACKGROUND: The relationship between the population doubling time and the quantity of silver-stained nucleolar organizer region (AgNOR) interphase proteins was studied in cell culture at three different temperatures used to modulate the cell cycle duration. METHODS: After MIB 1 and AgNOR combined staining, the quantity of AgNOR proteins was measured in cycling cells by image cytometry. RESULTS: Among the several parameters calculated, the AgNOR relative area showed a strong correlation with the changes of the population doubling time induced by different temperatures. CONCLUSIONS: The results support the hypothesis that the cell cycle time and the size of the ribogenesis machinery are coregulated and that measurements of AgNORs can thus be used as a static evaluation of the cell cycle duration in arbitrary units.

Regular form factor. A new concept and calculating method for quantitative form description.

OBJECTIVE: To quantitatively describe the shape of two-dimensional particles based on combined consideration of area (A), circumference (C), and major and minor diameter (d1 and d2, respectively). STUDY DESIGN: A regular form factor (RFF), a new concept and calculating method for form description, was defined as the extent to which a structure differs from an ellipse or circle of which the major and minor diameters, respectively, are equal to the described structure. Based on the definition of RFF, formulae were deduced in accordance with mathematical principles. RESULTS: The deduced formulae were as follows: RFF = A [1.5(a + b) - square root of (a b)]/(a b C) and RFF = A [3 (d1 + d2) - 2 square root of (d1 d2)]/(d1 d2 C). In the formulae, a = d1/2, and b = d2/2. RFF = 1 for ellipses and circles, and < 1 for irregular particles. The more similar to an ellipse or circle, the closer to 1 is RFF. The more irregular the structures are, the smaller the values of RFF. An error analysis was also made, and a rectified formula for RFF was deduced. An example is given to show how to calculate RFF. A practical test of RFF was made in colorectal adenoma and adenocarcinoma in human material, and the results show that RFF of glands of colorectal adenoma and adenocarcinoma have significant differences between normal, adenoma and adenocarcinoma. CONCLUSION: RFF was compared with the parameter Form AR, and the results showed that RFF is more sensitive in describing the irregularity of a structure than Form AR. RFF is of practical significance for describing structures in human material.

Nuclear image morphometry and cytologic grade of breast carcinoma.

OBJECTIVE: To correlate visual cytologic grade with automated nuclear morphometry of carcinoma of the breast. STUDY DESIGN: We randomly selected 24 histologically proven infiltrating ductal carcinomas of the breast and 10 benign breast lesions (fibroadenoma). Hematoxylin-eosin-stained fine needle aspiration cytology (FNAC) smears were selected for both cytologic grade and automated image morphometry. The same hematoxylin-eosin-stained FNAC smears were studied for area, convex area, standard deviation of nuclear area, diameter, perimeter and convex perimeters of nucleus. At least 100 cells from each case were measured with an image cytometer. RESULTS: Mean nuclear area, standard deviation of nuclear area, nuclear diameter, convex area, convex perimeter and perimeter were significantly increased from benign versus grade 1 carcinomas and grade 1 versus grade 2 and 3 carcinomas (one way ANOVA test). However, there was no significant difference in grade 2 versus grade 3 carcinomas. CONCLUSION: Automated image cytometry rapidly and successfully measures various nuclear parameters. The measurement of various nuclear parameters would be helpful in future applications of automated diagnosis and grading of breast carcinomas from cytologic material.

Automatic morphological sieving: comparison between different methods, application to DNA ploidy measurements.

The aim of the present study is to propose alternative automatic methods to time consuming interactive sorting of elements for DNA ploidy measurements. One archival brain tumour and two archival breast carcinoma were studied, corresponding to 7120 elements (3764 nuclei, 3356 debris and aggregates). Three automatic classification methods were tested to eliminate debris and aggregates from DNA ploidy measurements (mathematical morphology (MM), multiparametric analysis (MA) and neural network (NN)). Performances were evaluated by reference to interactive sorting. The results obtained for the three methods concerning the percentage of debris and aggregates automatically removed reach 63, 75 and 85% for MM, MA and NN methods, respectively, with false positive rates of 6, 21 and 25%. Information about DNA ploidy abnormalities were globally preserved after automatic elimination of debris and aggregates by MM and MA methods as opposed to NN method, showing that automatic classification methods can offer alternatives to tedious interactive elimination of debris and aggregates, for DNA ploidy measurements of archival tumours.

Total number and size distribution of hepatic metastases of carcinoma.

