Bartlomiej Rajwa

Analysis of orientations of collagen fibers by novel fiber-tracking software.

Recent evidence supports the notion that biological functions of extracellular matrix (ECM) are highly correlated to not only its composition but also its structure. This article integrates confocal microscopy imaging and image-processing techniques to analyze the microstructural properties of ECM. This report describes a two- and three-dimensional fiber middle-line tracing algorithm that may be used to quantify collagen fibril organization. We utilized computer simulation and statistical analysis to validate the developed algorithm. These algorithms were applied to confocal images of collagen gels made with reconstituted bovine collagen type I, to demonstrate the computation of orientations of individual fibers.

A non-parametric Bayesian model for joint cell clustering and cluster matching: identification of anomalous sample phenotypes with random effects

BackgroundFlow cytometry (FC)-based computer-aided diagnostics is an emerging technique utilizing modern multiparametric cytometry systems.The major difficulty in using machine-learning approaches for classification of FC data arises from limited access to a wide variety of anomalous samples for training. In consequence, any learning with an abundance of normal cases and a limited set of specific anomalous cases is biased towards the types of anomalies represented in the training set. Such models do not accurately identify anomalies, whether previously known or unknown, that may exist in future samples tested. Although one-class classifiers trained using only normal cases would avoid such a bias, robust sample characterization is critical for a generalizable model. Owing to sample heterogeneity and instrumental variability, arbitrary characterization of samples usually introduces feature noise that may lead to poor predictive performance. Herein, we present a non-parametric Bayesian algorithm called ASPIRE (anomalous sample phenotype identification with random effects) that identifies phenotypic differences across a batch of samples in the presence of random effects. Our approach involves simultaneous clustering of cellular measurements in individual samples and matching of discovered clusters across all samples in order to recover global clusters using probabilistic sampling techniques in a systematic way.ResultsWe demonstrate the performance of the proposed method in identifying anomalous samples in two different FC data sets, one of which represents a set of samples including acute myeloid leukemia (AML) cases, and the other a generic 5-parameter peripheral-blood immunophenotyping. Results are evaluated in terms of the area under the receiver operating characteristics curve (AUC). ASPIRE achieved AUCs of 0.99 and 1.0 on the AML and generic blood immunophenotyping data sets, respectively.ConclusionsThese results demonstrate that anomalous samples can be identified by ASPIRE with almost perfect accuracy without a priori access to samples of anomalous subtypes in the training set. The ASPIRE approach is unique in its ability to form generalizations regarding normal and anomalous states given only very weak assumptions regarding sample characteristics and origin. Thus, ASPIRE could become highly instrumental in providing unique insights about observed biological phenomena in the absence of full information about the investigated samples.

The design and construction of a cost-efficient confocal laser scanning microscope

The optical dissection ability of confocal microscopy makes it a powerful tool for biological materials. However, the cost and complexity of confocal scanning laser microscopy hinders its wide application in education. We describe the construction of a simplified confocal scanning laser microscope and demonstrate three-dimensional projection based on cost-efficient commercial hardware, together with available open source software.

Simplicity of Kmeans Versus Deepness of Deep Learning: A Case of Unsupervised Feature Learning with Limited Data

We study a bio-detection application as a case study to demonstrate that Kmeans -- based unsupervised feature learning can be a simple yet effective alternative to deep learning techniques for small data sets with limited intra-as well as inter-class diversity. We investigate the effect on the classifier performance of data augmentation as well as feature extraction with multiple patch sizes and at different image scales. Our data set includes 1833 images from four different classes of bacteria, each bacterial culture captured at three different wavelengths and overall data collected during a three-day period. The limited number and diversity of images present, potential random effects across multiple days, and the multi-mode nature of class distributions pose a challenging setting for representation learning. Using images collected on the first day for training, on the second day for validation, and on the third day for testing Kmeans -- based representation learning achieves 97% classification accuracy on the test data. This compares very favorably to 56% accuracy achieved by deep learning and 74% accuracy achieved by handcrafted features. Our results suggest that data augmentation or dropping connections between units offers little help for deep-learning algorithms, whereas significant boost can be achieved by Kmeans -- based representation learning by augmenting data and by concatenating features obtained at multiple patch sizes or image scales.

