Research Doctorate - Doctor of Philosophy (PhD) / In Western society, increased asthma prevalence over recent years has coincided with changes in dietary patterns, leading to the hypothesis that a Western diet increases susceptibility to asthma. Components of the diet that may be important are antioxidants (e.g. lycopene) and fatty acids. Lycopene and long chain n-3 polyunsaturated fatty acids (LCn-3PUFA) including eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) have anti-inflammatory effects. As asthma is a disease linked to oxidative stress and inflammation, it was hypothesised that these nutrients may have a beneficial effect individually, and may have a synergistic anti-inflammatory effect when used in combination. The aim was to examine the ability of lycopene and/or LCn-3PUFAs to protect against virus-induced inflammation, as rhinovirus infection is the primary cause of asthma exacerbation. The results presented demonstrate that both lycopene and DHA (but not EPA) individually decreased the inflammatory response of airway epithelial cells infected with rhinovirus. The results also showed that DHA supplementation increased the utilization of lycopene by cells. Furthermore, lycopene reduced rhinovirus replication. A combination of lycopene and DHA also reduced the inflammatory response of cells to rhinovirus infection, however, no synergistic anti-inflammatory effect was apparent. It is concluded that consumption of foods containing lycopene and DHA may exhibit a beneficial effect on the inflammatory response to rhinovirus infection. This may have important clinical implications, as increased dietary intake of foods rich in these nutrients may lead to a reduction in the frequency and severity of asthma exacerbations.

This dissertation is about efficient computation of the electromagnetic fields with the locally corrected Nyström (LCN) method as a point-based boundary element method (BEM). The concept of surface integral equations is discussed and the electric field integral equation (EFIE) is derived from the Maxwell’s equations. Due to its point-based nature, the LCN discretization of the EFIE has some advantages over discretizing the EFIE by the method-of-moments (MoM) which is an element-based BEM. On the other hand, due to maturity of the MoM, a large body of work is available to resolve the numerical issues arising in MoM while there has been less work related to the relatively new LCN. To combine the benefits of the LCN method and the classical Rao-Wilton-Glisson MoM, equivalence between these BEMs are established and their exact relationships are derived. Both the vector-potential EFIE and the mixed-potential EFIE are covered. Various aspects of achieving HO convergence to the correct answer using high-order (HO) LCN method are discussed. In particular, the patch size limitation, predicting the optimal degrees of freedom, and the effect of dynamic range in the solution are discussed both analytically and numerically to provide concrete motivations towards HO LCN. The benefits of an HO BEM can not be realized unless an HO geometry representation is used in conjunction with the BEM. Non-uniform rational b-spline (NURBS) surfaces are the most widely adopted HO geometry modelling technique in various disciplines due to their many advantages. However, a typical mesh created out of NURBS surfaces contain both triangular and quadrilateral elements while formulating LCN based on Gaussian quadrature rules on triangular elements have limitations. As a result, the LCN community has mostly adopted LCN based on curvilinear quadrilateral modelling of the geometry. A new class of Newton-Cotes quadrature rules for triangles is proposed to facilitate incorporating NURBS surfaces into the HO LCN. / May 2016

ELECTROMAGNETIC ANALYSIS OF PLANAR LAYERED STRUCTURES Fatma Caliskan 169 pages Directed by Dr. Andrew F. Peterson The electrical design of microelectronic devices and their packaging is complicated because of non-ideal attributes of the actual circuit realization. Electromagnetic modeling offers the possibility of accurately predicting the electrical performance of devices and reducing the cost associated with the design process. The proposed research concerns extensions of electromagnetic modeling techniques and their application to microelectronic package design. The method of moments (MoM) is utilized as a technique in modeling and analyzing these designs. Recently, an alternate approach called the locally corrected Nystrm method (LCN) has been applied to solve integral equations in electromagnetics. Recent research suggests that the LCN is well-suited for higher-order implementations and does not require cell-to-cell current continuity in the underlying representation. Thus it may offer advantages over the MoM, especially for problems involving complex 3-D structures. If cell-to-cell continuity is not required, nonconforming meshes may offer simpler geometrical modeling. In this proposal, we consider applying the above techniques to problems in package designs, which often involve multilayer structures, solid or perforated ground planes, embedded passive devices such as capacitors and spiral inductors, and interconnects in horizontal or vertical directions. Several examples will be used to illustrate the modeling.

Frequently evidentiary items contain an insufficient quantity of DNA to obtain complete or even partial DNA profiles using standard forensic gentotyping techniques. Here, various methods of post PCR purification were evaluated for their effects on the sensitivity of fluophore-based allelic detection. A method of post PCR purification is described which increases the sensitivity of standard 28 cycle PCR such that low copy number DNA templates (<100 pg DNA) can be analyzed. Full profiles were consistently obtained with as little as 20 pg template DNA without increased cycle number. In mock case type samples with dermal ridge fingerprints, genetic profiles were obtained by amplification with 28 cycles followed by post-PCR purification whereas no profiles were obtained without purification of the PCR product. Allele drop-out, increased stutter, and contamination (allele drop-in) typical of LCN analysis were observed. A single incident of contamination was observed in a reagent blank (not duplicated upon re-amplification) however, no contamination was observed in negative amplification controls.

Various error estimation and adaptive refinement techniques for the solution of electromagnetic integral equations were developed. Residual based error estimators and h-refinement implementations were done for the Method of Moments (MoM) solution of electromagnetic integral equations for a number of different problems. Due to high computational cost associated with the MoM, a cheaper solution technique known as the Locally-Corrected Nyström (LCN) method was explored. Several explicit and implicit techniques for error estimation in the LCN solution of electromagnetic integral equations were proposed and implemented for different geometries to successfully identify high-error regions. A simple p-refinement algorithm was developed and implemented for a number of prototype problems using the proposed estimators. Numerical error was found to significantly reduce in the high-error regions after the refinement. A simple computational cost analysis was also presented for the proposed error estimation schemes. Various cost-accuracy trade-offs and problem-specific limitations of different techniques for error estimation were discussed. Finally, a very important problem of slope-mismatch in the global error rates of the solution and the residual was identified. A few methods to compensate for that mismatch using scale factors based on matrix norms were developed.

