No description available.

No description available.

Raman spectroscopy is a powerful molecular fingerprinting method which measures the vibrational modes of molecules to identify and quantify chemical species. In biomedical spectroscopy, where samples are usually complex mixtures of many molecules, Raman spectra give a biochemical “portrait” that can be used to discriminate between distinct samples. One major technical challenge in implementing Raman spectrometer sensors is the technique’s low intrinsic signal to noise ratio. To amplify the Raman signal, a number of different approaches can be applied. In this thesis two techniques are used; surface enhanced Raman scattering (SERS) from metal nanoparticles along with light-matter interaction enhancement from co-coupling light and sample to a liquid core waveguide. In order to process the complex spectral data arising from these sensors, a robust signal processing method is required. To this end, we have developed and validated a machine learning spectral analysis platform based on genetically optimized support vector machines (GA-SVM). This work is the subject of Chapter 3. We found that the GA-SVM significantly outperformed the standard statistical based modelling approach, partial least squared, in regression tasks for several different biomedical Raman applications. Furthermore, we found that the use of more complex kernel functions in the SVM yielded superior results. The genetic optimization algorithm was necessary to use these more complex kernel functions because its computation time scales linearly with complexity, whereas the standard brute force approach scales exponentially. Chapter 4 concerns the development of a Raman sensor used to quantify and identify pathogenic bacteria. This device centres on a microfluidic flow cell which forces bacteria to flow through a hollow-core photonic crystal fiber (HC-PCF) to which the Raman excitation laser is also coupled. The bacteria are also mixed with silver nanoparticles to simultaneously achieve SERS and light-matter interaction enhancement in the sensor. Overall, the fiber and nanoparticles yield a bulk enhancement of 400x for the Raman spectrum. Bacteria are quantified in this system by counting the number of “spectral events” that occur as cells flow through the HC-PCF in a 15-minute window. This approach achieved very high linearity, as well as an average detection limit of 3.7 CFU/mL. In addition, bacteria are identified by using the same GA-SVM algorithm developed in the preceding chapter. These machine learning models achieved a discrimination accuracy of ~92% when comparing the spectra of the bacteria S. aureus, P. aeruginosa, and E. coli. In mixed samples of bacteria, the error of quantification increased significantly to 13.3 CFU/mL, but the output of the sensor was highly correlated with the ground-truth bacterial load. In Chapter 5 we outline the development of a diagnostic scheme for chemoresistance in ovarian cancer based on SERS measurements from cysteine-capped gold nanoparticles. Resistance to chemotherapy was determined based on three factors: the concentration of tumor derived exosomes, the chemical composition of the exosomes, and the concentration of exosome-derived cisplatin. Cisplatin is the drug of interest for this problem, as it is the most basic chemotherapy agent. The system works by first incubating the gold nanoparticles with tumor derived exosomes. The cisplatin therein causes the particles to destabilize slightly, resulting in the aggregation rate of the nanoparticles being proportional to the drug concentration. At steady state aggregation, the magnitude of the Raman spectrum is proportional to the exosome concentration, and the spectrum contains its chemical identity. Using in vitro cancer cell lines, we found that resistant cells tend to produce more exosomes and excrete a higher concentration of cisplatin within them. Overall, this sensor exhibited good diagnostic power for chemoresistance particularly in the most common subtype in ovarian cancer.

