91�Թ���

Ivan Sudakov

• 10/11/2026
  
• 4.30 p.m.
• Venue: Campus Drie Eiken, R3
• Supervisors: Sabine Van Doorslaer & Etienne Goovaerts
• Department of Physics

Abstract

Organic solar cells (OSCs) based on the bulk-heterojunction (BHJ) architecture represent a promising technology that combines high power conversion efficiency with low weight, mechanical flexibility, and solution processability. Non-fullerene acceptors (NFAs) have transformed organic photovoltaics through their highly tunable optical and electronic properties, enabling single-junction power conversion efficiencies beyond 20 %. Yet, insufficient long-term stability remains a major obstacle to the successful commercialization of OSCs. Understanding the degradation mechanisms is therefore essential for the development of more stable materials and devices. 

The first part of this thesis investigates the photodegradation kinetics of a model blend comprising the donor polymer poly(3-hexylthiophene) (P3HT) and the rhodanine-flanked small-molecule acceptor FBR. To investigate these degradation processes, absorption and photoluminescence spectroscopy were combined with electron paramagnetic resonance techniques, including pulsed EPR, optically detected magnetic resonance (ODMR), and spin trapping. The results show that neither P3HT nor FBR behaves in the blend as it does in its single-component film. In P3HT, photoinduced electron transfer to molecular oxygen produces partially reversible charge-transfer complexes, while its irreversible degradation is attributed mainly to radical-mediated photooxidation involving superoxide. By contrast, FBR acts as an efficient triplet sensitizer, generating singlet oxygen and undergoing rapid photobleaching in air. In the blend, charge separation at the donor–acceptor interface suppresses triplet formation and initially protects FBR, but simultaneously promotes superoxide formation and accelerates the degradation of P3HT. As degradation proceeds and charge separation becomes less efficient, triplet excitons repopulate in FBR and generate ¹O₂, which attacks both blend components. 

The second part focuses on triplet excitons in single-walled carbon nanotubes (SWCNTs). The optical and electronic properties of SWCNTs depend critically on their chiral structure, making them promising materials for a wide range of photovoltaic and optoelectronic applications. Although singlet excitons have been studied extensively, considerably less is known about optically dark triplet excitons. ODMR was used to characterize and distinguish these excitons in chirality-purified (6,5) and (7,5) SWCNTs. Analysis and simulation of the spectra revealed three triplet species. The first has axial symmetry and a zero-field splitting that depends on nanotube diameter, consistent with a spin density distributed around the nanotube circumference. The second is assigned to triplet excitons localized near structural defects, while the third exhibits unresolved zero-field splitting and is attributed to weakly interacting electron–hole pairs. These results demonstrate the ability of ODMR to distinguish optically dark excited states and provide information about their symmetry, confinement, and microscopic origin. ​ ​

Jindong Zhang

• 06/11/2026
  
• 4 p.m.
• Venue: Campus Drie Eiken, D.O.01
• ​​
• Supervisors: Raoul Van Damme & Kristina Wanieck
• Department of Biology

Abstract

Over billions of years, evolution has produced an immense diversity of biological structures, materials, mechanisms, and behaviours. Biomimetics draws on this diversity by analysing biological principles and transferring them into technical applications. Yet a comprehensive understanding of how biomimetic research accesses biological knowledge was still lacking. This thesis investigates which organisms biomimetic research draws on, who conducts such research, and which tools support it. It then developed methods to make the discovery and selection of biological models more systematic.

The first part of the thesis diagnosed the field along three dimensions – biological, human, and methodological dimensions. An analysis of more than 74,000 biomimetic publications identified 31,776 biological models. The field relies mainly on animals and on a limited set of repeatedly used organisms; fewer than 23% of models are specified at species level; more than 90% of publications with an identifiable model refer to a single organism; and new taxa are added more slowly than the field grows. An analysis of 436,010 author-affiliation records across biomimetic publications showed that technical researchers outnumber biological researchers by about two to one.

Interdisciplinary teams have grown more than twenty-fold since 2004 and are associated with higher citation performance; a broader, taxonomically more precise use of biological models is associated with biology expertise specifically. A total of 104 tools were identified, 24 of which were fully functional. Comparison of these 24 biomimetic tools against ten criteria revealed that few provide dedicated support for systematic biology-push discovery of unfamiliar biological models or for the explicit comparison of candidate models.

The second part of the thesis developed methodologies to address these gaps. A data-driven, phylogeny-aware framework treats biological model selection as an explicit comparison across data sufficiency, prior biomimetic use, phylogenetic characteristics, and project-specific constraints. In a microplastic-filtration case study with 35 suspension-feeding taxa, the developed framework distinguished candidates that remained competitive under different priorities from those whose suitability depended on context. A complementary discovery framework organises biological literature into functional atlases that show which taxa use which mechanisms to perform a given function, how these occurrences are distributed across phylogenetic lineages, and whether biomimetic research has already engaged them. In two pilot domains, filtration (257 taxa across 21 phyla) and attachment (838 taxa across 24 phyla), only 6% and 15% of taxa, respectively, had been engaged by biomimetic research. Together, this thesis provides an empirical and methodological basis for engaging biological diversity more strategically.

Joke Van Malderen

• 27/10/2026
  
• 5 p.m.
• Venue: Campus Drie Eiken, O.05
• Supervisors: Sarah Lebeer, Irina Spacova, Olivier Vanderveken & An Boudewyns
• Department of Bioscience Engineering

Abstract

Upper respiratory tract (URT) infections, including otitis media (OM), remain among the most common childhood infections and a major driver of antibiotic use. Microbiome research has shifted the understanding of these diseases from pathogen-centered infections towards disturbances of complex microbial ecosystems, creating opportunities for microbiome-based therapeutics. However, translation of probiotics and live biotherapeutic products (LBPs) into respiratory applications remains limited by an incomplete understanding of their ecological behavior and mechanisms of action. This thesis aimed to contribute to the rational development of mechanism-based respiratory microbiome therapeutics. The thesis addressed three questions: (1) Can orally administered probiotics reach the URT and influence its microbial ecosystem? (2) How does the URT microbiome develop during early life in children with cleft palate, a population at high risk for OM, and can this trajectory be influenced by probiotic supplementation? and (3) Which molecular mechanisms underlie the probiotic potential of respiratory commensals such as Streptococcus salivarius?

First, in a randomized controlled study of 37 children with persistent OM with effusion, LGG-like bacteria were detected in the URT in 4/19 treated children compared with 0/18 controls, whereas BB-12 was not detected. Thus, orally administered LGG can reach the URT, but transfer occurred only in a subset of children. Overall microbiome composition was primarily associated with anatomical site and host-related factors, with limited effects of probiotic treatment.

Second, a longitudinal randomized controlled study in children with cleft palate showed that URT microbiome development was primarily shaped by anatomical niche, inter-individual variation and temporal dynamics rather than probiotic supplementation. LGG-like bacteria were detected sporadically, consistent with transient presence rather than stable colonization.

Third, comparative genome mining identified five class II bacteriocin biosynthetic gene clusters in two nasopharyngeal S. salivarius isolates. Selected loci showed pathogen-specific inhibitory profiles, while cell-free supernatants confirmed the contribution of secreted antimicrobial compounds. A heterologous Escherichia coli expression platform was established for future bacteriocin characterization.

Collectively, these findings demonstrate that effective respiratory microbiome therapeutics require both ecological fitness and mechanistic understanding. They support a shift towards niche-adapted respiratory commensals, locally optimized delivery strategies and mechanism-based selection of next-generation LBPs.

