Van Thi Huyen Nguyen

• 20/12/2022
  
• 4 p.m.
• Venue: Campus Drie Eiken, Q.002
• ​Online PhD defence​
• Supervisors: Jan Sijbers & Jan De Beenhouwer
• Department of Physics

Abstract

X-ray Computed Tomography (CT) is a powerful technique for non-invasive evaluation and visualization of an object's internal structure. A CT image is computed from a set of X-ray radiographs of the object acquired from different directions relative to the object in a preset scanner geometry. To obtain an artifact-free 3D tomographic image of the object, it is crucial to establish a correct geometry description of the system configuration for CT reconstruction. Unfortunately, it is not possible to obtain accurate geometry information without repeating calibration for every new system setting in an experimental X-ray scanner. Geometry calibration typically involves describing and estimating the physical arrangement of the X-ray source, rotation stage, and detector prior to computing the CT image. If the geometry parameters are not derived correctly, the CT image is subjected to misalignment artifacts that destroy the internal detail of the object.

In this thesis, a phantom-based calibration method is presented to estimate the geometry setting of a biplanar X-ray CT system.

Successfully calibrating the system geometry opens possibilities for different applications using the X-ray radiographs acquired with the same X-ray system.

Chapter 1 and 2 provide background knowledge on X-ray imaging and X-ray CT, as well as foundations to build and train a deep neural model. A typical residual neural network (ResNet) is also presented in chapter 2. Chapter 3 focuses on calibrating an experimental biplanar X-ray cone-beam system using a LEGO phantom. More specifically, a procedure to construct the calibration phantom using the LEGO bricks and metal markers, followed by a deep learning-based marker tracking method and geometry alignment, will be presented. Chapter 4 discusses 2D/3D registration, one of the X-ray CT applications in biomedical imaging for the study of animal kinematics.

The registration method aligns the projections of reference 3D landmarks to a set of 2D landmarks for an estimation of the object's 3D pose. An automated 3D landmark extraction procedure and a trained deep neural network facilitate 2D landmark detection and tracking in fluoroscopy images.

Bram Vanden Broecke

• 16/12/2022
  
• 5 p.m.
• Venue: Campus Drie Eiken, Q.002
• Supervisors: Herwig Leirs & Erik Matthysen
• Department of Biology

Abstract

Emerging infectious diseases are a significant challenge for global public health. Understanding the transmission dynamics of such pathogens is crucial to reduce and control potential outbreaks. The rate at which an infection spreads within a population depends on the transmission coefficient, which can vary among individuals within a single population. The underlying mechanisms need to be studied more closely and could affect epidemiological models. One of these mechanisms is potentially the host’s personality.

Animal personality, or consistent individual variation in behaviour among individuals over time and/or context, has indeed been found to affect parasitic infection risk. Most studies, however, focused on macro-and ectoparasites, while relatively little attention has been paid to viruses. Nonetheless, such information is potentially crucial to increase our understanding of the disease dynamics of pathogens in wildlife, especially for those with zoonotic potentials. ​ ​

We focused on animal personality in the multimammate mice (Mastomys natalensis), how variation in these traits is maintained over generations and how they influence viral infection risk. We had three main questions that we wanted to investigate in this study: (i) Can we measure animal personality in M. natalensis? (ii) Is there a link between personality and MORV infection risk? (iii) Which factors are responsible for maintaining the variation in personalities? ​ ​

In short, (i) we were able to measure two personality traits: stress-sensitivity and exploration. (ii) Only the latter affected MORV infection risk, where less explorative individuals were more likely to be infected with MORV, potentially due to underlying co-infection patterns with other parasites, such as different helminth species and bacteria. (iii) Density played a significant role in maintaining the variation of these two personality traits. This study shows that animal personality in M. natalensis exists, is maintained by variation in density and that they influence the disease dynamics of MORV. ​ ​

Alice Presenti

• 02/12/2022
  
• 4 p.m.
• Venue: Campus Drie Eiken, Q.001
• ​Online PhD defence
• Supervisors: Jan Sijbers & Jan De Beenhouwer
• Department of Physics

Abstract

X-ray computed tomography (CT) is a widely used technique for the inspection of the external and the internal components of industrial objects. It requires the acquisition of a large number of projections (usually from several hundred to a few thousand) at different angular views, from which a volumetric representation of the object is reconstructed. In industrial inspection based on CT, a triangular mesh is extracted from the CT reconstruction and registered to a nominal geometry for comparison. However, mainly due to the long acquisition time, this procedure does not apply to inline inspection, where each product must be analysed in a few seconds.

For this reason, radiography-based methods not dealing with CT reconstruction, are preferred for inline applications. However, most of these methods lack accuracy and 3D spatial information. The aim of this thesis is to search for effective methods which allow for 3D fast and automatic inline inspection from very few radiographs.

The thesis is divided into four main parts. The first part is a general introduction to the basic knowledge required to understand the subsequent contents. It is divided into three chapters: quality control techniques (with a focus on X-ray radiography), image registration and neural networks. In the second part, an inspection method based on the dynamic registration between the acquired projections and simulated ones from the reference model of the object is presented. Subsequently, in the third part, a neural network is proposed to estimate the 3D pose of an object from one or multiple radiographs.

Finally, as the final goal of inspection is to determine the quality of objects, a neural network is described to automatically classify objects as defective or not defective based on a single radiograph.

Mehrdad Abdi

• 28/11/2022
  
• 3 p.m.
• Venue: Campus Middelheim, G.005
• ​Online PhD defence
• Supervisor: Serge Demeyer
• Department of Computer Science

Abstract

Effective testing is essential in today's digital society. Not only do effective tests enhance quality, speed up the development process, and reduce the risk, but ultimately they result in better software.

Effective testing is even more important in the context of software ecosystems. These are networks of technical components built by a loosely coupled heterogeneous group of software engineers. The high degree of interdependencies between the components, in combination with the constant evolution within the network, makes effective testing a real challenge.

In this dissertation, we investigate the use of test amplifiers in the context of software ecosystems. We exploit the symbiotic relationship between the test amplifier on the one hand and the network on the other hand. We conduct this investigation from the perspective of two feedback loops:

(1) The test amplifier is fed by knowledge extracted *out* of the ecosystem, like the source code, development history, interproject dependencies, and developers' activities.

(2) The test amplifier provides improvements *in* the available tests to reduce the impact of software defects.

In our research, we, therefore, applied test amplifiers in new ecosystems, identified points for improvement, and proposed possible solutions. This includes preliminary results concerning the transplantation of tests from one project to a similar project within the ecosystem.

As such, we have made important steps towards the ultimate dream: a ``zero-touch'' test amplifier that strengthens the tests within a software ecosystem without any human intervention.

Aäron Penders

• 25/11/2022
  
• 4 p.m.
• Venue: Campus Groenenborger, T.105
• ​Online PhD defence
• Supervisors: Nick Schryvers, Milan Konstantinović & Rik-Wouter Bosch
• Department of Physics

Abstract

The operating conditions of pressurized-water reactor (PWR) nuclear power plants (i.e. those stationed in Doel and Tihange) can bring about severe material degradation which may limit the plant’s structural integrity and its expected lifetime. One of these degradation mechanisms is the not fully understood phenomena known as irradiation-assisted stress corrosion cracking (IASCC). IASCC is an intergranular cracking effect that manifests from the interplay between several factors, including the aspects of heavy neutron-irradiation (emanating from the nuclear fission reaction), the high temperature pressurized water environment (around 320 °C at 150 bars; exhibited by the coolant water that flows through the PWR system) and the residual stresses that are imposed during power plant operation. In the framework of this Ph.D. thesis, the IASCC phenomenon was investigated in both tested and industrial specimens, wherein which some of the materials were extracted from the Tihange-2 power plant site. In particular, the characterization of the microstructure and the morphology of intergranular IASCC cracks and crack tips along fully oxidized grain boundaries of stainless steels is deemed important. The results obtained in this investigation provide crucial information in order to rationalize some of the existing cracking models and methodologies used in the analysis of the failures of nuclear reactor components. This investigation was a collaboration between the University of Antwerp and the Belgian Nuclear Research Centre SCKCEN, and was partially financially sponsored by Tractebel Engie.

Roy Aerts

• 22/11/2022
  
• 5 p.m.
• Venue: Campus Drie Eiken, Q0.02
• ​Online PhD defence
• Supervisors: Christian Johannessen & Wouter Herrebout
• Department of Chemistry

Abstract

Fundamental spectroscopic research revolves around comprehending the subject technique on a theoretical level and discovering in what chemical queries it is deployable. This dissertation attempts to do this for the technique called Raman optical activity (ROA).

ROA is the detection of a differential Raman intensity between sending in right- and left circular polarized light or the differential detection of right- or left circular polarized Raman scattered intensities. The preference for one of the circular polarization states finds its origin in the chirality of a system. The main application of ROA were found to be absolute configuration determination of small organic compounds and the secondary structure elucidation of proteins. Only from the unlocking of ROA spectral calculations onwards more sophisticated elucidations took place. Various polypeptides and (poly)carbohydrates became subjects of investigation by combining empirical work with spectral simulations. By 2018, when this thesis was initiated, ROA calculations and analyses were possible for quite sophisticated systems, finally permitting the investigation of molecules that are actually being used in pharmacy, an application area where ROA was believed to be a valuable future contributor. The question that remained, however, was: how exactly can ROA contribute and how should this be done? The main goal of this dissertation is to respond hereon, as such bringing together ROA and pharmacy.

The antibiotic glycopeptide molecular class was carefully selected as subject of investigation, the aim being the determination of the conformational ensemble of this molecular class by means of ROA. Besides ROA, also the closely related vibrational circular dichroism (VCD) technique is added to the study. Subsequently, derivatives of vancomycin -oritavancin, dalbavancin, teicoplanin- were recorded, so as to experimentally observe the spectral effect of changing the chemical structure and/or conformation of the glycopeptide. Finally, the vancomycin binding to its biological target, Lipid II -a bacterial cell-wall precursor- was investigated using ROA. The principal conclusion that can be derived from all the above studies is that the aromatic systems that adopt a fixed conformation define the spectral responses, whereas the carbohydrates barely contribute.

All combined, the obtained results provide important insights in the applicability of the VOA techniques in the conformational elucidation of challenging, real-world molecular systems. Both the strategies and the type of measurements contained herein represent a step in connecting the ROA field and pharmaceutical world. ​ ​

Saranya Thiruvottriyur Shanmugan

• 17/11/2022
  
• 9 a.m.
• ​Online PhD defence
• Supervisors: Karolien De Wael & Stanislav Trashin
• Department of Bioscience Engineering

Abstract

MicroRNAs (miRNAs) are small oligonucleotides (18-25 bases) that play a key role in epigenetic regulation. Since the discovery of miRNAs in 1993, their role in biological processes has been extensively investigated. By now, it has been evidenced that monitoring and detection of miRNAs can improve (early) disease diagnosis. The existing diagnostic approaches have limitations such as the need for complex multistep protocols for sample preparation, analysis, data interpretation, high cost of equipment, the need for highly qualified personnel, and high laboratory standards. As opposed to this, point-of-care biosensors and chips aim to facilitate the procedure and avoid sending samples into centralized laboratories, which saves time, reduces the chance of sample degradation, and enables analysis of patient samples in remote areas, directly at home or primary and secondary care facilities (i.e. general practitioners and specialists). The latter is essential for therapy assessment and follow-up monitoring of patients with chronic diseases and cancer. Nevertheless, the development of such sensors is lagging compared to the projections of 10 – 20 years ago, mainly due to insufficient sensitivity, poor reproducibility, and the complexity of sensors’ design reported in the research literature. This motivates the development of new detection strategies and technologies such as photoelectrochemical sensors that combine the best features of different sensing approaches.

The primary concern when developing detection technologies for miRNA is the need for a highly sensitive and selective platform. This thesis explores a novel photoelectrochemical (PEC) method that is distinctive owing to its sensitive nature and simple and robust design. Firstly, we focused on the usage of recently emerging commercial gold-sputtered electrode systems for the detection of short nucleic acid with enzymatic amplification. Importantly, cleaning such electrodes is a challenge since the standard procedures known for regular disk electrodes such as polishing cannot be employed here, since it will damage the protective layer on the electrode. However, the electrodes can be washed and pretreated chemically and/or electrochemically. Thus, a procedure to effectively clean and modify the gold-sputtered electrode has been developed, resulting in high-performance gold-sputtered nucleic acid sensors.

Next, the usage of molecular photosensitizers as an alternative to enzymatic amplification has been evaluated. We took advantage of the 1O2 production by photosensitizers upon photoexcitation, leading to a photocurrent response due to the 1O2-induced (electro)chemical conversions. Following the demonstration of the detection strategy, the analytical performance of the sensing system was evaluated using magnetic beads-based nucleic acid assay on disposable electrode platforms, with a focus to enhance the sensitivity and robustness of the technique in detecting complementary nucleic acid targets.