OBJECTIVE: To quantify the susceptibility of carcinoma to hepatic metastases by studying autopsy livers with carcinoma metastases, the primary sites of which were mostly the digestive organs. STUDY DESIGN: We developed a stereologic method of estimating the total number, N, and the size distribution of metastatic tumors in the liver based on a geometric model of spherical nodules with varying radii, r. This method proved to be sufficiently reliable by disector analysis simultaneously performed in some cases; it gave an approximate result. This method was applied to the 31 autopsy cases. Correlation and regression analyses were performed among N, the mean radius of nodules, rmean, and conventional pathologic features of the primary tumor. RESULTS: The estimates of N ranged from 10 to 3.2 x 10(5). A close negative correlation was confirmed between N and rmean. Neither significant correlation nor regression was observed among N and the other pathologic factors of the primary tumors. CONCLUSION: N turned out to serve as a useful index for evaluating the metastatic potential of a carcinoma. However, investigation has yet to be made to determine biologic factors in the primary tumor closely associated with N.

Application of confocal scanning laser microscopy to 3-D DNA image cytometry of prostatic lesions.

OBJECTIVE: To explore the potential of three-dimensional (3-D) image cytometry for the measurement of DNA content in prostatic specimens using confocal scanning laser microscopy (CSLM) and 3-D image analysis. STUDY DESIGN: Thick tissue slices (100 microns), stained for DNA with chromomycin A3, from four patients with benign hyperplasia, prostatic intraepithelial neoplasia (PIN), well- and poorly differentiated adenocarcinoma of the prostate, were studied. Two different blocks from the same slice were studied for each case. Cell nuclei were segmented automatically. DNA content and nuclear volume were measured. RESULTS: DNA histograms showed a single peak in the diploid range for the hyperplasia and PIN cases. For the case of well-differentiated carcinoma, two peaks were observed, one in the diploid range and one in the tetraploid range. The two peaks were observed on two different blocks of the same tissue slice. Poorly differentiated carcinoma was characterized by an aneuploid distribution. For the cases of PIN and carcinoma, we observed a considerable variation in nuclear volume. CONCLUSION: The results indicate the potential of 3-D image cytometry for the measurement of DNA content in prostatic specimens while preserving tissue architecture.

Glomerular hypertrophy as a prognostic marker in childhood IgA nephropathy.

A clinicopathological and morphometric analysis of glomerular hypertrophy (GH) was conducted using biopsies obtained from 52 selected pediatric patients with IgA nephropathy (IgAN). Of the 52 patients, consisting of 12 with chronic renal failure (CRF) and 40 without CRF, various clinical and morphometric parameters were compared to 10 controls with benign hematuria. The mean glomerular tuft size, mesangial area, and interstitial area all significantly increased in patients with poor prognosis when compared to the non-CRF-IgAN cases and the control cases. The glomerular capillary loop size was also significantly greater in CRF-IgAN than in non-CRF-IgAn patients (1.37 times) and the controls (1.55 times). The 10-year renal survival rates of patients with 'large' loop size (>1.55-fold) were significantly lower (p < 0.001) than those of patients with a smaller capillary loop size. The size of the capillary loops was directly related to the relative interstitial area (Aint) (r2 = 0.43, p < 0.001), to the degree of glomerulosclerosis (GS; r2 = 0.348, p < 0.001) and the mesangial area (r2 = 0.326, p < 0.001). Proteinuria tightly correlated with the capillary loop size (r2 = 0.374, p < 0.001). It was not unexpected that a strong relationship was detected between the serum creatinine level and Aint (r2 = 0.452, p < 0.001) and the percentage of GS (r2 = 0.342, p < 0.001). In IgAN the percentage of GS correlated significantly with Aint (r2 = 0.484, p < 0.001). GH, which was manifested by glomerular capillary loop dilatation, shows a close correlation with the interstitial expansion, degree of GS and mesangial enlargement. These data suggest that both extra- and intraglomerular hemodynamic changes followed by primary glomerular damage thus lead to capillary dilatation of the intact glomeruli as a morphological manifestation of GH and therefore such changes play a key role in the progression of IgAN.

Automated feature extraction and identification of colon carcinoma.

OBJECTIVE: To assess an automated algorithm, developed for the classification of normal and cancerous colonic mucosa, using geometric analysis of features and texture analysis. STUDY DESIGN: Twenty-one images were analyzed, 10 from normal and 11 from cancerous mucosa. The classification was based on a regularity index dependent on shape, object orientation for establishing parallelism and five texture features derived using the co-occurrence image analysis method. RESULTS: Geometric analysis yielded an overall classification accuracy of 80%. The corresponding sensitivity and specificity were 94% and 64%, respectively. Using texture analysis, the overall classification accuracy was 90%, with a sensitivity and specificity of 82% and 100%, respectively. CONCLUSION: This initial study demonstrated that geometric and texture analysis techniques show promise for automated analysis of colon cancer.