Mona Huyzentruyt

• 13/10/2026
  
• 4 p.m.
• Venue: Campus Drie Eiken, O.08
• ​
• Supervisor: Stijn Temmerman
• Department of Biology

Abstract

Tidal marshes are highly effective at mitigating climate change through soil organic carbon (SOC) sequestration, driven by vegetation productivity and sediment deposition. However, carbon sequestration rates vary widely between and within marsh systems. The patterns in this variability remain largely unknow. Additionally, tidal marshes are highly influenced by sea level rise, which generally increases sequestration by boosting vegetation growth and sediment deposition. In microtidal marshes (tidal range <2m) however, marshes cannot keep up with sea level rise and often show signs of degradation through pond formation. Whether these degrading marshes can maintain their carbon sequestration function remains unknown. This thesis investigates the variability of tidal marsh carbon sequestration under sea level rise, focusing on: 1) drivers of global variation in SOC content and accumulation rates, and 2) spatial heterogeneity of SOC dynamics and stability in degrading marsh systems. The first objective was investigated through a machine learning analysis of global SOC data. The second objective was investigated using field measurements in degrading marshes, geospatial analysis, and lab analyses for SOC stability assessments.

The global analysis showed that tidal range is main driver of SOC content, with macrotidal marshes (> 4m) typically having low SOC content, while high SOC marshes only occur in microtidal (< 2m) environments. The limited sediment deposition (due to low flow velocities) in microtidal marshes drives the reliance on organic matter accumulation, leading to high SOC contents. The analysis additionally confirmed that sea level rise rate positively predicts SOC accumulation, although there is no clear positive relationship between SOC accumulation and sea level rise.

In degrading microtidal marshes, remaining vegetated zones showed higher SOC accumulation than in less-degraded systems, likely due to increased inundation and redeposition of eroded SOC. However, once a threshold vegetation cover is passed, marshes become net carbon sources, as carbon losses from ponding outpace gains. Ponds form first in lower-elevation basins, while marsh levees, next to tidal channels, are more resistant to degradation. Marsh levees showed SOC accumulation rates five times higher than basins, due to greater sediment deposition, vegetation productivity, and auto-compaction. Including the difference in SOC accumulation rates between levees and basins increases the threshold vegetation cover at which marshes become net carbon sources.

Finally, our findings reveal a trade-off between SOC quantity and stability, with microtidal systems typically containing large amounts of unstable carbon. This highlights the vulnerability and the need for conservation of microtidal carbon stocks.

German David Goez Sanchez

• 12/10/2026
  
• 4 p.m.
• Venue: Stads91�Թ���, Hof van Liere - Prentenkabinet
• ​
• Supervisors: Johann Marquez-Barja & Miguel Camelo Botero
• Department of Computer Science

Abstract

The transition from 5G and 5G-Advanced toward 6G is redefining wireless communication systems as AI-native infrastructures, where intelligence is embedded throughout the communication architecture and pushed closer to the network edge. In this context, the PHY layer is a key domain for integrating data-driven models under strict constraints on latency, memory, energy consumption, and computational capacity.

However, a persistent gap remains between high-performance DL models developed in unconstrained computing environments and their practical deployment on heterogeneous edge devices. This thesis addresses this gap through a unified framework integrating efficient model design for the radio domain, complexity reduction, hardware-aware optimization, deployment, and experimental validation on computationally constrained RAN-oriented platforms. 

The research follows a closed methodological loop, progressing from efficient DL architectures through model compression, MTL, layer-wise quantization, and automated architecture search, culminating in deployment and experimental validation on embedded processors, accelerators, and programmable logic devices.

First, the thesis proposes a lightweight channel-gain prediction pipeline combining CSI compression, compression-aware training, and recurrent prediction enhanced with channel attention mechanisms. The approach reduces transport-layer overhead and computational complexity while maintaining bounded degradation in prediction accuracy.

Second, a methodology is developed for spectrum-based TC from I/Q samples using compressed multi-task architectures. Shared representations across related tasks reduce redundancy, memory footprint, inference time, and energy consumption while preserving accuracy compatible with real-time edge operation.

Third, a methodology for layer-wise quantization of DNNs for radio receivers uses layer-level sensitivity analysis to determine mixed-precision configurations that balance computational efficiency and predictive performance. Fourth, a resource-aware neural architecture search framework jointly optimizes convolutional architectures and adaptive quantization under hardware-relevant constraints. It combines PSO, surrogate modeling, and reduced-data evaluation to efficiently explore large design spaces and identify architectures that balance predictive performance and computational cost.

Finally, the resulting models are deployed and experimentally validated on computationally constrained devices, including accelerated GPU environments and FPGA-based platforms. The experiments demonstrate practical gains in latency, power consumption, throughput, and logical resource usage. 

Overall, this thesis provides a systematic framework bridging computationally demanding DL models and deployable AI solutions for edge intelligence, supporting the practical integration of efficient DL techniques into AI-native RAN architectures and future wireless communication systems.

Jari Temmermans

• 12/10/2026
  
• 4 p.m.
• Venue: Campus Middelheim, M.G.010
• ​
• Supervisors: Sarah Lebeer, Wenke Smets, Filip Kiekens & Barbara De Coninck
• Department of Bioscience Engineering

Abstract

As we are approaching 10 billion people by 2050, food production will need to increase by 50%. However, each year, up to 40% of crops are lost due to plant pests and diseases. While chemical plant protection products increase crop productivity, pathogen resistance, environmental and health concerns, sustained consumer demand for residue-free food, and stricter policies worldwide have increased the need for alternatives. Microbial biological control agents (MBCAs) have long been regarded as sustainable alternatives to synthetic agrochemicals, offering targeted suppression of plant pathogens while preserving ecosystem health. However, despite many MBCAs demonstrating strong in vitro efficacy, their in planta performance remains inconsistent. While this challenge is especially pronounced in the phyllosphere, the aerial surfaces of plants, this environment remains significantly less studied than the roots. Understanding why field performance is more challenging, particularly on flowers and leaves, is essential for guiding future research, regulation, and product development. Key barriers to field efficacy include ecological factors, (a)biotic stressors, technical limitations, such as ineffective formulation or application, and structural constraints, including regulation, funding, and stakeholder training. Growers and advisors often mention MBCA inefficacy under practical conditions as a disadvantage, expressing the need for comprehensive guidelines and hands-on training.

We conducted an MBCA isolation campaign on strawberry flowers, employing a citizen science approach, involving 50 growers from Belgium, the Netherlands, Germany, and Estonia. We explored the bacterial composition of strawberry flowers in agricultural contexts across a gradient of cultivation intensification, revealing that more intensive agriculture led to decreased diversity, the loss of core taxa, and greater variability, potentially reducing the plant's resistance against pathogens and creating opportunities for biocontrol. Furthermore, by focusing our MBCA isolation campaign on Lactobacillaceae, a large family of beneficial bacteria that are generally recognised as safe, have a niche in the phyllosphere, and have not yet been commercialised as MBCAs, we aim to accelerate the registration process, a bottleneck limiting the commercialisation of many potential biocontrol agents, especially in Europe. ​ ​

The acquired Lactobacillaceae isolates were screened for antipathogenic activity, adaptation to flowers, and survival in the phyllosphere. Notably, one isolate exhibited potent in vitro inhibitory activity against several plant pathogens, which was retained after formulation via spray-drying. Importantly, when sprayed preharvest on greenhouse-cultivated strawberry plants, this isolate significantly inhibited Botrytis cinerea. Ongoing and future work will focus on scaling up, elucidating the mode of action, and extending in planta efficacy to additional pathogen-cropping systems. ​ ​

Joran Van Turnhout

• 08/10/2026
  
• 11 a.m.
• Venue: Atlas, promotiezaal (Eindhoven University of Technology)
• ​​
• Supervisors: Annemie Bogaerts & Emiel Hensen
• Department of Chemistry

Abstract

Electrifying the chemical industry with renewable energy is a key strategy for reducing anthropogenic CO2 emissions, and non-thermal plasma is an emerging technology well suited to the intermittent nature of renewable energy supply. However, plasma-driven gas conversion is limited by poor product selectivity, particularly toward value-added products. Catalyst integration rarely improves selectivity in practice, and the fundamental mechanisms governing plasma–catalyst interactions remain poorly understood. This dissertation aims to advance that mechanistic understanding through experimental studies of CO₂ hydrogenation in dielectric barrier discharges (DBDs), the most common plasma type used with catalysts due to their low-temperature operation.