Following the fundamental evaluation of the singlet oxygen-based PEC detection of nucleic acids, we further optimized the assay and measurement parameters and employed the sensing strategy for a polymerase chain reaction-free (PCR-free) quantification of miRNAs related to prostate cancer. By successfully detecting and quantifying low-picomolar range concentrations (< 10 pM) in plasma samples from prostate cancer patients, we successfully showed the applicability of the novel sensing strategy. We have also compared and positioned the performance of our developed PEC strategy with an existing state-of-art technique, i.e. electrochemiluminescence (ECL). Our PEC strategy performed on par with ECL, both yielding low-picomolar detection limits in serum matrices, however quicker and cheaper than ECL.

Owing to the versatility of this PEC technique, the final study explored its multiplexing capability. As a starting point in this branch of the research, we have investigated two possible ways for multiplexing. To perform multiple measurements at the same time, constructing calibration plots and quantifying unknown miRNA concentrations in patient samples at the same time, we have developed a high-throughput detection with 96X multi-channel electrode systems and in-house designed and constructed 96XLED illumination sources. Secondly, to detect more than one target miRNA in a single measurement, intra-vial multiplexing where the samples were analyzed for different targets in one vial was also explored.

Altogether, this thesis presents the fundamentals, development and application of a novel PEC strategy for detecting short (< 25 bases) nucleic acid sequences, in particular, miRNA. With an aim to serve as a distinctive technique to function as a clinical testing platform without any need for PCR, this work adds value to the development of nucleic acid-based sensors for miRNAs and other short-stranded nucleic acid biomarkers, and benefits in the early detection of diseases like cancer.

Miguel De Clercq

• 26/10/2022
• 4.30 p.m.
• Venue: Campus Drie Eiken, O.03
• ​Online PhD defence​
• Supervisors: Sofie Cambré & Wim Wenseleers
• Department of Physics

Abstract

Single-wall carbon nanotubes (SWCNTs) exhibit unique and remarkably diverse electronic, optical and mechanical properties, that depend critically on their exact chiral structure. In particular, their hollow structure, with diameters in the nanometer range and lengths up to microns, make them unique one dimensional channels for aligning molecules in one dimension. In this PhD thesis I investigate the encapsulation of small optical switches inside SWCNTs. Such molecules can reversibly change their structure under the influence of external stimuli, such as visible light. By encapsulating optical switches we therefore anticipate to add new functionalities to the already unique properties of the SWCNTs, in particular the ability to switch their electronic and optical properties by visible light. I successfully encapsulated two different optical switches inside the SWCNTs, namely the spiropyran - merocyanine (SP/MC) optical switch and the dihydroazulene - vinylheptafulvene (DHA/VHF) switch. While for SP/MC the switching of the dye inside the SWCNTs could not yet be proven, for DHA/VHF the switching inside the SWCNTs results in shifts of the SWCNT optical transitions of up to 12nm. Finally, I also investigate the effect of ultraviolet (UV) light on the optical properties of SWCNTs and demonstrate the creation of luminescent defect sites along the SWCNT backbone. The results of this thesis will create a range of new possible applications of SWCNTs, from light-switchable transistors for sensing applications to the creation of memory elements that can be triggered with ultraviolet or visible light.

Marie Legein

• 26/10/2022
  
• 4 p.m.
• Venue: Stadscampus, W. Elsschotzaal, hof van Liere, Prinsstraat 13, 2000 Antwerpen
• Supervisors: Sarah Lebeer & Roeland Samson
• Department of Bioscience engineering

Abstract

We depend on plant protection products to secure our food production. However, our current use of these products threatens our health and the environment, and diseases acquire resistance to pesticides. Additionally, regulations on pesticide use are becoming more stringent, and consumers are demanding cleaner produce. Thus, there is a need for novel solutions to protect our crops.

These novel solutions could be based on the never-ending arms race between microbes. In this thesis, we explored the microbes inhabiting the phyllosphere, the above-ground parts of plants, looking for microbes that can improve the plant’s health. First, we described the known mechanisms by which phyllosphere bacteria protect plants against diseases. This knowledge is key in developing microbe-based solutions.

Secondly, we moved toward the field and explored the bacteria of greenhouse crops. Greenhouses, though highly effective agricultural environments, are characterized by reduced sources of bacterial diversity compared to more natural settings. Interestingly, bumblebees and predatory mites, introduced for pollination and pest control, were important bacterial sources and vectors. This opens possibilities for using these introduced bumblebees and mites to enrich the greenhouse microbiome. Next, we focused on isolating and characterizing beneficial microbes. First, we examined compost teas as an underexplored source of beneficial microbes. These liquid fermented extracts are traditional, microbe-based solutions that require little technology. However, we showed that storage, inevitable in commercial products, impacts the microbial communities and this raises questions on the efficacy of commercial compost tea products. Therefore, subsequent efforts in this thesis were directed toward more defined microbe-based solutions.

We zoomed in on two promising lactobacilli. Lactobacilli are known for their uses in fermentations and as probiotics, but their benefits for plants remain underexplored. First, we characterized the novel species Latilactobacillus fragifolii. Interestingly, this species produces a carotenoid, an uncommon trait that showed to be associated with flower and insect adaptation. Moreover, the novel species was also able to protect seedlings from disease.

The second strain of interest was Lactiplantibacillus plantarum AMBP214, a biocontrol agent that can be dispersed via bumblebees. Effective and timely dispersal of biocontrol agents is not straightforward in greenhouses. Here, we showed that this strain could be spray-dried and dispersed consistently onto strawberry flowers, in numbers exceeding previous reports.

In conclusion, this thesis contributed to our understanding of the phyllosphere microbiome of greenhouses and explored the application potential of lactobacilli in these ecosystems.

Kristina Stanković

• 18/10/2022
• 5 p.m.
• Venue: Campus Drie Eiken, O.003
• ​Online PhD defence​
• Supervisors: Jan Sijbers, Brian Booth & Toon Huysmans
• Department of Physics

Abstract

A human foot has a complex geometry which results in a variation in shape. For the creating and validating shoe and orthotic designs, a quantitative description of the foot shape is of particular use. These designs are usually based on simplified representations of the human foot. However, the inconsistency and incompleteness in these measurements led to the need for the analysis of higher dimensional (2D and 3D) foot shape representations. Although the analysis of 2D footprints provides a quantitative description of foot shape, it lacks information along the vertical dimension. In addition to foot shape measurements, the identification of abnormal foot shape regions is an area of open research. Typically, an experienced professional is required to assess foot shape anomalies. The automation of this foot assessment would be desirable to both reduce the barriers to foot shape assessment, and to eliminate subjectivity introduced through human influence during the assessment. These observations suggest that an automatic and a quantitative description of the entire 3D foot shape is needed. Moreover, the analysis of foot shape on a population level, such as normal feet, would be beneficial in terms of comparing an individual foot shape to this normative population to examine the shape differences. In this research, the focus is on the automated assessment of 3D foot shape as represented by a high-resolution triangular mesh. A population of these shapes could be quantitatively described through statistical shape analysis. This analysis provides a statistical shape model (SSM) that consists of the average 3D foot shape of the population and the main shape variations present in the population. Such an SSM could then be used to identify foot shape abnormalities. These techniques can then serve as a possible response to current challenges. This thesis presents the main challenges present in the field of foot assessment along with the proposed SSM approaches to address them.

Piergiorgio Panero

• 13/10/2022
• 4 p.m.
• Venue: Campus Middelheim, G.006
• ​Online PhD defence​
• Supervisors: Boris Shoikhet & Wendy Lowen
• Department of Mathematics

Abstract

Differentially graded categories (abbreviated: dg-categories) are fundamental objects in algebraic geometry and higher category theory. After their introduction by Kelly to homologous algebra, dg-categories were used by Bondal-Kapranov to enhance triangulated categories. Since then, the study of dg-categories and their category C_dg(k) has increased intensively.

Together with C_dg(k), “relaxed” versions of dg-categories were introduced, under the name A_∞-categories. Where a dg-category is a dg-graph with an associative and unital composition operation, an A_∞-category is a dg-graph with a unital composition that is only associative up to homotopy. Building on this, in his work on Homologous Mirror Symmetry, Fukaya introduced weak unital A_∞-categories by requiring that unitality holds only up to homotopy. Since then, weak unital dg- and A_∞-categories have been studied by many authors.

It would be advantageous to have a more relaxed model category (than the model category of small dg-categories), which consists of the small, weak unital dg-categories, and which is Quillen equivalent to the model category of small dg-categories. We completely solve this problem in the first part of this work.

In the second half of this thesis, we deal with a second problem. Given a monoidal k-linear category C, then there are two deformation complexes associated with it: the first is the well-known cohomological Hochschild complex CH^q(C, C) (associated to any k-linear or dg-category ) whose cohomology controls the deformations of the k-linear structure (i.e. the composition); the second is the Davydov-Yetter complex. The cohomology of this complex describes the infinitesimal deformations of the monoidal structure of a k-linear monoidal functor F : C → D or the tensor structure of a k-linear monoidal category (when F = Id_C).

So in order to check the complete deformation theory of C, one has to “merge” these two complexes. We achieve this in this work by constructing a functor A(F, F ) : Θ_2 → Vect(k) for a k-linear monoidal functor F : C → D such that the 2-cocellular totalization Tot_{Θ_2}(A(F, F )) is the desired deformation complex. Moreover we give an intrinsic homological algebra interpretation of our complex A(C, D)(F, G)(η, θ) and we show that the 2-cocellular totalization Tot_{Θ_2}(A(F, F )) has a structure of an algebra over an operad homotopically equivalent to Ch_q(E_2;k).

Mohammad Samanipour

• 26/09/2022
  
• 4 p.m.
• Venue: Campus Drie Eiken, O.04
• ​Online PhD defence
• Supervisors: Sabine Van Doorslaer & Hong Yue Vincent Ching
• Department of Physics

Abstract

The interest in environmentally friendly production of chemical products has increased in recent years. Electrocatalysis provides a clean and inexpensive method for chemical synthesis, which can play its role in green chemistry if it is employed at optimum conditions. Decreasing the energy cost is the crucial challenge in performing electrosynthesis in large-scale industrial productions. To this end, an in-depth understanding of what exactly happens during the reaction is necessary, and it requires the identification of the short-lived intermediates. These intermediates are formed due to electron transfer in the electrochemical process, so they often are paramagnetic species such as organic radicals and transition-metal complexes in their specific oxidation states. Electron paramagnetic resonance (EPR) spectroscopy is used to detect paramagnetic species. Using different EPR techniques, one can identify the molecular structures. The combination of electrochemistry and EPR (SEC-EPR) can help to detect the paramagnetic intermediates and unravel the reaction mechanisms, which eventually leads to overcoming the energy cost challenge. Despite its benefits and long history, the SEC-EPR field has not received a lot of attention due to its challenges. As mentioned above, the intermediates are often short-lived, so the conventional ex-situ SEC-EPR experiments are not practical in many cases. Hence, in many cases, in-situ experiments need to be performed to get real-time information from the reaction. Performing in-situ experiments are challenging due to the lack of commercial SEC-EPR cells. Other methods for stabilising the intermediates, such as spin trapping and freeze-quenching, can be combined with in-situ and ex-situ experiments. This thesis focuses on both in-situ and ex-situ experiments, introducing novel SEC-EPR cells and the challenges of designing them and investigating the electrochemical reactions. Spin-trap and direct EPR techniques are employed depending on the requirements for each case study. Moreover, the freeze-quenching of an SEC-EPR cell to trap intermediates is tested. DFT computations completed the EPR data in the identification of paramagnetic species.

Cristina Ariza Carricondo

• 23/09/2022
• 4 p.m.
• Venue: Campus Middelheim, G.010
• Supervisors: Ivan Janssens, Caroline Vincke & Goedele Verreydt
• Department of Biology

Abstract

Future extreme climate events are expected to increase in frequency and intensity with more rainfall events in the wet winter and longer droughts during summer with occasional extreme precipitation events. Such altered climatic conditions will potentially affect the soil's biochemical properties and structure and could cause event-driven carbon exports through dissolved losses. The increased frequency and intensity of high rainfall events can be expected to alter the production and leaching of dissolved organic carbon (DOC), impacting on the C balance at the ecosystem level and potentially at regional and global scale.

