Umeå University's logo

Analysis of numerous filamentous structures in an image is often limited by the ability of algorithms to accurately segment complex structures or structures within a dense population. It is even more problematic if these structures continuously grow when recording a time-series of images. To overcome these issues we present DSeg; an image analysis program designed to process time-series image data, as well as single images, to segment filamentous structures. The program includes a robust binary level-set algorithm modified to use size constraints, edge intensity, and past information. We verify our algorithms using synthetic data, differential interference contrast images of filamentous prokaryotes, and transmission electron microscopy images of bacterial adhesion fimbriae. DSeg includes automatic segmentation, tools for analysis, and drift correction, and outputs statistical data such as persistence length, growth rate, and growth direction. The program is available at Sourceforge.

Investigating individual red blood cells (RBCs) is critical to understanding hematologic diseases, as pathology often originates at the single-cell level. Many RBC disorders manifest in altered biophysical properties, such as deformability of RBCs. Due to limitations in current biophysical assays, there exists a need for high-throughput analysis of RBC deformability with single-cell resolution. To that end, we present a method that pairs a simple in vitro artificial microvasculature network system with an innovative MATLAB-based automated particle tracking program, allowing for high-throughput, single-cell deformability index (sDI) measurements of entire RBC populations. We apply our technology to quantify the sDI of RBCs from healthy volunteers, Sickle cell disease (SCD) patients, a transfusion-dependent beta thalassemia major patient, and in stored packed RBCs (pRBCs) that undergo storage lesion over 4 weeks. Moreover, our system can also measure cell size for each RBC, thereby enabling 2D analysis of cell deformability vs cell size with single cell resolution akin to flow cytometry. Our results demonstrate the clear existence of distinct biophysical RBC subpopulations with high interpatient variability in SCD as indicated by large magnitude skewness and kurtosis values of distribution, the "shifting" of sDI vs RBC size curves over transfusion cycles in beta thalassemia, and the appearance of low sDI RBC subpopulations within 4 days of pRBC storage. Overall, our system offers an inexpensive, convenient, and high-throughput method to gauge single RBC deformability and size for any RBC population and has the potential to aid in disease monitoring and transfusion guidelines for various RBC disorders.

We report a novel method for fabrication of three-dimensional (3D) biocompatible micro-fluidic flow chambers in polydimethylsiloxane (PDMS) by 3D-printing water-soluble polyvinyl alcohol (PVA) filaments as master scaffolds. The scaffolds are first embedded in the PDMS and later residue-free dissolved in water leaving an inscription of the scaffolds in the hardened PDMS. We demonstrate the strength of our method using a regular, cheap 3D printer, and evaluate the inscription process and the channels micro-fluidic properties using image analysis and digital holographic microscopy. Furthermore, we provide a protocol that allows for direct printing on coverslips and we show that flow chambers with a channel cross section down to 40 x 300 μm can be realized within 60 min. These flow channels are perfectly transparent, biocompatible and can be used for microscopic applications without further treatment. Our proposed protocols facilitate an easy, fast and adaptable production of micro-fluidic channel designs that are cost-effective, do not require specialized training and can be used for a variety of cell and bacterial assays. To help readers reproduce our micro-fluidic devices, we provide: full preparation protocols, 3D-printing CAD files for channel scaffolds and our custom-made molding device, 3D printer build-plate leveling instructions, and G-code.

We report an interpolation model to calculate the hydrodynamic force on tethered capsule-shaped cells in micro-fluidic flows near a surface. Our model is based on numerical solutions of the full Navier–Stokes equations for capsule-shaped objects considering their geometry, aspect ratio and orientation with respect to fluid flow. The model reproduced the results from computational fluid dynamic simulations, with an average error of <0.15 % for objects with an aspect ratio up to 5, and the model exactly reproduced the Goldman approximation of spherical objects close to a surface. We estimated the hydrodynamic force imposed on tethered Escherichia coli cells using the interpolation model and approximate models found in the literature, for example, one that assumes that E. coli is ellipsoid shaped. We fitted the 2D-projected area of a capsule and ellipsoid to segmented E. coli cells. We found that even though an ellipsoidal shape is a reasonable approximation of the cell shape, the capsule gives 4.4 % better agreement, a small difference that corresponds to 15 % difference in hydrodynamic force. In addition, we showed that the new interpolation model provides a significantly better agreement compared to estimates from commonly used models and that it can be used as a fast and accurate substitute for complex and computationally heavy fluid dynamic simulations. This is useful when performing bacterial adhesion experiments in parallel-plate flow channels. We include a MATLAB script that can track cells in a video time-series and estimate the hydrodynamic force using our interpolation formula.