The first study examines how surface oxygen vacancies (OVs) on CeO₂ affect plasma-catalytic CO₂ hydrogenation. Contrary to expectation, CO₂ conversion decreases with increasing OV concentration, with a simultaneous reduction in the number and intensity of microdischarges. Density functional theory calculations show that OVs lower the work function of CeO₂, altering discharge behavior; since CO₂ activation proceeds mainly through intense microdischarges, this directly suppresses conversion. Loading CeO₂ with Cu nanoparticles further reduces the conversion. These results indicate that discharge characteristics depend sensitively on catalyst composition and oxidation state, and that gas-phase chemistry dominates over surface catalysis in conventional packed-bed DBD reactors.

To probe catalyst surfaces during operation, a novel in-situ DBD transmission FTIR cell was designed, built primarily from PEEK for compatibility with high-voltage operation and atmospheric-pressure conditions without noble-gas dilution. Applied to CeO₂, Cu/CeO₂, and thermally pre-reduced Cu/CeO₂ under H₂/Ar plasma, the cell revealed plasma-driven OV formation via hydrogen spillover only on the pre-reduced Cu/CeO₂ catalyst, with a dynamic equilibrium between OV formation and healing established rapidly under H₂/CO₂ plasma. This establishes OV concentration as a highly dynamic, condition-dependent catalyst property.

A critical assessment of literature on plasma-catalytic CO2 hydrogenation to methanol, using direct online GC analysis, found methanol selectivities two to four orders of magnitude below published values, a discrepancy attributed to systematic overestimation by the indirect offline quantification methods common in the field; cross-laboratory validation confirmed the low online values. Co-feeding experiments further showed significant plasma-induced methanol decomposition even under mild conditions.

The final chapter identifies four key challenges facing plasma catalysis—catalyst design, exploitation of vibrational excitation, limited plasma–catalyst contact, and the need for rigorous, standardized measurement—and argues that in-situ product removal is essential for improving energy efficiency.

Stijn Van Rompaey

• 01/10/2026
  
• 4.30 p.m.
• Venue: Campus Middelheim, M.A.143
• ​​
• Supervisors: Annemie Bogaerts, Anton Nikiforov & Rino Morent
• Department of Chemistry

Abstract

Methane (CH₄) is a potent greenhouse gas and an important target for emission reduction and valorisation. Converting CH₄ into value-added products such as hydrogen (H₂), ethylene (C₂H₄), acetylene (C₂H₂), and solid carbon offers a promising alternative to conventional methane reforming, which generates CO₂. Methane pyrolysis avoids direct CO₂ formation and requires less reaction enthalpy, but conventional processes lack the flexibility needed to efficiently integrate intermittent renewable electricity. Nanosecond pulsed discharges (NPDs) provide an attractive alternative because they are electricity-driven, highly flexible, and combine intense electron-driven chemistry during short pulses with high gas temperatures in the subsequent afterglow. ​ ​

This thesis provides an experimental and modelling investigation of CH₄ conversion in NPDs, covering plasma physics, chemical kinetics, stable product formation, and solid carbon growth. First, Thomson scattering and Stark broadening of the Hα emission line were compared for electron-density measurements. Both methods provide consistent results shortly after the discharge, while Thomson scattering offers more reliable measurements at later times when incoherent scattering dominates. ​ ​

A complete characterisation of pure CH₄ NPDs subsequently revealed intense, narrow discharges with electron densities of 7.0–10.5 × 10¹⁷ cm⁻³ and pulse energies of 6–15 mJ, followed by afterglows exceeding 2000 K. Combining kinetic modelling with gas chromatography demonstrated that electron-impact reactions initiate CH₄ dissociation and generate reactive intermediates, while the subsequent high gas temperature strongly determines overall conversion and product distribution. CH₄ conversion therefore exhibits the characteristics of pyrolysis, with H₂ and C₂H₂ as the dominant products and near-complete CH₄ dissociation within the plasma region. ​ ​

Significant solid carbon formation was also observed: black carbon formed in the hot plasma core and brown soot accumulated in the cooler plasma periphery. Simulations using conventional combustion-based soot chemistry underpredicted the experimental carbon yield. Ab initio calculations showed that C₂H₂ addition to cationic polycyclic aromatic hydrocarbons can proceed barrierless. Incorporating this plasma-specific effect phenomenologically into the kinetic model substantially improved predictions of soot yield and particle size distribution, highlighting the importance of charged species in plasma soot formation. ​ ​

Finally, tailored pulse sequences were investigated to improve process performance. Burst-mode operation reduced the energy cost from approximately 480 to 400 kJ mol⁻¹ and increased C₂H₂ selectivity from 45% to 54%, without significantly changing CH₄ conversion. Overall, this thesis establishes a comprehensive understanding of the physical and chemical processes governing CH₄ NPDs and demonstrates opportunities to improve product selectivity and energy efficiency through plasma and pulse-pattern engineering.

Matthias Albrechts

• 29/09/2026
  
• 5 p.m.
• Venue : Campus Middelheim, M.A.143
• Supervisor: Annemie Bogaerts
• Department of Chemistry

Abstract

Plasma-based nitrogen fixation (NF) into NOx represents a promising electrified alternative to the Haber-Bosch process, with the potential for flexible, decentralized operation using air and renewable electricity. However, its energy efficiency remains limited. In this thesis, we develop a deeper understanding of the fundamental processes governing plasma-based NOx formation and identify pathways and reactor conditions that can improve its efficiency. First, we review the fundamental mechanisms of plasma-based NF, with particular attention to thermal, non-thermal, and warm plasma regimes and the different pathways through which N2 can be activated. We show that experimentally achieved energy efficiencies remain above industrial targets, motivating the development of predictive models that resolve both plasma chemistry and transport phenomena. We then reassess the potential of vibrational-translational (VT) non-equilibrium to enhance NO formation through the Zeldovich mechanism using a global model with updated reaction kinetics. We demonstrate that previously reported exceptionally low energy costs cannot be achieved when we properly account for reverse reactions and vibrational relaxation. We identify distinct regimes of NO formation, including a pathway in which NO forms under thermal equilibrium but chemical non-equilibrium conditions, which may be particularly relevant for nanosecond-pulsed discharges. ​ ​

Next, we develop a comprehensive multidimensional fluid model for (near-)atmospheric-pressure microwave air plasmas, coupling turbulent flow, electromagnetic heating, thermal chemistry, and species transport. We validate the model against experimental data over a wide range of operating conditions and demonstrate strong predictive capability for temperature, electron density, and NOx production without fitted parameters. We reveal the important roles of convection, diffusion, temperature gradients, pressure, and post-plasma quenching in determining NO formation and destruction.

Finally, we investigate strategies to improve warm plasma-based NF using open-torch and nozzle-enhanced reactor configurations combined with heat recovery and rapid post-plasma quenching. We show that feed-gas preheating, effective mixing, and fast cooling can substantially improve both NOx concentration and energy efficiency. We find that relatively simple open-torch designs can achieve efficient operation at lower energy inputs, whereas nozzle configurations become advantageous at higher energy inputs by enhancing mixing and quenching.