In order to quantify DOC leaching, it is essential to accurately quantify the amount of water passing from the surface to deeper layers in the soil. The method applied in this thesis was based on a self-designed 3D printed Zero Tension Lysimeter (ZTL3D) validated with a Water Bucket Model (WBM). DOC concentrations from the leachate captured by the ZTLs3D, together with the water drainage estimations by the WBM, provided a reliable method to quantify DOC exports. Once this methodology was achieved, I focused on studying the concentrations and fluxes of DOC produced under specific rain simulation experiments in the field.

The first rain simulation experiment took advantage of the natural summer drought in 2019. The main finding of this simulation during summer showed the importance of the soil's hydrological state prior to a heavy rain event. The second rainfall simulation showed a clear effect of vegetation dormancy and revealed a clear relation between the amount of water added and the size of the DOC flux during both, autumn and winter.

It can be concluded that DOC leaching was clearly affected by the extreme rainfall events. Comparison with previous research conducted under long-term average conditions showed that extreme events would increase up to double the DOC average flux calculated previously for the site, meaning a total of up to 1.4 t (C) ha−1 year−1 export from the ecosystem, representing up to 20% of the net ecosystem production (NEP). Bigger DOC exports caused by extreme rainfall events might affect the ecosystem functioning. The findings from this doctoral thesis show the importance of accounting not only for DOC exports in the C balance but also the need to account for the potential DOC export under future extreme events. ​ ​

Arne Mertens

• 22/09/2022
  
• 1.30 p.m.
• Venue: Campus Drie Eiken, O.01
• ​Online PhD defence
• Supervisor: Wendy Lowen
• Department of Mathematics

Abstract

Simplicial sets are geometric objects obtained by gluing simplices (vertices, edges, triangles, tetrahedra, ...) together. They were designed to do homotopy theory in a combinatorial way, and as such they are fundamental to the fields of algebraic topology and higher category theory.

Many mathematical objects with some kind of "homotopical" or "higher" structure (e.g. topological spaces, 2-categories, dg-categories, ...) can be turned into a simplicial set by means of so-called "nerve constructions". In this way, the homotopical properties of these objects can be studied through their associated simplicial sets. The simplest of these constructions is the ordinary nerve of a category.

The starting point of this thesis is to consider the nerve of categories which come equipped with more structure, called "enriched categories". It turns out that the classical nerve breaks down in this case. To resolve this issue we introduce a generalization of simplicial sets called "templicial objects". Following an observation of Leinster, these are defined as certain colax monoidal functors into a chosen monoidal category V. One could take V = Ab to be the category of abelian groups for example. In this case, the collection of n-dimensional simplices with fixed endpoints is now no longer a plain set, but an abelian group.

The bulk of the thesis is devoted to constructing templicial variants of classical nerve constructions, specifically the ordinary nerve, Cordier's homotopy coherent nerve and Lurie's differential graded nerve. The latter variant is called the "linear dg-nerve" and associates to every dg-category a templicial abelian group. It turns out that all of these nerves come equipped with natural extra data which we call "Frobenius structures". Especially when V = Ab, these have quite nice properties. In particular the linear dg-nerve provides an equivalence between non-negatively graded dg-categories and templicial abelian groups with Frobenius structures.

Finally, "quasi-categories" are special simplicial sets. They are models for higher categories which were developed by Joyal, Lurie and others. Similarly, we define "quasi-categories in V" as certain templicial objects and all of the nerves above are examples of these. A lot of questions about their homotopical properties remain. In the final chapter of the thesis we discuss some avenues for answering them.

In my talk I will present the main ideas and results appearing in the thesis. There will also be cats.

Nicolas Daans

• 21/09/2022
  
• 3 p.m.
• Venue: Campus Middelheim, G.010
• ​Online PhD defence
• Supervisors: Karim Johannes Becher & Philip Dittmann
• Department of Mathematics

Abstract

Answering a famous question of David Hilbert, it was shown by Davis, Putnam, Robinson and Matiyasevich in 1970 that there can never be an algorithm which can decide whether or not a given polynomial with integer coefficients has a zero consisting of integers. On the other hand, we know (by the work of Tarski around 1950, and subsequent developments in real algebra) that there exist algorithms which can decide whether a polynomial has a zero consisting of real numbers. Where exactly the boundary lies between decidable and undecidable problems in algebra, is a topic of ongoing research.

Existential first-order definability of subsets in rings and fields constitutes a crucial tool for studying these decidability problems. In this thesis, we study existential definability of subsets of fields. If K is a field and n a natural number, then a subset of K^n is called existentially definable if it is the projection of a zero set of a system of polynomials in K^(n+m) for some natural number m. This number m is called the number of quantifiers. To classify which subsets of K^n (even for n=1) are existentially definable, is in general a profoundly challenging problem.

We look at two aspects of existentially definable sets. In the first part of the thesis, we make a quantitative study of existentially definable sets. For example, it might be that a certain set is naively seen to be existentially definable with 10 quantifiers, but that it is in fact even existentially definable with 2 quantifiers by a more elaborate construction. In this thesis, we investigate necessary and sufficient conditions for a set to be existentially definable with a given number of quantifiers. The hope is that these types of quantitative questions can deepen our insight into which sets are existentially definable at all over certain fields. ​ ​

In the second part of the thesis, we study how modern techniques from the research area of quadratic form theory can be used to establish existential definability of subsets of fields, in particular of subrings or complements of subrings. We develop a methodological framework in which we are able to recover and extend signficantly several results concerning definability of subrings and their complements in fields (in particular number fields and function fields). Essential ingredients from quadratic form theory which we invoke include local-global principles and reciprocity laws.

Nick Gys

• 19/09/2022
  
• 4 p.m.
• Venue: Campus Drie Eiken, Building Q, Promotiezaal Q0.02
• Promotoren: Vera Meynen, Steven Mullens & Bart Michielsen
• Departement Chemie

Abstract

The recovery of Platinum-Group-Metals (PGMs) from secondary sources such as spent industrial catalysts is of increasing concern from both economic and environmental point of view. On industrial scale, PGMs are recycled by means of pyrometallurgical processes, followed by hydrometallurgical refining to acidic aqueous leachates. Given the low concentrations (ppm-ppb) of PGMs and complexity of these feed streams, selective adsorption is considered a promising recovery method.

Mesoporous SiO2 materials grafted with organosilanes containing N, S functional groups, are by far the best documented in literature. However, these materials require strict synthetic control to prevent the formation of a disordered organic layer, and under prolonged exposure to acidic conditions, leaching of the grafted groups becomes prominent. Ensuring a full monolayer coverage increases the stability, but reduces to some extent the tailoring of the surface properties. Grafting transition metal oxides with organophosphonic acids (PAs) has been reported as an interesting alternative, offering more stable bonds. Depending on the applied modification conditions, the grafting can be tuned towards different surface properties such as the surface coverage of functional groups, their bonding states and their distribution.

To adjust the surfaces to the application needs, the synthesis-properties-performance correlation needs to be understood. However, to date, in-depth insights into the synthesis-properties correlation are lacking for PAs containing hetero-groups. This know-how is crucial for a rationalized design of materials with clear insights into their applicability and which specific surface properties govern the performance. Therefore, the goal of this PhD is to unravel the synthesis-properties correlation on the grafting of mesoporous TiO2 powder with amino- and mercapto-alkylphosphonic acids, using a combination of experimental and computational approaches to strengthen our conclusions during in-depth surface characterization of these materials. Furthermore, palladium adsorption is envisioned as application domain to study the impact of the synthesis conditions and type of PA on the material’s performance.

In a second step, the knowledge transfer is made from powder to 3D-structured TiO2 materials. 3D shaping and the development of structured materials is a key step towards their successful implementation under industrial conditions. Recent advances in additive manufacturing has resulted in the formation of hierarchically 3D-structured materials with a high geometric flexibility and controllable meso-and macrostructures. As knowledge on the PA grafting of such materials is currently lacking, the impact of varying macroscopic properties of 3D printed titania structures on the extent of grafting throughout these materials is also studied in this PhD.

Jeremy Van den Eynde

• 09/09/2022
  
• 4 p.m.
• Venue: Campus Middelheim, G.010
• ​Online PhD defence
• Supervisors: Chris Blondia & Marc Moonen
• Department of Computer Science

Abstract

This thesis covers two parts: (I) the analysis of the end-to-end delay and delay variation of a strict priority scheduler for a particular combination of traffic inputs, and (II) cross-layer allocation of resources in shared systems.

In part (I) we develop expressions for the end-to-end (E2E) delay and delay variation distributions, for the different classes of aggregate traffic that are served by a strict priority scheduler. This can help to dimension the network and ensure the QoS is not violated. ​ ​

We characterize the busy period, taking low priority traffic into account, of all priorities in order to calculate the additional delay each traffic class encounters. In particular, we characterize the busy period of any aggregate of constant bit-rate sources. We use those distributions to obtain the E2E delay bound, taking the through-traffic and cross-traffic (CT) into account. Methods used in literature are usually limited to two priority classes or do not account for the through-traffic. We evaluate our approach using simulation of a network, and find that our expressions are able to upper bound the E2E delay and delay variation for all the considered traffic priorities. ​ ​

In part (II) we look at cross-layer resource allocation, in which information is shared beyond the usual OSI layers, in order to increase performance. The research presented here applies a novel cross-layer scheduler, called the minimal delay violation (MDV) scheduler, that can be used when the achievable data rate of a user depends on the data rates the other users receive.

We apply the MDV scheduler in a DSL context, where cross-talk between the copper cables of users reduces the maximum simultaneously achievable data rates, and long-term evolution (LTE) and 5G contexts where data rates can vary strongly from moment to moment. Through simulations, we show good performance of our scheduler, with respect to the delay and throughput. ​ 

We additionally implement and discuss an algorithm to constrain the service rates in these settings. Finally, one other cross-layer allocation algorithm is developed that can be used when service rates are assigned dynamically, but there is a large delay between requesting and receiving the data rate, such as for example in satellite communication networks.

Ashish Kumar Sultania

• 08/09/2022
• 10 a.m.
• Venue: Campus Middelheim, A.143
• ​Online PhD defence​
• Supervisor: Jeroen Famaey
• Department of Computer Science

Abstract

The Internet of Things (IoT) envisions billions of smart sensors and actuators to be connected to the Internet. These devices impose three main requirements on wireless networks: wide transmission range, low cost, and low power consumption. The last requirement is crucial for the devices to have long autonomy, thus reducing the need for device maintenance. Low-power networks, such as Low Power Wide Area Network (LPWAN) for long-range transmission and Wireless Personal Area Networks (WPAN) for short-range transmission, promise to fulfil the stated IoT requirements. The LPWAN Narrowband Internet of Things (NB-IoT) and WPAN Bluetooth Low Energy (BLE) radio technologies are popular and enable low-cost and low-power devices. Such communication technologies optimize energy consumption by introducing a sleep or deep sleep state, allowing the device to spend most of its time in a low-power state. This potentially enables IoT devices to operate for several years, powered by small coin cell batteries. Nevertheless, while the device is in this sleep state, it is unreachable and cannot immediately receive any downlink data. Thus, saving the battery life comes at the cost of downlink data latency. The uplink data communication is mostly controlled by the device and is therefore less impacted, as the device can wake-up when it needs to transmit data. Despite being energy-efficient technologies, most IoT devices still require frequent manual battery replacement. This increases operational costs and limits the feasibility to deploy in remote areas, and also affects the environment due to the harmful chemicals that the discarded batteries can leak into the soil. Therefore, the interest in energy harvesting from ambient sources and usage of a small capacitor to store the energy has been rising. A key challenge in ensuring sustainable energy harvesting is dealing with the time-varying energy supply and changing device power demand. The much smaller energy density of the capacitor and unpredictable availability of harvested energy result in intermittent on-off behaviour of the device. This means a batteryless device can turn on and off frequently, and that can further impact its data latency. This thesis studies the trade-off between energy consumption and communication latency for BLE Mesh and NB-IoT. Additionally, batteryless prototypes based on indoor light are also developed by leveraging the Bluetooth Mesh friendship and NB-IoT power saving features. The suitability of using Bluetooth Mesh and NB-IoT in the context of intermittently powered devices is also evaluated by analyzing their energy requirements in different radio states.

Jacob Snoeijer

• 30/08/2022
  
• 3 p.m.
• Venue: Campus Middelheim, G.010
• Supervisors: Karel in 't Hout & Wim Vanroose
• Department of Mathematics

Abstract

A lot of phenomena observed around us can be described in terms of mathematical problems or equations. Although the computational power to numerically solve these problems has extensively increased over the past decades, the mathematical equations to solve have become more and more complicated. ​ ​

The aim of this thesis is to study and develop efficient numerical methods to approximate solutions to high-dimensional partial differential equations (PDEs). The approximation techniques under consideration are based upon two, distinct approaches.