Den svenska hälso- och sjukvården står inför utmaningen med en ökande och åldrande befolkning. Överdiagnostisering och under-, över- och felanvändning av diagnostisk service bidrar till de ekonomiska utmaningarna. Utilization management kan bidra till att komma tillrätta med användningen. I ett samarbetsprojekt mellan Rådet för värdeskapande användning av Medicinsk diagnostik i Region Jönköpings län och en vårdcentral genomfördes ett förbättringsarbete för att minska förbrukningen av laboratorieanalyser; målet var att komma till rätta med över-, under och felanvändning. Parallellt genomfördes en studie med syftet att utvärdera effekter av fem insatser rådet gjort. Studien genomfördes med en mixad metod med förklarande sekventiell design. Då förbättringsarbetet aldrig tog fart i verksamheten, på grund av chefsomsättning, kan inga resultat redovisas. Studien påvisar få förändringar i beställningsmönster. I intervjuerna med remittenterna framkommer temat att vilja arbeta med beställningsmönster och förändringar nära verksamheten. Remittenterna efterfrågar tekniska lösningar och dialog med de diagnostiska specialiteterna. Litteraturen stödjer hittills gjorda insatser men i studien ses få resultat. Tidigare insatser har varit engångsföreteelser. Genom att nyttja förbättringskunskapen och arbeta med frågorna närmare remittenternas vardag kan engagemang skapas varpå förändringen lättare anammas. Kombinationen av förbättringsarbeten på mikrosystemnivå med stöd och förbättringar på mesonivå kommer sannolikt att bidra till framgång med värdeskapande användning. / Swedish healthcare is facing the challenge of an increasing and aging population making cost reductions necessary. Some of the challenges are based around the over diagnosis and the misutilisation of laboratory analysis. Utilisation management could contribute to proper use. A quality improvement project was performed as a collaboration between a primary care centre and the Council of Value-creating use of Medical Diagnosis. The aim was to reduce the number of analyses and misuse of laboratory analysis. A mixed method study was conducted with the aim to evaluate five interventions made by the Council. Because the improvement project never really started, results are limited. Interviews with physicians revealed that the studied units are open for dialogue with the diagnostic specialties and willing to change ordering patterns but few changes were detected. The literature supports the interventions made so far but few results are shown. One reason might be that previous efforts have been isolated events. By using the knowledge of quality improvement and bringing the questions closer to the units’, change might be easier to embrace. The combination of quality improvement in the microsystems with support and improvements in mesosystems will probably contribute to success.

In this dissertation, we extend several relatively new developments in statistical model selection and data mining in order to improve one of the workhorse statistical tools - mixture modeling (Pearson, 1894). The traditional mixture model assumes data comes from several populations of Gaussian distributions. Thus, what remains is to determine how many distributions, their population parameters, and the mixing proportions. However, real data often do not fit the restrictions of normality very well. It is likely that data from a single population exhibiting either asymmetrical or nonnormal tail behavior could be erroneously modeled as two populations, resulting in suboptimal decisions. To avoid these pitfalls, we develop the mixture model under a broader distributional assumption by fitting a group of multivariate elliptically-contoured distributions (Anderson and Fang, 1990; Fang et al., 1990). Special cases include the multivariate Gaussian and power exponential distributions, as well as the multivariate generalization of the Student’s T. This gives us the flexibility to model nonnormal tail and peak behavior, though the symmetry restriction still exists. The literature has many examples of research generalizing the Gaussian mixture model to other distributions (Farrell and Mersereau, 2004; Hasselblad, 1966; John, 1970a), but our effort is more general. Further, we generalize the mixture model to be non-parametric, by developing two types of kernel mixture model. First, we generalize the mixture model to use the truly multivariate kernel density estimators (Wand and Jones, 1995). Additionally, we develop the power exponential product kernel mixture model, which allows the density to adjust to the shape of each dimension independently. Because kernel density estimators enforce no functional form, both of these methods can adapt to nonnormal asymmetric, kurtotic, and tail characteristics. Over the past two decades or so, evolutionary algorithms have grown in popularity, as they have provided encouraging results in a variety of optimization problems. Several authors have applied the genetic algorithm - a subset of evolutionary algorithms - to mixture modeling, including Bhuyan et al. (1991), Krishna and Murty (1999), and Wicker (2006). These procedures have the benefit that they bypass computational issues that plague the traditional methods. We extend these initialization and optimization methods by combining them with our updated mixture models. Additionally, we “borrow” results from robust estimation theory (Ledoit and Wolf, 2003; Shurygin, 1983; Thomaz, 2004) in order to data-adaptively regularize population covariance matrices. Numerical instability of the covariance matrix can be a significant problem for mixture modeling, since estimation is typically done on a relatively small subset of the observations. We likewise extend various information criteria (Akaike, 1973; Bozdogan, 1994b; Schwarz, 1978) to the elliptically-contoured and kernel mixture models. Information criteria guide model selection and estimation based on various approximations to the Kullback-Liebler divergence. Following Bozdogan (1994a), we use these tools to sequentially select the best mixture model, select the best subset of variables, and detect influential observations - all without making any subjective decisions. Over the course of this research, we developed a full-featured Matlab toolbox (M3) which implements all the new developments in mixture modeling presented in this dissertation. We show results on both simulated and real world datasets. Keywords: mixture modeling, nonparametric estimation, subset selection, influence detection, evidence-based medical diagnostics, unsupervised classification, robust estimation.