Overall, we establish more realistic energy cost limits for plasma-based NF and identify elevated pressure, heat recovery, and rapid quenching as key elements for improving warm plasma-based NOx production.

Arthur Vienne

• 29/09/2026
  
• 10 a.m.
• Venue: Campus Drie Eiken, Q.002
• ​​
• Supervisor: Sara Vicca
• Department of Bioscience Engineering

Abstract

Rock Weathering as a Climate Solution: Promise and Complexity ​ ​

Limiting global warming in line with the Paris Agreement will require not only cutting emissions, but also actively removing carbon dioxide from the atmosphere at massive scale. Alongside familiar approaches like tree planting, scientists are exploring other strategies that harness naturally occurring chemical reactions to lock away CO₂. This PhD research investigates one such strategy: enhanced weathering, which involves spreading crushed volcanic rock — basalt or dunite — onto farmland. As these minerals dissolve, they react with CO₂ and, in theory, convert it into stable carbon forms stored in the soil or transported to rivers and oceans as dissolved carbon, while also delivering benefits for soil fertility and food production. ​ ​

Early computer models suggested enhanced weathering could remove billions tons of CO₂ annually. This research set out to test those predictions experimentally. Across several pot experiments, the expected build-up of carbon in the soil could not be clearly detected — any increases were too small to distinguish from normal measurement uncertainty. Digging deeper, the work found that other soil processes — including interactions with organic matter, the binding of nutrients (base cations) to soil particles, and the formation of new secondary minerals — can all interfere with the chemical pathway that is supposed to lead to CO₂ removal. These processes also introduce long delays before any carbon benefit shows up, which matters for anyone trying to build a credible business or policy case around the technology. The research also points to an underexplored dimension: enhanced weathering may affect not just inorganic carbon, but also the soil's existing organic carbon stocks, in ways that could work for or against the climate benefit. A newly developed model coupling rock weathering with soil organic matter dynamics reinforced this mechanism. ​ ​

Given these complications in agricultural soils, the research also examined an alternative setting: enhancing weathering in rivers directly, rather than in soil. Increasing pH in rivers by adding rocks could potentially remove between hundred millions to even billion tons of CO₂ per year at a lower estimated financial cost and energy footprint than technologies like direct air CO₂ capture, though environmental risks and practical limits still need careful evaluation. Together, these findings suggest enhanced weathering holds real climate potential, but that realizing it depends on far better understanding of the hidden chemical and biological processes governing carbon removal, and on long-term real-world testing.

Lisa Siciliani

• 25/09/2026
  
• 4 p.m.
• Venue: Campus Groenenborger, U.024
• Supervisors: Sara Bals & Eveline Verleysen
• Department of Physics

Abstract

Nanotechnology is driving innovation in the food sector, making robust regulatory oversight essential. Because nanoscale features confer size-dependent properties requiring evaluation beyond traditional chemical risk assessment through thorough physicochemical characterisation, specifically particle size distribution. In this context, this work applies electron microscopy (EM)-based methods to fill physicochemical data gaps, quantify measurement uncertainties, and evaluate materials in complex matrices and biological systems for two food-relevant case studies in the European Union: iron oxides and hydroxides (food additive E 172) and nanocellulose (a novel food candidate). Characterisation of commercially available E 172 revealed four distinct forms (spheroidal and elongated hematite, elongated goethite, and spheroidal magnetite with maghemite at the surface), with 86–100% of particles under 250 nm for all materials, indicating nanoscale properties must be considered under European Food Safety Authority (EFSA) guidance documents. The results of the characterisation of food additive E 172 can contribute to refining the European Union specifications. Based on the physicochemical properties of E 172 forms, grouping for risk assessment is not advisable without endpoint-specific justification. To address the lack of a standardized validation framework, the work established a robust metrological basis to quantify measurement uncertainties. It evaluated method precision and robustness, demonstrating that because particle sizes often overlap with the 100 nm threshold of the EC Recommended definition of a nanomaterial, integrating measurement uncertainty is critical for correct regulatory classification. An inter-laboratory comparison successfully validated the reproducibility and transferability of these EM protocols. Furthermore, the study developed EM-based protocols to extract and identify these forms in complex food matrices, demonstrating that the extracted particles retain their properties without transformation. For nanocellulose materials, four distinct nanoforms were identified (crystalline nanocellulose, nanofibrillated cellulose (NFC), TEMPO-oxidised NFC, and bacterial nanocellulose). Physicochemical characterisation of these materials confirmed that they meet the definition of engineered nanomaterials under the Novel Food Regulation. This work also established the first intra-laboratory validation study to quantify measurement uncertainties for nanocellulose fibre size by transmission electron microscopy. In addition, the work supports New Approach Methodologies (NAMs)-based risk assessment at three levels: (i) developing dispersion protocols in relevant in vitro media; (ii) investigating nanocellulose stability in an artificial human colon model; and (iii) optimising semi-automated EM analysis to reduce operator-based bias, and support the mechanistic interpretation of cellular uptake in toxicology. This work helps to contribute to more robust nano-safety assessments.

Ihtasham ul Haq

• 22/09/2026
  
• 4 p.m.
• Venue: Campus Groenenborger, G.US.024
• Supervisors: Nick Schryvers, Patrick Cordier & Hosni Idrissi
• Department of Physics

Abstract

This thesis investigates the deformation mechanisms of Earth’s mantle silicates and their significance for large-scale geophysical processes. Although phenomena such as mantle convection, plate tectonics, and fault deformation operate at geological scales, they are fundamentally governed by processes occurring at the grain and atomic scales. Understanding these mechanisms is therefore essential for linking microscale deformation to the macroscopic rheology of Earth’s interior. The present work employs advanced in-situ and ex-situ transmission electron microscopy (TEM) to investigate deformation in two important geological minerals: olivine and talc. Olivine, the most abundant mineral in the Earth's upper mantle, is a key phase controlling mantle rheology. Owing to its limited number of readily activated slip systems, olivine may activate intergranular deformation mechanisms to accommodate general strain. To investigate the grain boundary (GB) deformation mechanisms, PI-95 TEM Pico-indenter holder and the Push-to-Pull (PTP) device (Bruker. Inc) was used. The present study addresses these questions through systematic in-situ TEM nanomechanical experiments on both olivine end members, forsterite (iron-free) and fayalite (iron-bearing). The forsterite results showed that specimens deform exclusively by grain boundary sliding. Detailed microstructural investigation shows direct evidence of stress-induced amorphization in the sliding GBs. Moreover, the digital image correlation (DIC) inside TEM used to quantify the localized strain onset of GB sliding. Also, effect of the iron content on the GB sliding is addressed. The study also demonstrates the investigations on deformation behavior of talc using in-situ TEM nanomechanical testing. Talc is a hydrous magnesium silicate commonly found in mafic and ultramafic rocks and is of particular interest because of its exceptionally weak mechanical behaviour and potential role in weakening talc-bearing faults. In-situ nanomechanical results showed that talc does not exhibit conventional brittle behavior under the investigated loading conditions. Instead, the material displays significant plasticity, even for orientations in which the resolved shear stress on the basal planes is very low. When the basal (001) planes are oriented perpendicular to the tensile axis, however, deformation proceeds through stress-induced amorphization rather than conventional dislocation-mediated plasticity. These observations demonstrate that the deformation behavior of talc is strongly controlled by crystallographic orientation and reveals the coexistence of distinct deformation mechanisms, including dislocation-mediated plasticity and amorphization. The results provide new insights into the nanoscale mechanical behavior of talc and its potential role in controlling the strength, weakening, and deformation of talc-bearing geological materials.