The first idea replaces a single high-dimensional partial differential operator by a linear combination of multiple low-dimensional partial differential operators. Inspired by the principal component analysis (PCA), Reisinger and Wittum suggest a transformation of the covariance matrix that appear in the multi-dimensional Black-Scholes equation. In financial applications the eigenvalue corresponding to the first principal component is often dominant and this observation will be exploited in this first approximation approach. It turns out that neglecting all other principal components does not yield a good approximation but adding first-order corrections yields a good PCA-based approximation for the Black-Scholes operator. The main advantage of this PCA-based approximation approach is that an analytical approximation to the solution of the Black-Scholes PDE is obtained in terms of solutions to only one- and two-dimensional PDEs. These PDEs are independent of each other and can therefore be solved in parallel. ​ 

The second approach restricts the rank of the solution of a differential equation and derives a differential equation for the low-rank components. For example, a numerical representation of the solution of a two-dimensional problem on a certain grid can be represented by a matrix. From that matrix a singular value decomposition can be computed to obtain the singular values with the left- and right singular vectors. It is observed that the solution of certain Helmholtz problems that appear in scattering problems are of low rank. Thus instead of solving a differential equation on a full grid the differential equation is projected on the space spanned by the other factor matrices. This leads to an equation for the remaining low-rank factor of the solution. The equation for this low-rank factor can be related to equations the arise in the coupled channel technique. This idea for two-dimensional problems can be extended to larger dimensional problems where we obtain a low-rank Tucker tensor representation of the solution. ​ ​

Fatma Kader Bingöl

• 06/07/2022
  
• 3 p.m.
• Venue: Campus Middelheim, G.010
• ​Online PhD defence
• Supervisor: Karim Johannes Becher
• Department of Mathematics

Abstract

In this thesis, the main objects of our interest are central simple algebras over a field and involutions on these algebras. The focus is given largely to algebras of exponent 2.

In the first part of the thesis, we consider the problem of bounding several numerical invariants attached to a central simple algebra, such as the symbol length, the index of the algebra and the degree of a 2-extension splitting the algebra. ​ ​

In the second part of the thesis, we consider the problem to determine whether a given involution on a central simple algebra is isotropic. We also view involutions as adjoint involutions to non-degenerate hermitian or skew-hermitian forms. Therefore, the study of isotropy of involutions brings us to the problem of bounding the hermitian u-invariants, which we also work out. ​ ​

To study these problems, we benefit from the theory of quadratic forms.

​Ensar Zeljković

• 27/06/2022
  
• 9.30 a.m.
• Venue: Campus Middelheim, G.010
• ​Online PhD defence
• Supervisors: Steven Latré & Johann Marquez-Barja
• Department of Computer Science

Abstract

Wireless networks are a very complex and dynamic environment that requires proper management to optimise its performance. With a tremendous increase in the number of user devices and their data traffic, wireless networks are becoming denser. Therefore, optimising mobility management becomes a crucial challenge. In heterogeneous wireless networks, the usage of multiple wireless interfaces is inefficient and suboptimal. Solutions require modifications to the user device limiting the use cases. Reactive approaches are too slow to react and solutions often reside on the user side, which creates suboptimal perdevice optimisations. To tackle these challenges we use SDN. SDN monitors the network by collecting multiple metrics on the network side and creates a global overview of the network. This overview can be used in SON algorithms to proactively trigger global optimisations. We create and use different SDN frameworks for different wireless technologies and create SON algorithms on top to optimise mobility.

In heterogeneous wireless networks, we introduce ORCHESTRA, an SDN framework to manage different devices and do packetlevel dynamic and intelligent handovers, load balancing and replication. On top of ORCHESTRA we present a novel MIQP load-balancing optimization problem formulation for mobility management which can double the network-wide throughput across different scenarios.

To avoid user device modifications, we introduce the HuMOR framework for handover management in WiFi. On top of HuMOR we introduce ABRAHAM a handover algorithm that uses multiple metrics to predict the future state of the network and optimize the AP load. We show that ABRAHAM can achieve a 139% throughput improvement over the IEEE 802.11 handover algorithm. ​ 

To demonstrate that our WiFi solution is capable to address different wireless contexts, we use the same principles in mobile networks. On top of an OpenRAN compliant framework dRAX, we introduce MOLA_ADNA. This adaptive and QoS aware handover algorithm takes multiple metrics into account to do network level optimization of the cell load. We show that the mean throughput with MOLA-ADNA was higher by 25% compared to A3. ​ 

Finally, proper network element configuration is also crucial for network performance and mobility management. We, therefore, create the ALPACA algorithm on top of dRAX to optimise the PCI value of cells in mobile networks. ALPACA adapts to dynamic network topology changes and continuously optimizes the network. The results show that ALPACA can resolve all collisions and confusion for up to 1000 cells and minimize the effects of inevitable modulo PCI issues.

Rituraj Borah

• 21/06/2022
  
• 2 p.m.
• Venue: Campus Drie Eiken, O.04
• ​Online PhD defence
• Supervisors: Sammy Verbruggen & Silvia Lenaerts
• Department of Bioscience Engineering

Abstract

Photoactive nanoparticles and their light-driven applications have gained tremendous scientific attention towards remediation of the global environmental problems, meeting alternative energy demands, and other new technological discoveries. The research work presented in this dissertation includes a fundamental investigation of such nanoparticles to gain deeper insights that will ultimately benefit their application. In particular, the study of plasmonic metal nanoparticles and metal oxide nanoparticles for light driven applications is the major theme of this work. The investigation begins with isolated plasmonic Au and Ag nanoparticles, followed by a natural extension to nanoparticle clusters, and then further to nanoparticle films. Next, the application of such plasmonic nanoparticle films for gaseous phase sensing of volatile organic compounds is explored. Finally, the film formation of metal-oxide nanoparticles by self-assembly is investigated for the fabrication of photoactive functional interfaces.

The fundamental theoretical investigation of the isolated plasmonic nanoparticles encompasses alloy and core-shell nanostructures of Au-Ag bimetallic compositions. First, the optical properties of bimetallic alloy and core-shell nanoparticles are compared for different structures such as nanospheres, nanotriangles and nanorods. Based on the optical properties, the photothermal properties of these nanostructures are also evaluated for relevant light-driven applications. Further, to bridge the gap between the theoretical and experimental optical properties of colloidal plasmonic nanoparticles, the effect of different statistical parameters pertaining to the particle size distribution is studied. Going from isolated nanoparticles to nanoparticle clusters, the changes in the optical properties of plasmonic nanoparticles when they form finite clusters is investigated. A strong effect of clustering on the absorption intensities of the nanoparticles and hence, on the photothermal properties is found. Next, for the study of plasmonic nanoparticle infinite arrays, Au and Ag nanoparticles films are experimentally obtained by the self-assembly at the air-ethylene glycol interface. Upon further validation of the computational models with the experimental optical properties of these films, the near-field and far-field optical response of the plasmonic nanoparticle arrays is investigated. An application of the self-assembled Au nanoparticle film is then demonstrated in the sensing of volatile organic compounds (VOCs). Finally, the focus is shifted from plasmonic nanoparticles to metal oxide nanoparticles for their self-assembly at the air-water interface to obtain self-assembled films. For this, the hydrophobic functionalization of four metal oxides nanoparticles namely, TiO2, ZnO, WO3 and CuO is investigated.

The insights from this work is useful for the design and fabrication of functional nanoparticles and interfaces for light driven applications. ​ ​

Senne Van Putte

• 07/06/2022
• 4 p.m.
• Venue: Campus Groenenborger, T.148
• ​Online PhD defence​
• Supervisor: Pierre Van Mechelen
• Department of Physics

Abstract

The standard model of particle physics has booked enormous successes in the past few decades. It is able to describe experimental processes with astonishing precision and predicted the existence of several particles, such as the top quark, and the Higgs boson. Being a fundamental description of the processes of nature makes it one of most well-tested theories that ever existed. However, many observations indicate the presence of phenomena beyond the standard model. ​ ​

An example thereof are the many astronomical measurements indicating the presence of additional unseen sources of gravity. Movements of faraway galaxies cannot always be explained by the gravity of the observed matter in its environment, hinting the presence of dark matter. Whether it can interact with the standard model particles, besides through the gravitational force, remains an open question, although cosmological measurements indicate an additional interaction that cannot be much stronger than the scale of the weak force. Many expansions to the standard model have been proposed, attempting to provide particles fulfilling the role of dark matter. ​

In this work, a generalised model is examined, specifying the mediators between the hidden particles and the observable particles. The lightest particle of the dark sector corresponds to the dark matter candidate. Other content of the hidden sector is intentionally left out of the model description to widen its relevance. To abide the cosmological observations, an additional U'(1) local gauge symmetry is introduced, yielding an extra Z'-boson. The signature feature of the model is the inclusion of a new scalar field, spontaneously breaking this U'(1) symmetry, thereby generating the mass of the particles in the hidden sector, and adding a dark Higgs boson to the model content. 

Predictions of the dark Higgs model decaying the dark Higgs to two W-bosons in the semi-leptonic decay channel, are held against observed data from the LHC machine, collected with the CMS detector. The data sample was gathered over the run 2 period spanning three years 2016, 2017 and 2018, with an integrated luminosity of 137/fb. In this period the LHC was operating with a centre of mass energy of 13 TeV. ​ ​

Results are combined with the full-leptonic decay channel, observing no significant excess over the standard model predictions. A large portion of the dark Higgs model phase space is excluded.

Timour Ichmoukhamedov

• 03/06/2022
• 10 a.m.
• Venue: Campus Drie Eiken, Building R, Room R.2
• ​Online PhD defence​
• Supervisor: Jacques Tempere
• Department of Physics

Abstract

An impurity in a medium can become dressed by the excitations of the medium and form the polaron quasiparticle. This behavior is best known in the case of an electron moving through a crystal lattice, but has also been observed across various other physical systems. Recently, the Bose polaron where an atomic impurity is immersed into a Bose-Einstein condensate, has been experimentally observed. Spurring further extensive study of this system, the Bose polaron has attracted significant theoretical interest, as it exhibits a number of theoretical challenges that have previously not been encountered in other polaronic systems.

One of the seminal tools to study polaronic systems is Feynman’s variational path integral approach. In this formalism, the medium surrounding the impurity is integrated out such that the impurity is described through delayed interactions with itself at previous times. Hence, the polaron problem is closely related to the problem of capturing memory effects in the path integral approach. This thesis presents different extensions of treating general memory effects with path integrals to capture the challenging physics of the Bose polaron problem. ​ ​

Three questions are investigated in detail. In the literature, the path integral approach has already been applied to the Bose polaron in the past, but more recently this application was shown to be insufficient to capture the subtle UV behavior of the system. In the first part of the thesis, we investigate which extensions to the method should be made to correct this shortcoming. Next, we discuss the inclusion of the so-called extended Fröhlich interactions that become important at strong coupling between the impurity and the medium. Finally, we explore how these methods can also be extended to describe many impurities in a Bose-Einstein condensate.

Bertold Mariën

• 02/06/2022
  
• 2.30 p.m.
• Venue: Campus Drie Eiken, building R, room R3
• ​Online PhD defence
• Supervisor: Matteo Campioli
• Department of Biology

Abstract

Knowing the leaf senescence and fine-root phenology of deciduous trees is important to understand forest ecosystems and their feedback on the climate. To improve our knowledge on the phenology of deciduous trees a paradigm was suggested: the LEAF-FALL paradigm. This paradigm proposed that the onset of leaf senescence is not directly impacted by changes in the photoperiod and other environmental drivers (e.g. the temperature, and the light intensity and spectral quality) but rather indirectly through the impact of environmental drivers on tree growth, including fine-root growth. The cessation of the wood growth and the photoperiod would then act as the trigger of leaf senescence at the end of summer in favorable and unfavorable conditions, respectively. This work addresses here just two aspects: (I) determining the timing of the leaf senescence and (II) determining the fine-root phenology throughout the year to investigate a possible coupling between the phenology of the leaves and fine-roots. The objectives were investigated through observations of the below- and aboveground phenology in mature deciduous trees (Fagus sylvatica L., Quercus robur L., Betula pendula Roth. and Populus tremula L.) in Belgium. Additional leaf senescence data from the same species in Norway and Spain were also used. The leaf senescene and fine-root growth have been investigated mostly through measurements of the chlorophyll content index in leaves and through observations of fine-roots using a minirhizotron. Consequently, substantial attention has also been given to detecting phenological transition dates in longitudinal data of the chlorophyll content index and fine-roots. Especially the use of generalized additive models (GAMs) and generalized additive models for location, scale and shape (GAMLSS) has been explored in relation to the detection of phenological transition dates. The results show that the global radiation, temperature and vapor pressure deficit affected the timing of the onset of leaf senescence in all species. In Fagus sylvatica, unlike in Quercus robur and Betula pendula, the timing of the onset of leaf senescence was conservative (i.e. not varying much). A significant inter-annual trend in the onset of leaf senescence could not be established in any of the species. The seasonal trend in the fine-root growth was also inconsistent among species and years, although species with the same life strategy did show similar phenological trends and similar fine-root lifespan and turnover rate estimations. Our results about the two researched aspects do not provide immediate support for the LEAF-FALL paradigm.