Le système ostéocytaire soulève un intérêt croissant depuis quelques années car il est joue un rôle important dans l'adaptation de l'os. Le système ostéocytaire est inclus dans un réseau poreux dénommé le réseau lacuno-canaliculaire (LCN). L’observation du système ostéocytaire est difficile car les ostéocytes sont profondément enfouies dans la matrice osseuse et difficilement accessible par les techniques optiques. Récemment l’équipe de Creatis a montré la faisabilité d’imager le LCN en 3D grâce à la micro tomographie par rayonnement synchrotron. Toutefois, il n’existe actuellement pas de méthodes d’analyse permettant de quantifier, de façon automatique, le réseau lacuno-canaliculaire en 3D. L’objectif de cette thèse était de développer des méthodes d’analyse d’images permettant d’extraire des paramètres quantitatifs sur le réseau lacuno-canaliculaire. La première partie, consacrée à l’état de l’art. Le chapitre 1 présente les objectifs de ce travail. Le chapitre 2 rappelle les éléments de base sur le tissu osseux et présente les caractéristiques du réseau lacuno-canaliculaire. Le chapitre 3 présente les différentes méthodes d’imagerie utilisées jusqu’à présent pour étudier le réseau lacuno-canaliculaire. Le chapitre 4 présente l’état de l’art sur les paramètres qui sont classiquement utilisés pour caractériser le réseau lacuno-canaliculaire. La seconde partie est consacrée aux contributions de ce travail. Le chapitre 5 présente les deux systèmes expérimentaux de l’ESRF sur lesquels des images d’échantillons osseux ont été acquises. Le chapitre 6 décrit la méthode développée pour la quantification des lacunes ostéocytaires à partir d’images à l’échelle micrométrique. Elle propose de calculer des paramètres issus des moments géométriques ainsi que des paramètres basés sur la notion de volumes intrinsèques. Les méthodes sont appliquées à une série de 13 échantillons acquis en collaboration avec le Laboratoire d’Imagerie Paramétrique, Paris. Les résultats obtenus sont comparés et discutés par rapport à ceux de la littérature. Le chapitre 7 décrit la quantification des canalicules reliant les ostéocytes à partir d’images à l’échelle sous-micrométrique. En particulier, nous nous sommes intéressées à estimer le nombre de canalicules issues d’une lacune ostéocytaire, paramètre encore jamais mesuré en 3D. L’évolution de ce paramètre en fonction de la distance au centre de la lacune a permis de mettre en évidence et de quantifier la ramification des canalicules. Le chapitre 8 propose l’application des méthodes développées à une série d’échantillons acquis en collaboration le groupe de Sharmila Majumdar à l’université de San Francisco. Dans ce chapitre, nous avons travaillé sur une nouvelle méthode de segmentation du réseau lacuno-canaliculaire basée sur une méthode de chemins géodésiques. Les premiers résultats acquis sur 8 échantillons humains d’âges différents sont présentés. Finalement, le chapitre 9 conclut ce travail et présente des perspectives. / The osteocyte system has raised increasing interest in the recent years, since it is hypothesized to play an important role in orchestrating bone adaptation through mechanosensation and bone mechanotransduction mechanism. The osteocytes are deeply buried within the bone matrix, where their bodies are encysted in cavities called lacunae and their stellular processes are enclosed in tunnels called canaliculi. Together, they formed the lacuno-canalicular network (LCN). The geometry of the LCN is of importance since it is supposed to potentially affect and reflect the viability of the osteocyte and is supposed to be related to biomechanical constraints at the cell level. However, studying the LCN is quite challenging, due to limitations in an ideal imaging modality and the lack of quantitative analysis tools. In this thesis, we propose computational efficient and automated methods to quantify the 3D morphological properties of the LCN from synchrotron radiation (SR) micro / nano-CT images. For image acquisition, we used the SR micro/nano-CT setups installed on beamlines ID19 and ID22 at ESRF. A series of human cortical samples were imaged with spatial resolutions ranging between 3.5 µm to 60 nm. For the 3D assessment of lacunae, we used an image moment-based approach to calculate the volume, length, width, height and anisotropy of each osteocyte lacuna. We employed a fast algorithm to further calculate the surface area, the Euler number and the SMI of each lacuna. Validation of segmentation and experimental results on 13 bone samples are presented. For the 3D assessment of canaliculi, we propose a method to quantify the canalicular ramification around each lacuna. After segmentation, our method first labels each lacuna from the LCN. Then, a signature of the numbers of canaliculi at different distances from the lacunar surface is estimated through the calculation of topological parameters. Validation of this method and statistical results a large 3D SR micro-CT image of a human femoral bone sample are reported. We also improved the segmentation of the canaliculi and illustrated the feasibility of the application on a series of bone samples. We investigated a segmentation approach based on minimum cost paths and geodesic voting. A parallel computation scheme was implemented to reduce the computation times. The LCN was characterized by using the previous methods. Besides, we introduced the parameters from the Voronoi tessellation. Statistical results are reported on 8 large 3D micro-CT images, including around a hundred lacunae and the canaliculi. Future works will concern the improvement of canaliculi segmentation of from images at 300 nm as well as its evaluation and further characterization of LCN from SR CT images at both 300 nm and 50 nm. This work opens many perspectives for a better knowledge of the physiopathology of bone at the cellular scale.