abstract: Antibiotic resistance is a very important issue that threatens mankind. As bacteria are becoming resistant to multiple antibiotics, many common antibiotics will soon become ineective. The ineciency of current methods for diagnostics is an important cause of antibiotic resistance, since due to their relative slowness, treatment plans are often based on physician's experience rather than on test results, having a high chance of being inaccurate or not optimal. This leads to a need of faster, pointof- care (POC) methods, which can provide results in a few hours. Motivated by recent advances on computer vision methods, three projects have been developed for bacteria identication and antibiotic susceptibility tests (AST), with the goal of speeding up the diagnostics process. The rst two projects focus on obtaining features from optical microscopy such as bacteria shape and motion patterns to distinguish active and inactive cells. The results show their potential as novel methods for AST, being able to obtain results within a window of 30 min to 3 hours, a much faster time frame than the gold standard approach based on cell culture, which takes at least half a day to be completed. The last project focus on the identication task, combining large volume light scattering microscopy (LVM) and deep learning to distinguish bacteria from urine particles. The developed setup is suitable for pointof- care applications, as a large volume can be viewed at a time, avoiding the need for cell culturing or enrichment. This is a signicant gain compared to cell culturing methods. The accuracy performance of the deep learning system is higher than chance and outperforms a traditional machine learning system by up to 20%. / Dissertation/Thesis / Doctoral Dissertation Electrical Engineering 2020

This thesis investigates the use of Raman tweezers microspectroscopy to interrogate the biochemistry of single biological cells. Raman tweezers microspectroscopy is a powerful technique, which combines traditional Raman microspectroscopy and optical trapping, allowing the manipulation and environmental isolation of a biological cell of interest whilst simultaneously probing its biochemistry gleaning a wealth of pertinent information. The studies carried out in this thesis can be split into two broad categories: firstly, the exploitation of Raman tweezers microspectroscopy to study biological cells and secondly developments to the Raman tweezers microspectroscopy technique that extend its capabilities and the range of samples that can be studied. In the application of Raman tweezers, the stacking and interrogation of multiple cells is reported allowing a rapid representative Raman signal to be recorded from a small cell population with improved signal to noise. Also demonstrated is the ability of Raman spectroscopy to identify and grade the development of Human Papillomavirus induced cervical neoplasia with sensitivities of up to 96 %. These studies demonstrate the potential of Raman spectroscopy to study biological cells but it was noted that the traditional Raman tweezers system struggled to manipulate large cells thus a decoupled Raman tweezers microspectroscopy system is presented where a dual beam fibre optical trap is used to perform the trapping function and a separate Raman probe is introduced to probe the biochemical nature of the trapped cell. This development allowed the trapping and examination of very large cells whilst opening up the possibility of creating Raman maps of trapped objects. Raman tweezers microspectroscopy could potentially become an important clinical diagnostic and biological monitoring tool but is held back by the long signal integration times required due to the weak nature of Raman scattering. The final study presented in this thesis examines the potential of wavelength modulated Raman spectroscopy to improve signal to noise ratios and reduce integration times. All these studies aim to demonstrate the potential and extend the performance of Raman tweezers microspectroscopy.

abstract: This work describes efforts made toward the development of a compact, quantitative fluorescence-based multiplexed detection platform for point-of-care diagnostics. This includes the development of a microfluidic delivery and actuation system for multistep detection assays. Early detection of infectious diseases requires high sensitivity dependent on the precise actuation of fluids. Methods of fluid actuation were explored to allow delayed delivery of fluidic reagents in multistep detection lateral flow assays (LFAs). Certain hydrophobic materials such as wax were successfully implemented in the LFA with the use of precision dispensed valves. Sublimating materials such as naphthalene were also characterized along with the implementation of a heating system for precision printing of the valves. Various techniques of blood fractionation were also investigated and this work demonstrates successful blood fractionation in an LFA. The fluid flow of reagents was also characterized and validated with the use of mathematical models and multiphysics modeling software. Lastly intuitive, user-friendly mobile and desktop applications were developed to interface the underlying Arduino software. The work advances the development of a system which successfully integrates all components of fluid separation and delivery along with highly sensitive detection and a user-friendly interface; the system will ultimately provide clinically significant diagnostics in a of point-of-care device. / Dissertation/Thesis / Masters Thesis Biomedical Engineering 2018