Understanding dynamic mechanisms, morphology and behavior of bacteria are important to develop new therapeutics to cure diseases. For example, bacterial adhesion mechanisms are prerequisites for initiation of infections and for several bacterial strains this adhesion process is mediated by adhesive surface organelles, also known as fimbriae. Escherichia coli (E. coli) is a bacterium expressing fimbriae of which pathogenic strains can cause severe diseases in fluidic environments such as the urinary tract and intestine. To better understand how E. coli cells attach and remain attached to surfaces when exposed to a fluid flow using their fimbriae, experiments using microfluidic channels are important; and to assess quantitative information of the adhesion process and cellular information of morphology, location and orientation, the imaging capability of the experimental technique is vital.

In-line digital holographic microscopy (DHM) is a powerful imaging technique that can be realized around a conventional light microscope. It is a non-invasive technique without the need of staining or sectioning of the sample to be observed in vitro. DHM provides holograms containing three-dimensional (3D) intensity and phase information of cells under study with high temporal and spatial resolution. By applying image processing algorithms to the holograms, quantitative measurements can provide information of position, shape, orientation, optical thickness of the cell, as well as dynamic cell properties such as speed, growing rate, etc.

In this thesis, we aim to improve the DHM technique and develop image processing methods to track and assess cellular properties in microfluidic channels to shed light on bacterial adhesion and cell morphology. To achieve this, we implemented a DHM technique and developed image processing algorithms to provide for a robust and quantitative analysis of holograms. We improved the cell detection accuracy and efficiency in DHM holograms by developing an algorithm for detection of cell diffraction patterns. To improve the 3D detection accuracy using in-line digital holography, we developed a novel iterative algorithm that use multiple-wavelengths. We verified our algorithms using synthetic, colloidal and cell data and applied the algorithms for detecting, tracking and analysis. We demonstrated the performance when tracking bacteria with sub-micrometer accuracy and kHz temporal resolution, as well as how DHM can be used to profile a microfluidic flow using a large number of colloidal particles. We also demonstrated how the results of cell shape analysis based on image segmentation can be used to estimate the hydrodynamic force on tethered capsule-shaped cells in micro-fluidic flows near a surface.

Background: Studying individual red blood cells (RBCs) is critical to understanding hematologic diseases, as pathology often originates at the single-cell level. Measuring biophysical properties, such as deformability, of RBCs is also clinically important as RBC deformability is pathologically altered in numerous disease states including sickle cell disease (SCD), beta-thalassemia major (β-thal), and the "storage lesion" that occurs over time in stored samples of otherwise normal RBCs1-3. While ektacytometry-based systems measure RBC deformability, these assays do not have single-cell resolution4. As evidenced by flow cytometry, single cell measurements are important to detect pathologic cellular subpopulations in hematologic diseases. Recently, microfluidic devices that model the microvascular environment have enabled single RBC deformability measurements, but truly high-throughput measurements have remained elusive due to technical issues such as image processing inaccuracy or aberrant signaling when reconstructing RBC velocity profiles.

Methods: To that end, here we introduce a novel integration of a microfluidic device coupled with our UMUTracker, an innovative MATLAB-based automated particle tracking program that is run on a standard desktop computer5. This innovative pairing of the two technologies results in high-throughput velocity tracking of single RBCs through a microfluidic model of a capillary bed that translates to a single deformability index (sDI) for each RBC (Fig 1). To demonstrate the sDI heterogeneity of RBCs across different conditions, data was obtained from healthy volunteers, SCD and β-thal patients, and stored packed RBCs.

Results: sDI distribution curves were obtained for healthy RBCs and glutaraldehyde-stiffened RBCs (as positive controls) in which the latter resulted in an expected "left shift" and decrease in the mean sDI (Fig 2). Interestingly, while the mean sDI for SCD RBCs was also expectedly decreased, the sDI distribution was clearly non-normal, indicating the existence of heterogeneous RBC subpopulations with different deformabilities. These heterogeneous sDI distributions were observed in multiple SCD patients, including those on hydroxyurea (HU) and not, although patients on HU exhibited RBCs subpopulations with relatively higher sDIs (Fig 3). While stored RBC samples showed an expected drop in mean sDI over time, our system detected marked shifts in peak sDI with high temporal resolution over only 4 day intervals (Fig 4). Finally, each time a β-thal patient was transfused, the peak sDI shifted to the right and then gradually decreased over the course of the transfusion cycle, while the mean sDI exhibited minimal change over time (Fig 5).