Kenzo Yasaka

• 18/09/2026
  
• 3 p.m.
• Venue: Campus Middelheim, G.010
• Supervisors: Sonja Hohloch & Tom Mestdag
• Department of Mathematics

Abstract

In recent years, there has been increased interest in a coordinate-independent geometric theory for studying solutions of systems of PDEs, known as the theory of exterior differential systems (EDS). This theory originates from the ideas of several mathematicians, such as Pfaff, Darboux, Goursat and, most prominently, Cartan. Here, the PDEs are represented by a collection of differential forms on a manifold, satisfying the condition that this collection is closed under exterior differentiation. The main motivation to study EDS is to compute integral manifolds, which are submanifolds satisfying the property that all differential forms in the collection vanish. Since any system of partial differential equations can be written as an EDS (with an independence condition), finding solutions to a system of PDEs can be given the desired geometric interpretation, namely, finding integral manifolds of its associated EDS. ​&�Բ�����;

In this thesis, we extend the theory of EDS from manifolds (and their tangent bundles), to Lie algebroids. To be more precise, we extend the theory so that instead of limiting ourselves to finding submanifolds such that a collection of differential forms on a manifold vanish (i.e., a form on the tangent bundle), we will find submanifolds so that certain forms on a vector bundle vanish. However, to make sense of the fact that the collection of forms should be differentially closed, we need to assume that this vector bundle has some additional structure, namely, that of a so-called Lie algebroid, where an exterior derivative operator is defined. We also focus on techniques to determine the existence of these special submanifolds for a given EDS.

Robin Smeyers

• 16/09/2026
  
• 4 p.m.
• Venue: Campus Groenenborger, T.105
• ​​
• Supervisors: Lucian Covaci, Milorad Milošević & Lino da Costa Pereira ​
• Department of Physics

Abstract

This thesis investigates how controlled deformations and heterostructures in graphene can be exploited for electronic applications by combining molecular dynamics (MD) simulations with tight-binding (TB) calculations. ​ ​

First, the pseudo-vector potential in strained graphene is analyzed by comparing continuum elasticity and atomistic MD approaches. While continuum models capture only acoustic displacements, MD reveals significant contributions from optical modes, which effectively screen the pseudo-gauge field and substantially reduce predicted pseudo-magnetic fields. ​ ​

Next, indentation-induced doping in graphene on a Cu(111) substrate is studied. TB calculations based on MD-relaxed structures reproduce experimentally observed Fermi energy shifts of up to 0.3~eV and attribute them to resonant states arising from local bond defects. ​ ​

Moving to larger scales, graphene nanobubbles on Pt(111) are modeled by combining MD simulations with scanning tunneling microscopy data. The resulting electronic structure shows quasi-confinement due to both electrostatic and pseudo-magnetic effects, without forming pseudo-Landau levels inside the bubble, while such states emerge outside within a limited spatial range. ​ ​

Finally, moiré engineering in graphene--hBN heterostructures is explored. Relative layer alignment and electrostatic gating enable tuning of primary and secondary Dirac point gaps and the formation of flat bands. We further show that this mechanism underlies the observed negative differential resistance and subsequent single-photon emission in identical experimental setups, providing a platform for future optoelectronic applications.

Björn Loenders

• 08/09/2026
  
• 3.30 p.m.
• Venue: Campus Middelheim, A.143
• Supervisor: Annemie Bogaerts
• Department of Chemistry

Abstract

Our society is strongly dependent on fossil resources for production of energy and chemicals. Yet, their combustion results in the emission of large amounts of CO2, which together with other greenhouse gasses (GHG’s) causes global warming and climate change. Additionally, the reserves of fossil resources are limited and will eventually run out.

A potential solution to this issue is carbon capture and utilization (CCU), which involves collecting CO2 and converting it into new chemicals and fuels. This can be achieved through various reactions, one of the most interesting of which is the dry reforming of CH4 (DRM). This reaction converts the two most important GHG’s, namely CO2 and CH4, into new feedstock chemicals for industry. However, DRM is a very energy-intensive reaction, due to the high thermodynamic stability of its reactants, and thus requires high temperatures using conventional thermal catalysis.

Plasma technology is an interesting alternative for gas conversion applications, as it can make use of renewable energy. Moreover, non-thermal plasma (NTP) can convert thermodynamically stable molecules at mild temperatures and pressures, due to its strong non-equilibrium character and high reactivity. However, a downside to this high reactivity is that NTP is typically not very selective. Plasma catalysis therefore combines plasma with a catalyst to improve the selectivity towards the desired products. This thesis explores how the chemistry inside the plasma and on the catalyst is affected by the exchange of reactive species between both, using chemical kinetics modelling. First, the effect of vibrationally excited molecules and radicals on the surface rates and pathways of a Pt(111) catalyst in CH4 partial oxidation is studied. Our results indicate that radicals dominate the surface chemistry and their effect on the formation of different products is discussed. Next, C2H6 dry reforming in a dielectric barrier discharge (DBD) without catalyst is studied and the main reaction pathways are identified. Subsequently, a combined plasma-surface model is presented that includes both plasma and gas phase reactions. Using this model to study plasma-catalytic DRM, we illustrate that presence of a catalyst can also have a negative effect by facilitating the recombination of adsorbed radicals back to the reactants. Lastly, plasma-catalytic CO2 splitting in a low-pressure discharge is studied and the performance of different catalysts is evaluated for various reaction conditions (e.g., temperature, pressure and flow rate). Hence, the findings presented in this thesis contribute to a better understanding of plasma-catalysis, with a focus on GHG conversion.

Laura Steinwidder

• 07/09/2026
  
• 3 p.m.
• Venue: Stads91�Թ��� S.R.007
• Supervisor: Sara Vicca
• Department of Bioscience Engineering

Abstract

Climate change remains one of the most pressing global challenges, requiring rapid and sustained reductions in greenhouse gas emissions. However, even under ambitious mitigation scenarios, emission reductions alone are not sufficient to meet climate targets, making the deployment of CO₂ removal technologies necessary. Among these, enhanced weathering has emerged as a particularly promising approach, given its potential co-benefits for soil health. ​ ​

Yet, current assessments of CO₂ removal through enhanced weathering remain incomplete, as they predominantly focus on two inorganic pathways during which CO₂ removal occurs — bicarbonate leaching and pedogenic carbonate precipitation — while largely neglecting organic processes. This narrow perspective limits our understanding of the overall carbon balance and may lead to misestimation of the CO₂ removal potential of enhanced weathering. To address this research gap, in three experiments, this thesis adopted a more holistic approach, investigating the full carbon cycle, considering both inorganic and organic pathways of CO₂ removal and loss. ​ ​

Results of this thesis demonstrate that the current focus on inorganic pathways captures only a limited fraction of the processes governing carbon dynamics during enhanced weathering. Future research will therefore need to move beyond the narrow focus on inorganic processes and instead consider the coupled dynamics of mineral weathering and soil organic carbon cycling. Aside from its mechanistic insights, the assessments carried out for this thesis also underscore the substantial methodological challenges associated with quantifying CO₂ removal. The scaling of enhanced weathering will require adapted experimental designs and sampling strategies alongside more sophisticated statistical approaches (e.g. Bayesian inference) to overcome low signal-to-noise ratios. The development of standardised carbon crediting frameworks tailored to noisy datasets will be critical not only for enhanced weathering but also for other soil-based CO₂ removal technologies that face similar challenges. ​ ​