Gilles De Meester

• 31/05/2022
• 6 p.m.
• Venue: Campus Drie Eiken, O.06
• ​Online PhD defence​
  
• Supervisors: Raoul Van Damme & Panayiotis Pafilis
• Department of Biology

Abstract

The evolution of cognition is one of the most enigmatic topics within biology. In particular, which exact (socio-)ecological forces shape the evolution of cognition remains unclear. Both the spatial complexity and the temporal variability of the environment have been hypothesized to be major selective drivers behind cognitive evolution, but evidence from previous research is sparse and mixed. During my PhD, I studied the role of ecology in cognitive evolution, by looking at variation both across and within species of lacertid lizards. The first part of my thesis investigated cognitive variation at the intraspecific level. I looked at the evolution of relative brain size across Squamata (lizards + snakes). In contrast to expectations based on literature, brain size was unrelated to habitat complexity, and social species had relative smaller brains than solitary species. Next, I measured and compared five aspects of cognition across thirteen species of lacertid lizards. Albeit species varied considerably in their performance on all five tests, this variation was largely unrelated to differences in their ecology and life-history. The sole exception was that species from more seasonal habitats tended to exhibit lower behavioural flexibility. The second part of my PhD focused on the link between environment and cognitive variation among populations of the Aegean wall lizard (Podarcis erhardii). Wall lizards from a more seasonal island habitat performed better on a spatial learning task than conspecifics from a less variable mainland location, but also demonstrated lower cognitive flexibility. Secondly, lizards from structural complex habitats were superior spatial learners compared to lizards from more simple environments. Behavioural covariance between lizard personality and cognition was also often year- and habitat-dependent. Finally, I investigated variation at the individual level. I specifically tested whether cognition would be more advantageous in complex habitats. Lizards with known cognitive abilities and personality traits were released in large outdoor enclosures characterized by either complex or simple vegetation for 11-12 months to measure their fitness. Spatial learning and problem-solving were indeed associated with survival, albeit in unexpected ways, but there was no link between cognition and reproductive success. I also found no evidence for habitat-dependent selection on cognition. The link between ecology and cognition is thus not straightforward, and may depend on the cognitive trait and taxonomic level under investigation. Altogether, this thesis illustrates how an integrative approach, looking at both macro-evolutionary patterns and selection within species, can provide valuable new insights in the evolution of animal cognition.

Ken Schoutens

• 30/05/2022
• 4 p.m.
• Venue: Campus Drie Eiken, Promotiezaal Q002
• Supervisor: Stijn Temmerman
• Department of Biology

Abstract

Global climate changes impose multiple challenges, including the increasing risk of coastal flood hazards due to sea level rise and increasing impact of storm activity. Social and economic cost of flood hazards are huge since low elevated coastal zones often house densely populated and industrialized communities. Hence, there is a strong urge to implement sustainable climate adaptation strategies. In this context, nature-based shoreline protection approaches such as conservation or (re)creation of tidal marsh ecosystems provide multiple opportunities. Nevertheless, effective implementation of tidal marsh ecosystems as a complementary shoreline protection is hampered by the uncertainties about their effectiveness and reliability. This thesis elaborates on the role of species-specific plant traits, how they vary spatially and over time, how they interact with hydrodynamics and sediment dynamics and, how these feedbacks contribute to the shoreline protection capacity of tidal marshes. Field monitoring and field experiments were done along the brackish part of the Elbe estuary, Germany and flume experiments took place in the Large Wave Flume, Hannover, Germany and the Mesodrome tidal flume facility, Antwerp, Belgium. This thesis provides new insights in the role of species-specific plant traits within the mutual interactions between hydrodynamic forces, sediment dynamics and vegetation that are crucial in understanding the spatial-temporal shoreline protection efficiency of tidal marshes. We show that hydrodynamic forces from waves and currents form a not to be neglected additional stressor for plant growth and survival and we illustrate how these forces constrain suitable conditions for successful tidal marsh conservation, restoration and creation. Moreover, we argue how providing enough space to allow marsh expansion will increase the resilience and reliability of the nature-based shoreline protection function of tidal marsh in a changing climate.

Sándor Hajdú

• 30/05/2022
• 4 p.m.
• Venue: Campus Middelheim, G.010
• ​Online PhD defence​
• Supervisors: Tom Mestdag & Sonja Hohloch
• Department of Mathematics

Abstract

Similar to the case of Riemannian geodesics, one may define  conjugate points for systems of second-order ordinary differential equations (SODEs). In this dissertation we provide a method to find such conjugate points and we apply our results to locally symmetric SODEs and to Lagrangian systems that admit a symmetry Lie group. Sprays are a specific type of SODEs. Two sprays are said to be projectively equivalent if they have the same geodesics, when viewed as point sets. We exploit the freedom in the choice of a representative of a projective class of sprays in the search for their conjugate points. In the context of Finsler geometry, we use the theory of cut points to draw a conclusion about the existence of conjugate points for a class of Randers-type Finsler metrics. ​ ​

Next, we investigate nonlinear splittings on fibre bundles. These can be thought of as generalizations of Ehresmann connections. We investigate both the similarities and the differences between nonlinear splittings and Ehresmann connections. We show how certain structure-preserving submersions relate to nonlinear splittings. We also define a curvature map for nonlinear splittings and show how it can be used to investigate questions about the submersiveness of Lagrangian systems of magnetic type. Finally, we apply our results to the context of Finsler geometry, and we provide new, interesting examples of Finsler manifolds.

Gerardo Fracasso

• 24/05/2022
• 4.30 p.m.
• Venue: Campus Drie Eiken, O.03
• ​Online PhD defence​
• Supervisors: Erik Matthysen & Dieter Heylen
• Department of Biology

Abstract

Parasites and hosts are dynamic interactions exerting reciprocal selective pressures. Thus, host-parasite interactions are ideal systems for the study of ecological and coevolutionary processes. However, while the effects of parasites on hosts have been extensively investigated, host-induced parasite evolution and parasite life history have been neglected. In particular, knowledge on the amount of among- and within-individual variation between parasite traits is mostly unknown despite its fundamental importance to understand parasite performance and evolution. In this dissertation, I report four experimental studies investigating several aspects of parasite individual variation in a tick-songbird system, namely the bird-specialized tree-hole tick Ixodes arboricola and its main host, the great tit Parus major.

First, I investigated the behavioural preferences for tick attachment sites on the host body. Experiments were carried out using three tick species differing in ecology and host specificity both with and without grooming restrictions. The experimental findings as well as the literature evidence suggest that ticks prefer to attach to the host head and actively move to this area. I hypothesize that this pattern is consistent throughout ixodid ticks feeding on birds. 

Second, I report how fundamental life-history traits affect individual tick success at every stage, and estimate their phenotypic and genetic correlation between and within stages as well as the trait evolutionary potential (using animal models) for feeding time, engorgement weight, moulting time, and number of hatched eggs. Additionally, I account for the effect on the abovementioned traits of tick sex, maternal effect, host identity, fasting time and batch. Results suggest differences in tick individual quality, for which engorgement weight seems to be a good proxy.

Third, I report a study investigating variation and heritability of host quality from the parasite perspective. Here, I measured to what extent hosts can affect tick performance and life history of larvae and nymphs both on- and off-host. I show that host individual characteristics significantly influenced larva and nymph attachment success. Additionally, hosts had an heritable effect on tick feeding time and, to a lower extent, on several other traits and success parameters. ​ ​

Lastly, I investigated whether I. arboricola males prefer to mate with heavier engorged females in order to obtain a higher fitness. Surprisingly, male mate choice experiments carried out in two different setups showed a lack of preference for heavier females. However, males seem to remember and avoid the mating partners they previously met.

Mohamed Rashad Darwish

• 24/05/2022
  
• 3 p.m.
• Venue: Campus Groenenborger, Z.223
• Supervisors: Albert De Roeck & Nick Van Remortel
• Department of Physics

Abstract

This thesis project conducts a search for a new type of particle using the data of the CMS experiment at the Large Hadron Collider at CERN, in Geneva, Switzerland. The data have been collected in 2016, 2017 and 2018, before the long shutdown started. The new particle is a so called Heavy Neutral Lepton (HNL), a potential new family member of the Standard Model neutrino. Along with the known left-handed neutrinos that interact with the W and Z particles, the νMSM model postulates the existence of three right-handed neutrinos with a small mass, labelled as N1 (a light stable dark matter candidate particle), N2 and N3, the latter two being potentially long-lived particles. The existence of these particles would restore the symmetry in the standard model such that all left handed particles have a right handed partner. The νMSM model, when realized in Nature could e.g. explain the existence of Dark Matter in the Universe, the baryon asymmetry after the Big Bang and the non-zero observed neutrino masses. The exact values of the masses and couplings of these new particles are unknown but in a large part of the preferred theory phase space the N1 is much lighter than a GeV while N2 and N3 can be in the GeV to tens of GeV range, and can be searched for at the LHC.

The thesis contains a summary of the theoretical framework of the SM and νMSM extension, along with a comprehensive description of the CMS experiment at the LHC accelerator complex. This thesis focuses on the search for long lived Heavy Neutral Leptons, both as possible right-handed Dirac or Majorana neutrinos, and is conducted using final states that contain two charged leptons (electrons or muons), jets and displaced vertices. These particles can be produced in proton-proton collisions at the LHC in decays of the produced W and Z bosons through mixing with the standard model neutrinos. Their lifetimes depend on the mass and on the strength of the coupling with standard model neutrinos. For low masses and small mixing parameters, the decay length can be sufficiently large to produce a detectable secondary vertex.

No excess in the collision data is observed with respect to the predicted standard model backgrounds in the searches for all channels, and hence a statistical interpretation of the results is performed to set upper limits in the production cross sections on the νMSM particles. These limits extend in the region beyond previously derived results.

Ece Arslan Irmak

• 23/05/2022
  
• 4 p.m.
• Venue: Campus Groenenborger, T.103
• ​Online PhD defence
• Supervisors: Sandra Van Aert & Sara Bals
• Department of Physics

Abstract

The structure-property relationship of nanoparticles is linked to the positions of the atoms and even small changes in the local structure of a given particle may significantly affect its performance for a specific application. However, when these particles are exposed to application-relevant conditions, such as elevated temperatures or intense light illumination, rapid structural transformations and changes in elemental distribution are observed. In order to gain control over the structure-dependent properties and performance of nanomaterials, an atomic-scale understanding of the ongoing transformations is of crucial importance. In situ transmission electron microscopy (TEM) experiments provide useful information to analyze nanoparticle changes down to the atomic scale during heating. Nonetheless, these investigations are often performed based on 2D TEM images which are usually inadequate to analyze the structure-property relation of nanomaterials because they only provide a projected image of a 3D structure.

This thesis is devoted to presenting robust approaches by joining the scanning TEM (STEM) imaging technique and theoretical calculations to capture 3D atomic-scale transformations of metallic NPs in response to changes in their environment. Within this framework, the research presented in this thesis is conducted twofold. The first part aims to present a refined method to investigate the 3D dynamics of metallic NPs lying on an oxide support based on 2D STEM images acquired during in situ experiments at high temperatures, which is of importance to understand their behavior during catalytic reactions. On the other hand, atomic-scale transformations cannot be captured by only experimental techniques when in situ STEM experiments do not provide the necessary time or spatial resolution or when the specific environmental trigger cannot be applied. Therefore, in the second part of this thesis, 3D experimental characterization techniques are combined with atomistic simulations to extract missing atomic-scale dynamics of NPs. In this manner, the complex atomistic rearrangements upon heating and intense light illumination that cannot be achievable by experimental observations are unraveled for Au-Pt bimetallic and mesoporous silica-coated Au nanoparticles, which are of interest for several plasmonic and catalytic applications.

Ferre Knaepkens

• 20/05/2022
  
• 2 p.m.
• Venue: Campus Middelheim, G.010
• ​Online PhD defence
• Supervisors: Annie Cuyt & Wen-shin Lee
• Department of Mathematics

Abstract

Multidimensional exponential analysis, also known as sparse interpolation or harmonic analysis, is the central theme of the thesis. The overall aim is not only to provide valuable theoretical contributions, but also to bridge the gap between this theoretical foundation and the numerical implementation, to ultimately advance several relevant engineering applications. For this reason, this thesis is divided into four different parts.