Conclusions: Our novel combined microfluidic/portable image analysis system demonstrates the high-throughput capability to detect distinct RBC subpopulations, at the single cell level, of different deformabilities in SCD, β-thal, and aging stored RBCs. This heterogeneity indicates that, in these disease states, RBC deformability cannot be fully characterized with mean or bulk biophysical measurements such as those obtained with ektacytometry. Ongoing studies will determine how changes in sDI profiles are associated with clinical events and different therapies as well as the biological significance of these RBC subpopulations with varied deformabilities and the underlying mechanisms for these differences.

Phase retrieval and the twin-image problem in digital in-line holographic microscopy can be resolved by iterative reconstruction routines. However, recovering the phase properties of an object in a hologram requires an object plane to be chosen correctly for reconstruction. In this work, we present a novel multi-wavelength iterative algorithm to determine the object plane using single-shot holograms recorded at multiple wavelengths in an in-line holographic microscope. Using micro-sized objects, we verify the object positioning capabilities of the method for various shapes and derive the phase information using synthetic and experimental data. Experimentally, we built a compact digital in-line holographic microscopy setup around a standard optical microscope with a regular RGB-CCD camera and acquired holograms of micro-spheres, E. coli, and red blood cells, which are illuminated using three lasers operating at 491 nm, 532 nm, and 633 nm, respectively. We demonstrate that our method provides accurate object plane detection and phase retrieval under noisy conditions, e.g., using low-contrast holograms with an inhomogeneous background. This method allows for automatic positioning and phase retrieval suitable for holographic particle velocimetry, and object tracking in biophysical or colloidal research. 

1. Behavioral analysis based on video recording is becoming increasingly popular within research fields such as; ecology, medicine, ecotoxicology, and toxicology. However, the programs available to analyze the data, which are; free of cost, user-friendly, versatile, robust, fast and provide reliable statistics for different organisms (invertebrates, vertebrates and mammals) are significantly limited.

2. We present an automated open-source executable software (ToxTrac) for image-based tracking that can simultaneously handle several organisms monitored in a laboratory environment. We compare the performance of ToxTrac with current accessible programs on the web.

3. The main advantages of ToxTrac are: i) no specific knowledge of the geometry of the tracked bodies is needed; ii) processing speed, ToxTrac can operate at a rate >25 frames per second in HD videos using modern computers; iii) simultaneous tracking of multiple organisms in multiple arenas; iv) integrated distortion correction and camera calibration; v) robust against false positives; vi) preservation of individual identification; vii) useful statistics and heat maps in real scale are exported in image, text and excel formats.

4. ToxTrac can be used for high speed tracking of insects, fish, rodents or other species, and provides useful locomotor information in animal behavior experiments. Download ToxTrac here: https://toxtrac.sourceforge.io  (Current version v2.61).

Particle and object tracking is gaining attention in industrial applications and is commonly applied in: colloidal, biophysical, ecological, and micro-fluidic research. Reliable tracking information is heavily dependent on the system under study and algorithms that correctly determine particle position between images. However, in a real environmental context with the presence of noise including particular or dissolved matter in water, and low and fluctuating light conditions, many algorithms fail to obtain reliable information. We propose a new algorithm, the Circular Symmetry algorithm (C-Sym), for detecting the position of a circular particle with high accuracy and precision in noisy conditions. The algorithm takes advantage of the spatial symmetry of the particle allowing for subpixel accuracy. We compare the proposed algorithm with four different methods using both synthetic and experimental datasets. The results show that C-Sym is the most accurate and precise algorithm when tracking micro-particles in all tested conditions and it has the potential for use in applications including tracking biota in their environment.

Wide field-of-view imaging of fast processes in a microscope requires high light intensities motivating the use of lasers as light sources. However, due to their long spatial coherence length, lasers are inappropriate for such applications, as they produce coherent noise and parasitic reflections, such as speckle, degrading image quality. Therefore, we provide a step-by-step guide for constructing a speckle-free and high-contrast laser illumination setup using a rotating ground glass diffuser driven by a stepper motor. The setup is easy to build, cheap, and allows a significant light throughput of 48%, which is 40% higher in comparison to a single lens collector commonly used in reported setups. This is achieved by using only one objective to collect the scattered light from the ground glass diffuser. We validate our setup in terms of image quality, speckle contrast, motor-induced vibrations, and light throughput. To highlight the latter, we record Brownian motion of micro-particles using a 100x oil immersion objective and a high-speed camera operating at 2000 Hz with a laser output power of only 22 mW. Moreover, by reducing the objective magnification to 50x, sampling rates up to 10,000 Hz are realized. To help readers with basic or advanced optics knowledge realize this setup, we provide a full component list, 3D-printing CAD files, setup protocol, and the code for running the stepper motor.