Sophie Gresham

• 04/09/2026
  
• 5 p.m.
• Venue: Stads91�Թ���, Promotiezaal Klooster van de Grauwzusters, Building S
• ​
• Supervisors: Hannes Svardal & Milan Malinsky
• Department of Biology

Abstract

Speciation has classically been viewed as a process requiring complete reproductive isolation between species. However, genomic studies increasingly show that gene flow, the transfer of genetic material between species via hybridization, can promote adaptation and speciation by boosting genetic variation within lineages. For adaptive radiations, where many ecologically diverse species arise rapidly from a common ancestor, gene flow may be an important source of variation for selection to act upon. ​ ​

The adaptive radiations of cichlid fish found in the East African Great Lakes, which show extraordinary species diversity, are an excellent system for studying the effect of gene flow in the evolution of biodiversity. In each of the major radiations, those of Lake Tanganyika, the Lake Victoria region, and Lake Malawi, gene flow has been shown to be a contributing factor in the generation of species. For the Lake Malawi radiation, which consists of around 800 species that evolved within the last 800 thousand years, only one gene flow event was previously known: the ancestral Malawi lineage hybridized with a riverine species related to Astatotilapia sp. "Ruaha blue," forming a ‘hybrid swarm’ that initialized the radiation. Still, little is known about the true extent, impact and genomic footprint of gene flow in the Malawi radiation. ​ ​

Using a whole-genome sequencing dataset of 239 Malawi and 76 non-Malawi African cichlid species, we tested for additional gene flow events that affected the Malawi radiation, both before and after the radiation’s onset. Alongside A. sp. "Ruaha blue," we found that three further lineages, related to the Congo South Africa (CSA), Pseudocrenilabrus, and Orthochromis groups, also likely contributed to the Malawi hybrid swarm. We identified and characterized hundreds of long genomic tracts of foreign CSA and Pseudocrenilabrus ancestry which were segregated amongst Malawi species. We found signals of divergent selection among Malawi species, supporting adaptive contributions from CSA- and Pseudocrenilabrus-derived ancestry. We found complex patterns of ancestry sharing between not just the Malawi radiation and neighbouring riverine species, but also with species from the Lake Victoria region, indicating previously unrecognized genetic connectivity among East African cichlids. ​ ​

Overall, these findings suggest the Malawi radiation formed through repeated intermingling of diverse cichlid lineages and that Malawi genomes are comprised of a complex mosaic of ancestries varying among species and individuals. Our results support the growing recognition that gene flow is an important driver of biodiversity. ​ ​

Stein Maerivoet

• 04/09/2026
  
• 4.30 p.m.
• Venue: Campus Drie Eiken, Q.002
• ​​
• Supervisors: Annemie Bogaerts & François Reniers
• Department of Chemistry

Abstract

Hydrogen (H₂) is expected to play an important role in the transition towards a more sustainable energy and chemical industry. However, most hydrogen is currently produced from fossil fuels, resulting in substantial CO₂ emissions. Plasma technology offers a promising alternative for converting molecules such as methane (CH₄) and ammonia (NH₃) into hydrogen and other valuable products. In plasma, a partially ionized gas, high-energy electrons and reactive species can drive chemical reactions that are difficult or inefficient under conventional conditions. Developing efficient plasma reactors requires a detailed understanding of the complex interactions between plasma chemistry, gas flow and heat transfer.

This thesis develops predictive plasma models from first principles, aiming to reproduce experimental observations without relying on experimental input as part of the model development. Two complementary approaches are considered: global models, which focus primarily on detailed plasma chemistry and kinetics, and multidimensional models, which additionally account for spatial variations, gas flow and heat transfer. This thesis develops the first warm plasma multidimensional model incorporating an extensive chemistry set demonstrates that spatial effects can strongly influence plasma behaviour and reaction outcomes, highlighting limitations of conventional global models for warm plasmas. Global modeling is subsequently used to investigate optimal operating conditions for plasma-based NH₃ conversion. The results show that continuous warm plasma operation can achieve lower energy costs than pulsed operation, while feed-gas preheating can substantially improve the efficiency of pulsed systems. Plasma-assisted NH₃ cracking is identified as a promising route for hydrogen production.

Further developments improve the agreement between multidimensional simulations and experimental observations for both NH₃ and CH₄ plasmas. These models reveal the importance of multidimensional phenomena, including the detrimental effect of turbulent transport in NH₃ plasmas, and provide initial guidance for improved reactor design. In addition, a two-zone global model is developed for nanosecond pulsed CH₄ plasmas. By calculating gas temperatures self-consistently, this approach reduces the need for user-defined model inputs while providing detailed insight into chemical pathways in both the plasma and the surrounding colder gas.

Overall, this thesis advances predictive modeling tools for plasma-based chemical conversion. By improving our understanding of plasma chemistry, transport and energy transfer, these models support the computational design and optimization of plasma reactors, potentially accelerating the development of more efficient and sustainable technologies for hydrogen production and chemical conversion.

​

Julia Camacho García

• 02/09/2026
  
• 10 a.m.
• Venue: Stads91�Թ���, Klooster van de Grauwzusters (Building S)
• ​​
• Supervisor: Hannes Svardal
• Department of Biology

Abstract

In aquatic environments, water depth represents a major axis of ecological divergence. Increasing depth is accompanied by pronounced changes in light, temperature and oxygen availability, which can promote ecological specialization and, ultimately, reproductive isolation. Understanding how organisms adapt to such environmental gradients can therefore provide insight into the evolutionary processes underlying biodiversity. The adaptive radiations of cichlid fishes provide powerful systems for studying these processes, particularly because similar adaptations have evolved repeatedly, constituting “evolutionary replicates”. This thesis investigates adaptation along the water-depth gradient in the Lake Malawi cichlid radiation, focusing on two evolutionary lineages that have diversified into deepwater habitats: the pelagic genus Diplotaxodon and the benthic subradiation.

First, I examine diversification of the visual system in Diplotaxodon, where species differ substantially in eye size, potentially linked to differences in depth ranges. Combining genome-wide association analyses, selection scans and retinal transcriptomics, I identify widespread signals of adaptive evolution across the visual system, supporting a role for visual adaptation in depth-associated ecological differentiation within Diplotaxodon.

Second, investigate the genomic basis of depth diversification in the benthic subradiation. Genome-wide association analyses identify depth-associated variants across the genome, including within chromosomal inversions previously implicated in benthic diversification. Candidate variants are enriched for nonsynonymous changes and occur in genes involved in vision, neurotransmission and circadian regulation. Comparisons among benthic lineages further reveal signals of parallel evolution, suggesting that similar selective pressures have repeatedly shaped the genome during depth diversification.

Finally, I quantify parallel genetic changes between Diplotaxodon and deep-living benthic cichlids at the genome-wide level. This analysis reveals extensive genomic parallelism, with deep-living species in both lineages repeatedly sharing allele-frequency shifts at identical genetic loci. Analyses of excess derived allele sharing and gene haplotype genealogies suggest that these parallel-evolving variants have multiple evolutionary origins, including shared ancestral variation, introgression and independent mutation. ​ ​

Altogether, this thesis shows that deepwater adaptation in Lake Malawi cichlids is shaped by both lineage-specific genetic changes and the recurrent recruitment of shared genes, pathways and variants. Visual pathways emerge as a central component, with adaptive changes spanning morphology, gene expression and protein-coding variation. These findings suggest that ecological adaptation involves coordinated changes across functionally connected pathways. More broadly, it contributes to our understanding of the repeatability of evolution and the genomic basis of ecological diversification.