The theoretical results are based on the further development of the connections between exponential analysis, Padé approximation theory and tensor decomposition. In particular the connection with Padé approximants is used to address the infamous ill-conditioning and sensitivity to noise of Prony-like methods.

These new theoretical findings are then applied to various challenging application domains. The logical starting point of this endeavour is a one-dimensional problem before moving on to the multidimensional setting. Hence, as starting point we tackled one-dimensional direction of arrival estimation. Subsequently, we explored some multidimensional applications. First, we consider two two-dimensional problems, namely antenna position estimation in the nearfield and the denoising of structured images. Finally, we explored inverse synthetic aperture radar, which is a three-dimensional application. ​ ​

To conclude, we cover a few additional topics that are independent of the previous sections. First, we describe how lower-dimensional exponential models can be blended into a single model in a higher-dimensional space. Next, we examine the feasibility of using exponential analysis to solve problems in dimensions higher than three. Finally, we discuss our efforts to share new developments using the mathematical toolbox Sparsimatics. ​ ​

Jeffrey Cornelis

• 18/05/2022
  
• 3.15 p.m.
• Venue: Campus Middelheim, A.143
• Supervisors: Wim Vanroose & Siegfried Cools
• Department of Mathematics

Abstract

Solving a linear system of equations is undoubtedly one of the most fundamental tasks in numerical linear algebra and optimization. Not only are these linear systems often important on their own, they also appear frequently as sub-problems when attempting to solve a more complex problem. Many applications in science and industry can be modelled using a sparse matrix, i.e. containing only a limited amount of non-zero entries. Moreover, in many cases it suffices to compute an approximate solution to the linear system of equations. Instead of computing the exact solution, Krylov subspace methods iteratively improve an approximate solution until some predefined tolerance is satisfied. The main computational cost of these methods are the matrix-vector products that need to be computed in each iteration. The total number of floating point operations for a matrix-vector product is relatively modest for sparse matrices, even when the number of unknowns is very large, which makes Krylov subspace methods efficient solution methods for sparse linear systems of equations. ​ ​

In the first part of the thesis we develop a reformulation of the Conjugate Gradient method, one of the most widely used Krylov subspace methods, with improved parallel scalability on large distributed-memory computers compared to the standard implementation. The main idea of our approach is to overlap the global communication phase of the dot-product operation with computational work, which can be achieved by introducing suitable auxiliary variables and recurrence relations. ​ ​

In the second part of the thesis we propose new techniques to exploit (generalized) Krylov subspaces in specialized optimization routines, with a special focus on ill-posed inverse problems. We develop a new hybrid regularization method that simultaneously solves a regularized inverse problem and finds the corresponding regularization parameter such that the discrepancy principle is satisfied. We also develop a mixed-precision technique to speed up an interior-point method for solving linear programming problems by using a Cholesky factorization in single precision as preconditioner for the Conjugate Gradient method implemented in double precision.

Pieter Livens

• 17/05/2022
• 4 p.m.
• Venue: Campus Groenenborger, US025
• Supervisor: Joris Dirckx
• Department of Physics

Abstract

In the first step of the hearing process, the eardrum captures sound vibrations that are transmitted by three ossicles to the fluid-filled cochlea. The combined system of eardrum and ossicles is called the middle ear and acts as an acoustic-mechanical transformer. Computer modeling plays a vital role in understanding how the middle ear transports sound energy to the cochlea. Moreover, these simulations allow us to vary the material properties, such as mass distribution and stiffness, of this system to investigate which properties play an important role and at which frequencies. Such variations are difficult or impossible to achieve experimentally, so simulations lead to valuable insights. However, such models depend on well-defined input parameters to obtain valid results.

First, the middle ear of the anolis lizard (Anolis sagrei) was examined using finite element modeling. The computer simulations showed that these animals use internally coupled middle ears to localize sounds. It was found that simulating the effect of the fluids of the inner ear improved the accuracy of the simulations and it is important to include in future studies.

Next, the displacement of the eardrum of the gecko (Gekko gecko) was measured when slowly varying pressure was applied. The largest displacements to date were observed, showing that this middle ear is highly elastic.

Next, the stiffness of the human eardrum was calculated using a new method. A human eardrum was made to vibrate under different sound frequencies. From this vibration data, the displacement and deformation of the eardrum were calculated. Then the so-called virtual-field method was developed for these thin vibrating membranes, which led to the identification of the stiffness of the eardrum. When the displacements become large, the virtual-field method must be extended to include the theory of finite deformations. This new method was validated on rubber membranes, but should certainly be tested on eardrums in the future to find out material values of the eardrum in this case as well.

Finally, the presence of so-called preload was investigated. When the eardrum is at rest, tensile forces from the middle ear can still create internal tension on the eardrum. By making several small incisions across the rabbit's eardrum, it was shown that prestress was definitely present. Future work should investigate whether this effect also occurs in the human eardrum.

Matthew Houtput

• 13/05/2022
  
• 4 p.m.
• Venue: Campus Drie Eiken, Promotiezaal Q0.02
• ​Online PhD defence
• Supervisor: Jacques Tempere
• Department of Physics

Abstract

A conduction electron in a solid is usually described as a free, non-interacting electron. However, in a polar crystal the ions are charged, meaning the electron can interact with the lattice vibrations (phonons). Usually, the electron and phonon cloud are combined into a new quasiparticle, called the polaron. A thorough understanding of electron-phonon coupling is important in solid state physics, since it influences many material properties such as the DC conductivity, optical absorption, and thermal conductivity, among others. Electron-phonon coupling also causes pair formation of the electrons, which leads to superconductivity.

One of the oldest quantum mechanical models for the description of polarons is the Fröhlich Hamiltonian. Although today electron-phonon coupling can be described more accurately with ab initio methods, the Fröhlich interaction is still used because of its relative simplicity. The Fröhlich Hamiltonian assumes a linear electron-phonon interaction, but recently it has become clear that in some materials higher order interaction terms are also important. In this thesis, Fröhlich theory is extended by adding 1-electron-2-phonon interaction, which is the lowest order anharmonic interaction. The central result of this thesis is an analytical expression for the interaction strength of an electron interacting with two longitudinal optical phonons, in the continuum approximation. This interaction strength only depends on one scalar parameter for cubic materials: which significantly simplifies the investigation of this Hamiltonian.

In the remainder of the thesis, the properties of the “anharmonic” polaron are investigated using this extended Hamiltonian. Firstly, the ground state energy and effective mass of the new polaron are calculated using the Greens function formalism and the path integral formalism. With both methods, it is shown that the polaron energy is significantly lowered due to the additional interaction. Furthermore, the optical conductivity of the anharmonic polaron gas is studied. The optical absorption spectrum shows a characteristic secondary peak, which can be used as an experimental fingerprint to measure 1-electron-2-phonon interaction. Finally, we investigate the possibility of bipolaron formation within this new polaron model. In the Fröhlich model bipolarons are only stable when the electron-phonon coupling constant exceeds a certain critical value. When 1-electron-2-phonon interaction is introduced, this critical value is lower, meaning the potential region of bipolaron stability is enlarged.

Kevin van 't Veer

• 10/05/2022
  
• 2 p.m.
• Venue: Campus Drie Eiken, O.01
• Supervisors: Annemie Bogaerts & François Reniers
• Department of Chemistry

Abstract

Ammonia (NH3) synthesis is crucial for the production of artificial fertilizer and is carried out through the Haber-Bosch process. With an energy consumption of 30 GJ/t-NH3 and the emission of 2 kg-CO2/kg-NH3, ammonia is the chemical with the largest environmental footprint. Haber-Bosch operates under high pressure and high temperature conditions. Plasma technology potentially allows greener ammonia production. Dielectric barrier discharges are a popular plasma source in which a catalyst is easily incorporated. The combination of plasma and catalyst can circumvent the harsh reaction conditions of the Haber-Bosch process.

Plasma kinetics modelling is used to gain insight into the mechanisms of such plasma-catalytic systems. Special attention is given to the instantaneous power absorbed by the electrons, the relevant fraction of the microdischarges and the discharge volumes. The importance of vibrational excitation is investigated. Depending on the exact discharge conditions, it was found that both the strong microdischarges and vibrational excitation can be simultaneously important for the ammonia yield.

The temporal behavior of filamentary dielectric barrier discharges was explicitly taken into account. Ammonia was found to decompose during the microdischarges due to electron impact dissociation. At the same time atomic nitrogen and other excited species are created. Those reactive species recombine to ammonia in the afterglow through various elementary Eley-Rideal and Langmuir-Hinshelwood surface reaction steps with a net ammonia gain.

Finally, the concept of the fraction of microdischarges was generalized. It directly represents the efficiency with which the applied electric power is transferred to each individual particle in the plasma reactor. It is argued that any type of spatial or temporal non-uniformity of the plasma will cause unequal treatment of the gas molecules in the reactor, corresponding to a lower efficiency at which the power is transferred to the gas molecules.

All of those insights aid in an increased understanding of plasma-catalytic ammonia synthesis as a potential green chemistry solution to the synthesis of ammonia on small scale.

Janne Spanoghe

• 05/05/2022
  
• 4 p.m.
• Venue: Campus Drie Eiken, Building Q, Promotiezaal Q0.02
• ​Online PhD defence
• Supervisor: Siegfried Vlaeminck
• Department of Bioscience Engineering

Abstract

The human population is projected to grow to 9.7 billion by 2050, resulting in an estimated increase in protein demand of 50%. From an environmental perspective, the current and future demand of protein cannot be sustainably met as the conventional food production chain is severely altering biogeochemical cycles of nitrogen and phosphorus, biodiversity and land-use, with flows towards the biosphere and oceans that are exceeding the planetary boundaries. Microbial protein (protein derived from microorganisms) has been suggested as an excellent sustainable protein source, a fortiori when produced in a land- and fossil free manner. The photoautohydrogenotrophic cultivation (i.e. with light, CO2 and H2) of purple bacteria links up perfectly with the upcoming green electrification of industry (green H2) and the need for carbon capture and utilization. However, this metabolism represented a gap in literature, and thus this thesis aimed to establish a basic knowledge platform on its kinetic, stoichiometric and nutritional performance. At first, three originally photoheterotrophically enriched purple bacteria were studied of which Rhodobacter capsulatus reached the highest protein productivity of 0.16 g protein/L/d, which aligned well with the commonly-known photoautotrophic microalgae. Moreover, a full dietary essential amino acid match was found for human food, while the fatty acid content was dominated by the health-stimulating vaccenic acid (82-86%). Lastly, the achieved protein yield in photoautohydrogenotrophic purple bacteria was 2.3 times higher compared to hydrogen oxidizing bacteria, indicating a resource-efficient use of H2. Next, a photoautohydrogenotrophic enrichment of wastewater treatment microbiomes was performed in search for specialist species. While the isolates of this enrichment showed improvements in their performance during acclimation, the kinetic and nutritional performance of Rhodobacter capsulatus still excelled. Subsequently, the influence of nutrient limitations (C or N) and nitrogen gas fixation was studied on the nutritional tuning potential. Both the limitations as well as the N2 fixation resulted in the shift of the essential amino acid profiles. Additionally, the limitations significantly decreased the pigment content, while an increase in the storage of poly-P was seen in case of carbon limitations. The next major challenge was the production intensification in a photobioreactor of which the design was linked to minimizing both H2 and light limitations. The chosen bubble-column photobioreactor already resulted in a doubled biomass productivity. Finally, the remaining technological and non-technological challenges ahead for the production of a high-value, cost-efficient, environment-friendly microbial protein that complies with legislative requirements and appeals to future consumers were discussed.

Jonas Verspeek

• 27/04/2022
• 5 p.m.
• Venue: Campus Drie Eiken, O.06
• ​Online PhD defence​
• Supervisor: Jeroen Stevens
• Department of Biology

Abstract

Cooperation is a key component of social life but seems an evolutionary puzzle as it involves behaviours that benefit others. Because cooperative behaviours involve an immediate cost to the actor, natural selection has produced mechanisms to regulate cooperation to overcome adverse effects of these costs. The main proximate mechanisms that regulate cooperation are prosociality and inequity aversion (IA). In this thesis, I combined behavioural and physiological measures in different experimental paradigms to explain the variability in these proximate mechanisms of cooperation in bonobos (Pan paniscus). Before focusing on prosociality and IA in bonobos, I implemented three methodological studies. First, I investigated the food preference of the bonobos to decide which food items to use in the experimental paradigms. Second, I provided a biological validation for the use of salivary cortisol to measure stress and arousal in bonobos. Third, I investigated whether bonobos prefer to bond with more similar individuals. To study prosociality, I used a novel juice-provisioning experiment, the prosocial choice task and the group service paradigm and showed that the Zoo Planckendael bonobos mainly behaved out of self-interest and, like chimpanzees, behaved indifferently to the welfare of others. This contrasts with the popular image of the prosocial and food sharing bonobo, who is often portrayed as a “hippie of the primate world”. I concluded that this popular image is mainly the result of an age bias in previous experimental studies that looked for evidence of prosociality in bonobos. To study IA in bonobos, I used the standard token exchange task. To complement the standard behavioural measures with the emotional component of IA, I also investigated a behavioural, rough self-scratching, and a physiological measure, salivary cortisol increase, of arousal. The bonobos reacted to receiving less than a partner while they never refused trials when receiving more than a partner. Also, stronger bonded individuals were more tolerant towards inequity. Further, subjects were more aroused when receiving a better reward than a partner, suggesting that bonobos do notice when being favoured but do not respond to it behaviourally. The results of this thesis provide supporting evidence for the nuanced view of the prosocial, food-sharing and tolerant hippie ape. I demonstrated that adult bonobos do not behave prosocially in food-related paradigms, which can be explained by the competitive nature around the highly preferred food items, and which corresponds to the food-related behaviour of bonobos in the wild.