Elevated pulmonary artery pressure (PAP) is a significant healthcare risk. Continuous monitoring for patients with elevated PAP is crucial for effective treatment, yet the most accurate method is invasive and expensive, and cannot be performed repeatedly. Noninvasive methods exist but are inaccurate, expensive, and cannot be used for continuous monitoring. We present a machine learning model based on heart sounds that estimates pulmonary artery pressure with enough accuracy to exclude an invasive diagnostic operation, allowing for consistent monitoring of heart condition in suspect patients without the cost and risk of invasive monitoring. We conduct a greedy search through 38 possible features using a 109-patient cross-validation to find the most predictive features. Our best general model has a standard estimate of error (SEE) of 8.28 mmHg, which outperforms the previous best performance in the literature on a general set of unseen patient data.

There is an ever growing demand for low cost biosensors in medical diagnostics. A well known commercially successful example is glucose biosensors which are used to diagonize and monitor diabetes. These biosensors use electrochemical analysis (electro analysis) as transduction mechanism. Electro analytical techniques involve application of electrical stimulus to the chemical/biochemical system under consideration and measurement of electrical response due to the oxidation and reduction reactions that occur because of the stimulus. They offer a lot of advantages in terms of sensitivity, selectivity, cost effectiveness and compatibility towards integration with electronics. Besides glucose, there are several biomolecules of significance for which electro analysis can potentially be used to develop low cost, rapid, easy to use biosensors. One such biomolecule is Glycated Hemoglobin (GHb). It is a post translational, non-enzymatic modification of hemoglobin with glucose and is a very good biomarker that indicates the average value of blood glucose over the past 120 days. It is always expresses as a percentage of total hemoglobin present in blood. Monitoring diabetes based on the value of percentage Glycated hemoglobin is advantageous as it gives an average value of glucose unlike plasma glucose values which vary a lot on a day to day basis depending on the dietary habits and the stress levels of the individual. This thesis is focused on the development of a low coat, easy to use, disposable sensor for measuring percentage Glycated hemoglobin. The first challenge in developing such a sensor is isolation of hemoglobin. Unlike glucose which is present in blood plasma (liquid content of blood), hemoglobin resides inside red blood cells also known as erythrocytes. O isolate hemoglobin, these cells have to be broken or lysed. All the existing approaches rely on mixing blood with lysing reagents to lyse erythrocytes. Ideal biosensors should be devoid of liquid reagents. Keeping this in perspective, in this thesis, this challenge is addressed by developing two entirely buffer/reagentless techniques to lyse erythrocytes and isolate hemoglobin. In the first technique, cellulose acetate membranes are embedded with lysing reagents and are used for lysing reagents and are used for lysing application. In the second techniques, commercially available nylon mesh nets are modified with lysing reagents to lyse and isolate hemoglobin. These membranes or mesh nets can be easily integrated on top of a disposable strip. After isolating hemoglobin, the next challenge is to selectively detect Glycated hemoglobin. Boronic acid conjugates are known to bind Glycated hemoglobin. Using this principle, a new composite is sysnthesized to specifically detect glc\ycated hemoglobin. The composite (GO-APBA) is a result of functionalization of Graphene Oxide (GO) with 3-aminophenylboronic acide (APBA). Detection of Glycated hemoglobin is achieved by modifying screen printed electrode strips with the synthesized compound, thus taking a step forwards achieving the objective. Since Glycated hemoglobin is always expressed as a percentage of hemoglobin, the next challenge is to detect total hemoglobin. In this thesis a low cost way of detecting hemoglobin is achieved by using GO modified or surfactant modified screen printed electrode strips. Furthermore, the potential interferences that blood plasma can cause in these measurements are eliminated with the help of permselective coatings. Thus using the technologies developed in this thesis, measurements of percentage Glycated hemoglobin can be potentially made on handheld electronic devices akin to glucose meters by using just a drop of blood.