Callie Ndayirinde

• 31/08/2026
  
• 4.30 p.m.
• Venue: Campus Groenenborger, G.T.148
• ​​
• Supervisor: Annemie Bogaerts
• Department of Chemistry

Abstract

The rise in global average temperature is largely driven by anthropogenic greenhouse gas emissions, including carbon dioxide (CO₂) and methane (CH₄). Among the major emitting sectors, the chemical industry is particularly energy-intensive, relying heavily on fossil fuels to sustain the high temperatures and pressures required for large-scale production. To meet the European Union’s target of climate neutrality by 2050, significant efforts are required to reduce the carbon intensity of chemical manufacturing. Electrification represents a promising pathway in this transition, with plasma technology emerging as a viable route. Specifically, we consider three key reactions: (i) NH₃ synthesis, (ii) the non-oxidative coupling of CH₄ (NOCM), and (iii) vinyl chloride production.

First, we investigate the electrification of NH₃ synthesis via plasma catalysis in a dielectric barrier discharge (DBD) reactor. Five cobalt-based catalysts are synthesized, extensively characterized, and evaluated. Although the CoLa/Al₂O₃ catalyst exhibits the best performance, differences in NH₃ concentration, energy consumption, and production rate remain limited. In contrast, significant variations are observed in plasma characteristics, including plasma power and current profiles. These results indicate that, in our system, catalysts primarily act as plasma modifiers, altering discharge behavior rather than solely driving surface chemistry, a phenomenon often overlooked in plasma catalysis.

Next, we investigate NOCM in a gliding arc plasma, producing valuable chemicals such as acetylene, ethylene, and hydrogen. A major challenge is carbon deposition, leading to reactor fouling and reduced efficiency. We investigate strategies to mitigate solid carbon formation and find that the specific energy input affects CH₄ conversion and absolute carbon formation, whereas hydrogen addition effectively suppresses carbon formation. Although process conditions influence the properties of the formed solid carbon, all samples mainly consist of turbostratic carbon and show contamination due to electrode erosion. Overall, we demonstrate the importance of suppressing carbon formation at the source to enable selective plasma-driven hydrocarbon production.

Finally, we combine insights from plasma catalysis and CH₄ conversion to develop a novel electrified route for vinyl chloride monomer (VCM) production. Plasma-driven NOCM to acetylene is integrated with post-plasma catalysis and hydrochloric acid injection, where the heat of the plasma effluent activates a downstream Au/AC catalyst for hydrochlorination to VCM. A VCM selectivity of 5% is achieved, which, while modest, indicates potential for further optimization. Overall, this work provides a proof of concept for a more sustainable VCM synthesis route with simultaneous CH₄ valorization.

Tobias Vente

• 30/06/2026
  
• 3.30 p.m.
• Venue: Campus Middelheim, A.143
• ​​
• Supervisors: Bart Goethals & Joeran Beel
• Department of Computer Science

Abstact

The unprecedented growth of digital data underscores the critical need for recommender systems (RecSys) to filter and organize relevant content. However, despite the success of RecSys, designing and developing effective research pipelines for RecSys remains a labor- and resource-intensive task. At the foundation of RecSys research experiments lies the critical task of identifying appropriate datasets for empirical evaluation. To simplify this, researchers frequently default to a narrow subset of popular datasets, which undermines robustness by constraining evaluations and risking overfitting. To ensure more robust evaluations, this dissertation first tackles the dataset selection problem. By proposing novel, data-driven methodologies and visualization tools, we enable practitioners to easily compare and select RecSys datasets. This mitigates the risk of cherry-picking and ensures generalizable benchmarking. To overcome the manual engineering of the RecSys research pipeline, this work investigates the application of Automated Machine Learning (AutoML) principles to the recommendation domain; formally known as Automated Recommender Systems (AutoRecSys). Motivated by AutoML’s success in streamlining model development, we specifically focus on solving the automated model selection problem. We evaluate the potential of out-of-the-box AutoML tools and propose the first fully automated AutoRecSys framework capable of simultaneously executing automated algorithm and hyperparameter selection.  Additionally, we provide the first systematic analysis of ensembling RecSys models to improve predictive performance and system robustness. Beyond automation and empirical evaluation, we conduct the first comprehensive investigation of the environmental impact of RecSys experiments. Our analysis reveals a concerning ecological trajectory: modern deep learning-based research papers emit approximately 42 times more CO2 equivalents than traditional approaches used 10 years ago. Given this trend, applying computationally demanding automated frameworks might seem counterintuitive. To summarize, this dissertation lays a comprehensive foundation for future research. This work advances data-driven dataset selection, pioneers AutoRecSys model selection frameworks, and quantifies ecological footprints. In doing so, it empowers the RecSys community to build highly performant, automated systems while actively minimizing their environmental impact for a sustainable future.

Denis Šabani

• 29/06/2026
  
• 4 p.m.
• Venue: Campus Groenenborger, T105
• Supervisor: Milorad Milošević
• Department of Physics

Abstract

The doctoral thesis “Origin of magnetic interactions in two-dimensional materials” addresses the microscopic mechanisms governing magnetic exchange interactions in atomically thin systems, with particular emphasis on the interplay between electronic structure, spin-orbit coupling, and low-dimensionality. The discovery of intrinsic magnetism in two-dimensional (2D) crystals has challenged long-standing theoretical paradigms rooted in the Mermin-Wagner theorem and has opened new directions in spintronics and quantum materials research. In this context, a quantitative and conceptually transparent understanding of magnetic exchange interactions is essential to predict, engineer, and control magnetic order in low-dimensional systems.

The central objective of this work is to establish a rigorous theoretical framework that connects first-principles electronic structure calculations to effective magnetic Hamiltonians capable of describing experimentally observable magnetic phenomena in 2D materials. The thesis combines density functional theory (DFT), including spin-polarized, non-collinear, and DFT+U formalisms, with two complementary exchange-mapping techniques to extract isotropic Heisenberg exchange, antisymmetric Dzyaloshinskii-Moriya interaction (DMI), symmetric anisotropic exchange (SAE), and single-ion anisotropy (SIA). Emphasis is placed on clarifying the microscopic origin of these interactions in terms of orbitally resolved contributions to the magnetic exchange between atoms and electronic hopping processes and exchange pathways that lead to these contributions.

Collectively, this thesis provides a comprehensive theoretical investigation of magnetic exchange interactions in 2D materials, bridging fundamental quantum mechanical principles and state-of-the-art first-principles simulations. The principal contributions include: (i) the development of an orbitally-resolved decomposition scheme for both isotropic and anisotropic exchange; (ii) the clarification and extension of Goodenough-Kanamori-Anderson concepts to the 2D limit; (iii) the quantitative analysis of spin-orbit-induced anisotropic interactions; (iv) the demonstration of defect-induced chiral magnetism in atomically thin systems, and (v) the comparison of two methodologies for extraction of exchange parameters from DFT.

The results presented herein deepen the microscopic understanding of magnetism in reduced dimensions and provide practical guidelines for the predictive design of 2D magnetic materials. By establishing explicit links between electronic structure, orbital hybridization, spin-orbit coupling, and effective spin Hamiltonians, this work contributes to the broader effort of engineering low-dimensional quantum materials for spintronic, magnonic, and topological applications.