Liselotte Neven

• 30/03/2022
  
• 10 a.m.
• Venue: Campus Drie Eiken, O.01
• ​Online PhD defence
• Supervisors: Karolien De Wael & Sabine Van Doorslaer
• Department of Chemistry

Abstract

Phenolic compounds can be found everywhere in our daily lives but exhibit high toxicity, low (bio)degradability and hormone-disrupting effects when they are released in the environment. It is for this reason imperative to develop detection strategies for these pollutants. A promising approach involves the use of a photoelectrochemical (PEC) sensor. In this sensor, a photosensitiser (PS) type II, which generates 1O2 under illumination, is used to oxidise phenolic compounds present in the sample. The oxidised phenols are reduced at the electrode surface leading to the generation of an electrocatalytic redox cycle.

In this thesis, an in-depth understanding, through the identification of the reactive oxygen species (ROS) in the PEC sensing mechanism, is obtained. The detection strategy is optimised by choosing the PS with the highest 1O2 production and by optimising the detection parameters so that the PEC sensor can be successfully applied for the detection of phenols in industrial samples.

First, it was determined that the use of highly fluorinated zinc phthalocyanine derivatives, F52PcZn and F64PcZn, as photocatalysts was optimal for the sensing of phenol due to their high 1O2 production and improved single-site isolation. However, next to 1O2, it was shown that the ROS O2•- and H2O2 were also generated in the PEC sensor. Their contribution to the photocurrent response was studied by rotating disk electrode measurements in function of the pH and applied potential. After this, the PEC detection strategy was optimised in terms of pH and applied potential for the detection of doxycycline, cefadroxil, and phenol. It was found that the use of alkaline pH-levels led to nmol L-1-level detection limits. The combination with square wave voltammetry (SWV) was, also, proposed to allow the quantification and identification of phenolic compounds in a specific sample. At last, the developed PEC and SWV sensors were applied for the measurement of phenolic compounds in industrial water samples. The PEC sensor could follow the decrease of the phenolic concentration throughout the wastewater treatment process while the SWV sensor provided the electrochemical fingerprints of these samples. The thesis concluded that the use of the PEC sensor was advantageous in the measurement of lower concentrated phenolic samples due to its high sensitivity and fast measurement time in comparison to commercial test kits.

Michiel Van Tendeloo

• 28/03/2022
  
• 4 p.m.
• Venue: Stadscampus, Promotiezaal Klooster van de Grauwzusters, gebouw S, Lange Sint-Annastraat 7, 2000 Antwerpen
• ​Online PhD defence
• Supervisor: Siegfried Vlaeminck
• Department of Bioscience Engineering

Abstract

Adequate removal of pollutants from sewage is important to protect the environment and public health. Today, sewage treatment plants are operational in many parts of the world, and although the used technologies are effective in removing pollutants from wastewater, they are energy- and resource-intensive. Reshaping sewage treatment into a two-stage system, with separated organic carbon and nitrogen removal, facilitates the transformation towards energy-positive sewage treatment. This thesis will focus on resource-efficient nitrogen removal from sewage via partial nitritation/anammox (PN/A), with reduced organic carbon and oxygen consumption compared to conventional techniques.

PN/A relies on the teamwork between two microbial groups to convert ammonium into nitrogen gas. Several other groups of microbes however can proliferate in the sludge, competing for substrate with the key players, lowering the nitrogen removal efficiency and increasing the energy demand. To obtain the desired microbial community, control tools should be applied to selectively promote the desired microbes while suppressing the unwanted competitors. In this thesis, multiple control tools were studied to establish a workable framework for successful implementation of PN/A in the main stream of a sewage treatment plant. These tools can be divided into three categories: i) kinetic tools, regulating substrate availability (e.g., oxygen availability control and residual ammonium concentration), ii) physical tools, revolving around sludge retention and selection (e.g., sludge age control and sludge aggregation form), and iii) chemical tools, exposing the sludge to stress conditions for which the unwanted microbes are vulnerable (e.g., sludge treatments with a single stressor such as free ammonia).

The first research chapter focussed on oxygen availability control and single-stressor sludge treatments. The following two chapters covered the development of a novel multi-stressor concept combining substrate starvation and exposure to sulphide and free ammonia. In the final research chapter, the previously obtained knowledge was combined into a demonstration study on pilot-scale.

The combination of these control tools was found effective in achieving nitrogen removal via PN/A, both on lab- and pilot-scale. Consequently, the obtained results in this thesis can catalyse the implementation of mainstream PN/A by providing a toolbox with multiple control tools and clever reactor design, thus advancing the concept of energy neutrality and resource efficiency in sewage treatment plants.

Tim Van Everbroeck 

• 28/03/2022
  
• 2 p.m.
• Venue: Campus Drie Eiken, R1
• ​Online PhD defence
• Supervisor: Pegie Cool
• Department of Chemistry

Abstract

One of the big problems with gasoline-powered cars is that the exhaust gas contains pollutants which are detrimental for human health and the environment. For this reason cars are equipped with a three-way catalytic converter which has the function to convert carbon monoxide (CO), hydrocarbons and nitrogen oxides (NOx) to molecules that are not harmful for human health. The materials that are used to do this are the platinum group metals (PGMs), platinum, palladium and rhodium, which are very rare, expensive, critical raw materials. The increasing demand for new cars and progressively stricter emission standards drives the price even higher. So, there is a need to reduce the amount of PGMs in three-way catalytic converters and replace them by materials that are more common and cheap. A good candidate to do this is copper oxide (CuO) because it is common and shows some catalytic activity. While it is not as active as the PGMs it can be used in much larger quantities for compensation. To keep the CuO particles small, which is more efficient, they need to be supported on a material with a large surface area. Furthermore, the catalytic activity can be altered by interactions with other materials. For this thesis CuO is deposited on different support materials such as alumina, titania and ceria. The catalytic performance of the final materials are evaluated against its characteristics. Another part of the research focuses on the precipitation of copper with other elements is to obtain intimate mixtures of metal oxides and mixed metal oxides. The spinel-type materials are evaluated against pure CuO to find synergies between the different elements. A final study makes use of layered double hydroxides to obtain mixtures of metal oxides and mixed metal oxides. Again, we evaluate how certain properties and elements in the composition affect the catalytic performance. At the end of the thesis the conclusions are presented and the developed materials are compared to a commercial three-way catalyst. Furthermore, the limitations of this work are discussed and an outlook on the future of the automotive industry is presented and how the developed materials can play a role in this.

Pedro Heleno Isolani

• 25/03/2022
  
• 3 p.m.
• Venue: Stadscampus, Promotiezaal Klooster van de Grauwzusters, Gebouw S, Lange Sint-Annastraat 7, 2000 Antwerpen
• ​Online PhD defence
• Supervisors: Steven Latré, Johann M. Marquez-Barja & Lisandro Z. Granville
• Department of Computer Science

Abstract

Over the years, wireless technologies have enabled innovation in the industry and leveraged social and economic changes. With the so-called Industry 4.0, future digital industries are moving towards the distributed organization of production, with connected goods, low energy processes, collaborative robots, and integrated manufacturing and logistics. 5G is expected to guarantee stringent Quality of Service (QoS) delivery and for that to happen, the proper employment of wireless Medium Access Control (MAC) protocols is essential for efficient and reliable wireless communication. In indoor scenarios, given its low cost and easy deployment, IEEE 802.11 networks are still the default access choice. Among such stringent and heterogeneous requirements, network slicing is envisioned as the architecture to carve out multiple virtual networks with significantly different performance, security, and traffic isolation characteristics from a common physical infrastructure. However, deciding how to efficiently allocate, control, and manage users and slices remains challenging. Besides, although Software-Defined Networking (SDN) has enabled new levels of innovation and automation for the creation of Resource Allocation (RA) mechanisms, latency-related metrics are often neglected. In this thesis, we analyze the evolution and programmability of wireless MAC protocols and present the questions that should be answered to determine the MAC programmability needed. With our lessons learned, we identify the challenges and opportunities for future-proof MAC designs. Next, we propose an airtime-based slice orchestration approach and RA modeling for IEEE 802.11 Radio Access Network (RAN). By orchestrating airtime portions at the IEEE 802.11 upper MAC layer, we show how our network slicing algorithm can enhance the QoS at the RAN. We present our airtime-based RA modeling for network slicing in IEEE 802.11 RAN and the limitations of performing such optimizations at runtime. To support 5G Mission-Critical Applications (MCAs), we propose a delay-aware approach for MAC management via airtime-based network slicing and traffic shaping, as well as user association using Multi-Criteria Decision Analysis (MCDA). Through experimentation in a real-world testbed, our approach maintains the queueing delay requirements of 5ms under varying traffic demands and for most of the experiments run. Last, we provide a complete Software-Defined-RAN interactive management approach using In-band Network Telemetry (INT) in IEEE 802.11 networks. With INT, the end-to-end flow dynamics, state of the wireless links, and per-hop reliability can be fully assessed.

Danh Bui-Thi

• 17/03/2022
  
• 4 p.m.
• ​Online PhD defence
• Supervisors: Kris Laukens & Pieter Meysman
• Department of Computer Science

Abstract

Biological data mining has been an active research area in bioinformatics in recent years. It is expected to unlock a new stage of biomedical research by discovering knowledge from the huge amount of available biological data using computational methods. This knowledge will generate novel insights into the mechanisms of biological systems. Furthermore, it will support the design of new drugs and development of improved solutions for informed clinical decision making.

In this dissertation, we present machine learning techniques for mining interesting patterns and useful knowledge from biological data for several case studies. More specifically, the dissertation elaborates on the following three problems: mining unexpected patterns from transaction data, building associative classifiers based on association rule mining and identifying compound-protein interactions using deep neural networks. The first problem focuses on finding unexpected patterns from data. These patterns identify a failing in prior knowledge or may suggest an aspect of data that deserves further investigation. We propose a novel approach based on association rule mining along with a clustering algorithm to discover the unexpected patterns. The second problem concerns mining reliable patterns, constructing an interpretable classification model which can be understood. Interpretability of machine learning models is critical in several domains with significant social or financial impact such as healthcare, disease diagnosis. The proposed classification model is a rule list, making a single prediction based on multiple rules. We built the model using association rule mining and multi-objective optimization. The last problem we investigated concerns the problem of compound-protein interaction prediction. Identifying interactions between compounds and proteins is an essential task in drug discovery and development. Such prediction tools can be used to screen compound libraries for given protein targets to achieve desired effects or in testing given compounds against possible off-target proteins to avoid undesired effects. To tackle the problem, we developed a novel approach combining a graph convolutional network and a one-dimensional convolutional neural network. These neural networks encode the data objects, i.e. the compounds or proteins, into intermediate representations which are then used to predict the interaction. We also applied an explanation technique to visualize the contributions of the protein regions on the prediction outcome.

We conclude with an overview of our main contributions as well as a discussion of potential future actions that can be taken to improve our proposed methods.

Yannick Gullentops

• 16/03/2022
  
• 2 p.m.
• Venue: Campus Middelheim, G.010
• ​Online PhD defence​
• Supervisor: Sonja Hohloch
• Department of Mathematics
  

Abstract

Hamiltonian systems are dynamical systems which have at least one conservation law. These systems are of particular interest because they allow us to use geometric tools to obtain dynamical results. This thesis focuses on two distinct types of Hamiltonian systems: proper S1−systems and Hamiltonian PDEs.