Application of detector precision characteristics and histogram packing for compression of biological fluorescence micrographs

Modern applications of biological microscopy such as high-content screening (HCS), 4D imaging, and multispectral imaging may involve collection of thousands of images in every experiment making efficient image-compression techniques necessary. Reversible compression algorithms, when used with biological micrographs, provide only a moderate compression ratio, while irreversible techniques obtain better ratios at the cost of removing some information from images and introducing artifacts. We construct a model of noise, which is a function of signal in the imaging system. In the next step insignificant intensity levels are discarded using intensity binning. The resultant images, characterized by sparse intensity histograms, are coded reversibly. We evaluate compression efficiency of combined reversible coding and intensity depth-reduction using single-channel 12-bit light micrographs of several subcellular structures. We apply local and global measures of intensity distribution to estimate maximum distortions introduced by the proposed algorithm. We demonstrate that the algorithm provides efficient compression and does not introduce significant changes to biological micrographs. The algorithm preserves information content of these images and therefore offers better fidelity than standard irreversible compression method JPEG2000.

A Statistical Modeling Approach to Computer-Aided Quantification of Dental Biofilm

Biofilm is a formation of microbial material on tooth substrata. Several methods to quantify dental biofilm coverage have recently been reported in the literature, but at best they provide a semiautomated approach to quantification with significant input from a human grader that comes with the graders bias of what is foreground, background, biofilm, and tooth. Additionally, human assessment indices limit the resolution of the quantification scale; most commercial scales use five levels of quantification for biofilm coverage (0%, 25%, 50%, 75%, and 100%). On the other hand, current state-of-the-art techniques in automatic plaque quantification fail to make their way into practical applications owing to their inability to incorporate human input to handle misclassifications. This paper proposes a new interactive method for biofilm quantification in Quantitative light-induced fluorescence (QLF) images of canine teeth that is independent of the perceptual bias of the grader. The method partitions a QLF image into segments of uniform texture and intensity called superpixels; every superpixel is statistically modeled as a realization of a single 2-D Gaussian Markov random field (GMRF) whose parameters are estimated; the superpixel is then assigned to one of three classes ( background, biofilm, tooth substratum) based on the training set of data. The quantification results show a high degree of consistency and precision. At the same time, the proposed method gives pathologists full control to postprocess the automatic quantification by flipping misclassified superpixels to a different state (background, tooth, biofilm) with a single click, providing greater usability than simply marking the boundaries of biofilm and tooth as done by current state-of-the-art methods.

Multispectral Flow Cytometry: Next Generation Tools for Automated Classification

Flow cytometry has moved from a relatively simple technology 30 years ago, to a very sophisticated and high-speed detection technology today. However, the number of simultaneous fluorescence dyes that can be separated is limited by the difficulty in overlapping spectra and the complexity of resolving this spectral overlap problem. High-speed multianode PMTs may change this situation. The system we propose utilizes such a technology to allow full spectral analysis of cells and particles as they flow past the light source. Making these measurements is very complex and the necessity for advanced spectral overlap calculations creates a number of difficult problems to solve in a very short period of time. Next-generation instruments can either increase the number of detectors or modify the principles of collection. If the detector system were simplified, the overall cost and complexity of single-cell analytical systems might be reduced. This requires changes in both hardware and software that allow for the analysis of 30 or more spectral signals. Analysis of complex data sets requires some completely new analytical approaches, particularly in the area of multispectral analysis. This presentation discusses a next-generation instrument, which can collect simultaneously 32 bands of fluorescence from a particle in less than 5 microseconds. This opens new opportunities for analysis of bioparticles in a very fast and high content fashion.

Using Scattering to Identify Bacterial Pathogens

New advances in elastic light scattering technology allow for faster and more accurate identification of bacteria. By using globally networked libraries of unique scattering patterns produced by bacterial colonies, researchers have developed an efficient method of identifying pathogens that has potential applications in food and water safety, health care and biodefense.

BiofilmQuant: a computer-assisted tool for dental biofilm quantification.

Dental biofilm is the deposition of microbial material over a tooth substratum. Several methods have recently been reported in the literature for biofilm quantification; however, at best they provide a barely automated solution requiring significant input needed from the human expert. On the contrary, state-of-the-art automatic biofilm methods fail to make their way into clinical practice because of the lack of effective mechanism to incorporate human input to handle praxis or misclassified regions. Manual delineation, the current gold standard, is time consuming and subject to expert bias. In this paper, we introduce a new semi-automated software tool, BiofilmQuant, for dental biofilm quantification in quantitative light-induced fluorescence (QLF) images. The software uses a robust statistical modeling approach to automatically segment the QLF image into three classes (background, biofilm, and tooth substratum) based on the training data. This initial segmentation has shown a high degree of consistency and precision on more than 200 test QLF dental scans. Further, the proposed software provides the clinicians full control to fix any misclassified areas using a single click. In addition, BiofilmQuant also provides a complete solution for the longitudinal quantitative analysis of biofilm of the full set of teeth, providing greater ease of usability.