Proper S1−systems are Hamiltonian systems on four dimensional manifolds, where we have two conservation laws one of which is a proper map inducing an S1−action. The presence of the S1−action allows us to link the minimal period (dynamical feature) of that S1−action to the local shape of hyperbolic fibers (topological feature). This allows us to establish a one-to-one correspondence between the topology of hyperbolic fibers and a graph theoretical construction, called a generalized bouquet. After the theoretical classification of hyperbolic fibers we focus on explicit examples. We study the bifurcation behaviour of a family of proper S1−systems and discuss what happens locally around hyperbolic fibers.

For the investigation of Hamiltonian PDEs, we start with a ‘triholomorphic’ Dirac-type equation, called the Cauchy-Riemann-Fueter equation, on a so-called hyperkähler manifold that can be transformed into a Hamiltonian PDE. Then, we discuss scale manifolds as preferred underlying function space for the study this equation. Finally, we describe the problems concerning convergence behaviour.

Lies Teunen

• 15/03/2022
  
• 4 p.m.
• Venue: Campus Drie Eiken, Q.002
• ​Registration required via this link
• ​Online PhD defence
• Supervisors: Lieven Bervoets & Ronny Blust
• Department of Biology
• Wearing a face mask is mandatory during the defence
• You must present a valid CST and ID to attend the defence

Abstract

Aquatic chemical pollution, mainly of anthropogenic origin, is a global issue. A specific group of persistent pollutants with distinct hydrophobic/lipophilic characteristics tend to biomagnify (i.e. reaching high concentrations in higher trophic levels) and exhibit low detection rates in water samples. Therefore, within the Water Framework Directive, Environmental Quality Standards were derived for 11 priority compounds and their derivatives. These are to be measured in biota specifically (EQS_biota), in order to assess the risk of secondary poisoning of top predators (including humans) and include hexachlorobenzene (HCB), hexachlorobutadiene (HCBD), mercury (Hg), brominated diphenyl ethers (PBDE), perfluorooctane sulfonate (PFOS), hexabromocyclododecane (HBCD), dicofol, dioxins and dioxin-like compounds, heptachlor and heptachlor epoxide, fluoranthene and benzo(a)pyrene.

In this PhD, the relevance of the current EQS_biota with regards to ecological quality of aquatic freshwater and brackish ecosystems was evaluated. The study was built around the EQSbiota monitoring in Flanders on 44 sampling locations. The abovementioned compounds and PCBs were analysed in indigenous European perch (Perca fluviatilis) and eel (Anguilla anguilla) in its juvenile ‘yellow eel’ stage. However, benzo(a)pyrene and fluoranthene were analysed in transplanted caged bivalves because of their fast metabolization in fish. Accumulated concentrations were checked for compliance against the current standards and compared to passive sampler data. The main motivation for this PhD was that, despite frequent exceedances of the some EQS_biota (often with a large factor), no apparent effects on the aquatic ecosystem (including secondary poisoning) are reported in literature.

Next, two case studies were included on accumulation patterns of Hg and PFAS. In contrast to the other lipophilic compounds, they show a high affinity for proteins. For Hg, accumulation in muscle and liver tissue was compared between the two fish species and linked to size (i.e. proxy for age). PFAS profiles were compared between mussels and fish, allowing for the interpretation of trophic magnification. Furthermore, the effects of environmental (water and sediment) concentrations and abiotic characteristics on bioaccumulated concentrations were investigated. Eventually, the ecological relevance of the EQS_biota for the ecological quality, based on the macro-invertebrate community was studied. The human health risk was assessed for all compounds as well.

Finally, results for exceedances of the standards, human health risk and ecological relevance were combined to interpret the overall relevance and protection level of the current EQS_biota. These results might serve as an important first indication for the need to revise or fine-tune the standards for specific compounds.

Stephen Pauwels

• 25/02/2022
  
• 3 p.m.
• Venue: Campus Middelheim, A.143
• ​Online PhD defence
• Supervisor: Toon Calders
• Department of Computer Science

Abstract

In our highly automated world, where manufacturing processes are often performed by autonomous robots, the need for describing and checking these ongoing processes is high. The field of Business Process Monitoring tries to monitor running executions of tasks and tries to indicate possible bottlenecks or errors. In this thesis, we look at predicting next events, detecting anomalies, and detecting concept drift in an event log.

We introduce the use of Dynamic Bayesian Networks as a model for monitoring business processes. We extend this model to describe the typical behavior found in these event logs. We show the use of our model for predicting the next events, given a partially completed case in an event log. We then expand the model further to detect anomalies or deviations from the original process. One of the advantages of our Bayesian Network-based approach is its explainability. The impact of taking measures to deal with a predicted anomaly might be high, therefore people need to trust the model before they will take action to correct the running process. We show how we can use this explainability to explain drifts and differences in event logs after the model has identified them. ​ ​

Predictive Monitoring techniques are proposed in a static situation, where the data used for training and testing is fixed. This leads to using models that are no longer up-to-date for predicting new activities. We propose various strategies for updating existing models while new events arrive. We do this by first using these events for testing our model, and then adding them to the training set which is used to update the model. To allow for quick training times we propose the use of a simple neural network, consisting mainly of a single dense layer. We show that this network performs on par with existing methods with higher complexity. ​ ​

We also look at how evaluations for next activity prediction are performed in the literature. We show some dangerous practices where results are being copied from papers, resulting in comparisons that are based on different datasets. As we show that small changes to the data can lead to different results. We propose some basic guidelines to improve the reproducibility of next activity prediction methods.

Diana Lorena Giraldo Franco

• 18/02/2022
  
• 3 p.m.
• ​Online PhD defence
• Supervisors: Jan Sijbers, Ben Jeurissen & Eduardo Romero
• Department of Physics

Karthik Gadde

• 02/02/2022
  
• 3.30 p.m.
• ​Online PhD defence
• Supervisors: Bert Maes & Kourosch Abbaspour Tehrani
• Department of Chemistry

Abstract

A major issue within chemistry is the use of expensive, non-renewable and toxic reagents/reactants, catalysts or solvents to carry out organic reactions. The replacement of these by inexpensive, abundant and benign alternatives is an ongoing challenge for the chemical community. The research of this doctoral thesis is focused on the development of novel sustainable methodologies for the synthesis of α-substituted homoallylamines via 2-aza-Cope rearrangement and subsequent functionalization of its alkene moiety via visible light photocatalysis.

α-Substituted homoallylamines are valuable synthetic building blocks and precursors for the synthesis of a variety of nitrogen-containing heterocycles, natural products and pharmaceutical compounds in organic synthesis and medicinal chemistry. In this doctoral thesis, a metal-free or base-metal-catalyzed 2-aza-Cope rearrangement strategy has been developed for the synthesis of α-substituted homoallylamines by using easily accessible aldehydes and 1,1-diphenylhomoallylamines.

Alkenes are one of the most fundamental functional groups in organic chemistry and are often derived from simple chemical feedstock. In the past few decades, the functionalization of alkenes has caught tremendous attention, especially strategies to achieve functionalization in an anti-Markovnikov manner were the focus of fundamental research. The installation of sulfonyl (R1SO2-) and sulfenyl (R2S−) moieties specifically is of application interest as they appear in natural products, bioactive molecules, and pharmaceuticals. Besides, they are attractive functionalities in organic synthesis as they are easily transformed into other functional groups. Direct functionalization methods of alkenes, which allow to introduce two different groups in a single reaction step such as the thiosulfonylation are therefore highly desired. In this doctoral thesis, a metal-free methodology for the vicinal thiosulfonylation of unactivated alkenes has been developed by using thiosulfonates as a reactant and the acridinium salt 9-mesityl-10-methylacridinium perchlorate as a photo-organocatalyst with visible-light irradiation. To illustrate the synthetic potential, the method was applied for homoallylamines and late stage functionalization of olefins in active pharmaceutical ingredients.

Dante Horemans

• 12/01/2022
  
• 10 a.m.
• ​Online PhD defence
• Supervisors: Patrick Meire & Tom Cox
• Department of Biology

Abstract

Estuaries often show regions in which suspended particulate matter (SPM) and/or phytoplankton accumulate. Predicting the location of these regions and the corresponding magnitude of the SPM and phytoplankton concentration is of great importance for managing the estuary; it may prevent the system from evolving towards a (hyper-)turbid state, a condition in which phytoplankton growth, which forms the basis of the food chain, is very limited.

To predict the location and magnitude corresponding to accumulation of SPM and phytoplankton, we have to understand the interconnection between SPM and phytoplankton concentration. On the one hand, SPM is often a limiting factor for estuarine phytoplankton growth by deteriorating the light climate and thereby limiting photosynthesis. On the other hand, various authors showed that phytoplankton may determine the SPM concentration by, for example, the excretion of sticky substances. These substances may impact flocculation and thereby the settling velocity and dynamics of SPM flocs. Additionally, they may stabilize the bed and thus change the erosion properties, which also impacts the SPM concentration. While most literature focuses on the small-scale impact of biological flocculants, its influence on the SPM profiles on the large spatial- and temporal scale is still largely unknown.

In this thesis, we study the interconnection of SPM and phytoplankton on the large temporal- and spatial scale, applied to a turbid, tide-dominated, and nutrient-rich estuary, being the Scheldt estuary. To this end, we combine a model approach and analysis of multi-annual observations covering the entire domain of the Scheldt estuary.

First, we extend a hydro-sediment transport model by a flocculation model and showed that flocculation might significantly impact the estuary-scale SPM distribution. Next, we apply the model to show that the impact of biotically-induced flocculation and erosion on the estuary-scale seasonality in the SPM distribution is rather limited. Seasonality in freshwater discharge might explain the observed seasonality in SPM. Next, we construct a conceptual model to show that temporal variability in light climate (cf. SPM) may drastically reduce the time-averaged PP and exponential phytoplankton growth and delay the onset of a spring bloom by ~ weeks. Last, we combine our previous results to show that a multi-annual change in mortality rate, and not in the SPM alone, may explain the observed multi-annual evolution in phytoplankton blooms in spring in the Scheldt estuary. Although we apply our modeling framework to the Scheldt estuary, it may also be applied to other turbid, tide-dominated, and nutrient-rich estuaries.

Charlotte Segers

• 10/01/2022
  
• 5.15 p.m.
• ​Online PhD defence
• Supervisor: Sarah Lebeer
• Department of Bioscience Engineering

Abstract

Pelvic cancers are amongst the most frequently diagnosed cancers worldwide. Patients undergoing pelvic radiotherapy often report a wide diversity of complications, which reduce patients’ quality of life. Currently, there are no effective therapies available to mitigate these injuries, which is partly due to a lack of insight into the events causing intestinal mucositis and dysbiosis. Apart from traditional pharmacological compounds, there is a need for research on adjuvant therapies.

This PhD explored the potential of Limnospira indica PCC8005 as innovative treatment strategy for pelvic irradiation-induced mucositis and dysbiosis, in comparison to probiotic Lacticaseibacillus rhamnosus GG ATCC53103.

To achieve this, first, an in vivo irradiation-gut-microbiome model was developed. Herein, the structural and functional impact of pelvic irradiation on the intestinal ecology of healthy mice was characterized so that side effects encountered by patients following pelvic radiotherapy were mimicked. Pelvic irradiation evoked structural and functional changes in the intestine, which secondarily resulted in a microbiome shift. Members of Ruminococcacceae, Lachnospiraceae and Porphyromonodaceae were differentially impacted and identified as biomarkers for pelvic irradiation.

Subsequently, unprocessed, fresh biomass of L. indica PCC8005 or L. rhamnosus GG were investigated for their effect on the gut microbiome of healthy mice. Both appeared to transiently shift the microbial community, characterized by a higher relative abundance of butyrate-producing members of the Lachnospiraceae and Porphyromonadaceae families, respectively. This could indicate a potential for both supplements to sustain or restore the intestinal ecology following pelvic irradiation-induced intestinal mucositis.

Therefore, a randomized, placebo-controlled preclinical trial was set up in which daily supplementation with fresh L. indica PCC8005 and L. rhamnosus GG before and after pelvic irradiation was investigated in our in vivo irradiation-gut-microbiome model. They were evaluated for their radioprotective effects on the intestinal ecosystem. Although both could not confer barrier protection, L. rhamnosus GG was attributed some anti-inflammatory capacities by partly reducing mucosal myeloperoxidase levels. In addition, L. rhamnosus GG appeared more effective in preventing pelvic irradiation-induced dysbiosis when compared to L. indica PCC8005.

In conclusion, this PhD contributed to a better understanding of the pathogenesis of pelvic irradiation-induced mucositis and dysbiosis. Furthermore, we obtained insights in the capacity of Lacticaseibacillus rhamnosus GG ATCC53103, and to a lesser extent Limnospira indica PCC8005, to grant radioprotection. This work thus provides evidence to further explore the potential of both food supplements as a mitigator counteracting pelvic irradiation-induced intestinal mucositis and resultant dysbiosis.