Evangelia Charalampopoulou

• 24/06/2026
  
• 4 p.m.
• Venue: Campus Groenenborger, U.408
• Supervisors: Nick Schryvers, Dirk Lamoen, Rémi Delville & Konstantina Lambrinou
• Department of Physics

Abstract

This doctoral thesis presents a detailed investigation of the liquid metal corrosion (LMC) behaviour of the MYRRHA austenitic stainless steels,i.e., the DIN 1.4970 fuel cladding steel and the 316L structural steel, when exposed to the liquid lead-bismuth eutectic (LBE) MYRRHA primary coolant. The study involves the systematic investigation of dissolution and oxidation corrosion mechanisms, considering the influence of the steel microstructure and thermomechanical history, the oxygen concentration in liquid LBE, and the exposure temperature and duration. In oxygen-poor LBE environments (Co < 10⁻⁸ mass%), dissolution corrosion is dominant. Fine-grained cold-drawn steels with high density of deformation twins corrode faster than coarse-grained solution-annealed steels. Dissolution corrosion often involves selective leaching of the austenite stabilisers Ni and Mn, destabilising the austenitic phase and leading to ferritisation. An interesting finding regards the fcc-to-bcc (γ→α) phase transformation resulting from the steel exposure to liquid LBE, which follows the Pitsch orientation relationship (OR) rather than the more commonly reported Kurdjumov–Sachs (K-S) and Nishiyama–Wassermann (N-W) ORs. In the case of oxygen-rich LBE (Co > 10⁻⁶ mass%), oxidation corrosion becomes dominant. It has been found that complex oxide scales form on the DIN 1.4970 fuel cladding steel surface, including a nanostructured internal oxidation zone (IOZ). Due to the inherent limitations of the active oxygen control LMC mitigation strategy that relies on the in situ formation of protective oxide scales on the steel surface, this PhD study also tested candidate protective coatings as an alternative LMC mitigation strategy.

MohmmadSadegh Mokhtari

• 24/06/2026
  
• 4 p.m.
• Venue: Stads91�Թ���, S.R.218
• ​
• Supervisors: Jeroen Famaey & Bram Vanderborght
• Department of Computer Science

Abstract

Energy management in robotic systems has traditionally been driven by the limitations of onboard energy. In collaborative multi-robot systems, it is not only about conserving energy but also about improving how efficiently robots use their time, how reliably they operate, and how long they remain autonomous. This thesis investigates energy management strategies for collaborative multi-robot systems, with emphasis on scenarios where both energy consumption and time of use are equally important. The work combines system development, realistic energy modeling, scheduling optimization, and energy-aware path planning, supported by real robot deployment and simulation. The motivation arises from ambiently-powered multi-robot systems harvesting energy from renewable sources. Although on-board harvesting can extend operation, it remains insufficient to fully sustain robotic systems, and robots must rely on charging infrastructure such as docking stations or opportunistic wireless energy transfer.

In such environments, energy management becomes less about conservation and more about intelligently coordinating energy usage, task execution, and charging activities to sustain long-term autonomous operation. Efficiency refers to maintaining continuous task execution and maximizing productivity while minimizing operational interruptions caused by charging or idle states. These challenges are particularly relevant in long-duration applications such as warehouse automation, environmental monitoring, agriculture, and surveillance, where idle energy consumption can significantly reduce operational persistence.

The first research question investigates how idle energy consumption can be managed to improve long-term autonomy. A low-power Energy Management Module (EMM) was designed and deployed on a real TurtleBot3 robot, reducing idle-mode power consumption from approximately 7 W to 26.2 µW.

The second research question examines energy-aware scheduling in multi-robot systems. Incorporating charging decisions directly into the scheduling process reduced total mission makespan by up to 15%, highlighting the importance of integrating energy replenishment into task allocation.

The third research question explores scalability. A hybrid multiverse-based parallel scheduling framework was developed, distributing computation across the robotic swarm and making scheduling up to 5 times faster.

The fourth research question extends energy-aware scheduling to UAVs. The EA-MAOSS framework reduced mission makespan by up to 30% and charging events by more than 50% compared to fixed-threshold approaches.

The fifth research question focuses on exploiting energy harvesting during navigation. The H-CMARL-DE framework achieved up to 240% higher energy-harvesting efficiency compared to classical planners, while maintaining path-length efficiency within 10% of shortest-path solutions.

Lucas Deckers

• 23/06/2026
  
• 5 p.m.
• Venue: Stads91�Թ���, Klooster van de Grauwzusters, Promotiezaal (Building S), Lange Sint-Annastraat 7, 2000 Antwerpen
• Supervisors: Steven Latré & José Antonio Oramas Mogrovejo
• Department of Computer Science

Abstract

Recent advances in hardware and algorithms have enabled AI systems to tackle tasks once considered uniquely human. However, the growing size of modern AI models brings substantial energy demands, raising environmental concerns. This thesis explores Spiking Neural Networks (SNN) as a promising low-power alternative. Inspired by the brain, SNN can operate 10 to 1.000 times more efficiently than conventional AI when deployed on neuromorphic hardware, though they often fall short in performance on complex tasks. To address this, we propose several biologically motivated extensions to SNN models, focusing on learning mechanisms and performance improvements.

First, we propose three biologically inspired enhancements to the Liquid State Machine (LSM) framework: balancing excitatory and inhibitory input currents, incorporating SFA, and introducing neuronal heterogeneity. These extensions collectively enrich the reservoir’s temporal dynamics, leading to improved class separation and enhanced classification performance across multiple speech recognition benchmarks. We demonstrate that the resulting ELSM model approaches state-of-the-art accuracy, rivaling end-to-end trained AI models, while maintaining the efficiency and simplicity of the LSM architecture.

Next, we explore end-to-end training of SNN using. We introduce the cAdLIF neuron model, which enables joint optimization of homeostatic adaptation parameters alongside a novel temporally-aware learning rule for synaptic transmission delays. This combination allows neurons to explicitly correlate temporally dispersed input patterns, enhancing temporal processing capabilities. We demonstrate that these biologically inspired enhancements lead to substantial performance gains across benchmark datasets and show that the proposed model approaches or surpasses state-of-the-art results in SNN.

Next, we introduce Twin Network Augmentation (TNA), a new method for improved model generalization and compression to binary/ternary weight precision, through a twin-network setup. Two identical architectures are co-trained using individual cross-entropy losses and a shared logit matching loss, encouraging consistent decision boundaries. We show that TNA improves performance over the baseline across all evaluated datasets and outperforms a related technique, knowledge distillation. Finally, we extend the TNA framework to non-spiking ANN, including more complex architectures such as CNN and transformers. We validate the method on popular vision benchmarks and conduct extensive ablation studies to isolate the source of performance gains. Our results show that improvements stem from mutual learning via logit matching rather than an increase in trainable parameters. Furthermore, analysis of internal representations reveals that TNA-trained models develop more discriminative high-level features and exhibit enhanced robustness to input corruptions.

Sarah Leyder

• 22/06/2026
  
• 4 p.m.
• Venue: Campus Middelheim, A.143
• Supervisors: Tim Verdonck & Jakob Raymaekers
• Department of Mathematics

Abstract

This dissertation advances the field of robust multivariate statistics by developing new methods that are well-suited for data containing outliers and heavy tails. By introducing targeted robust procedures, including specialized estimators and tailored data transformations, we adapt four existing statistical methods and demonstrate their effectiveness through simulations and real-world applications. ​ ​

First, we propose a computationally efficient robust principal component analysis method (PCA) based on decomposing the generalized spatial sign covariance matrix. This approach overcomes the computational bottlenecks and sample-size limitations of existing robust PCA methods while maintaining flexibility through various radial functions. Second, in the domain of causal inference, we introduce TSLiNGAM. By replacing the non-robust ordinary least squares regression in the original LiNGAM algorithm with Theil-Sen regression, this method achieves superior causal discovery on heavy-tailed and skewed data. Third, addressing the need for a highly robust independence measure, we introduce the biloop data transformation to robustify distance correlation. The resulting biloop distance correlation achieves a bounded, redescending influence function and a non-zero breakdown point. Fourth, we leverage the biloop distance correlation to develop a new approach for independent component analysis (ICA). By applying a novel transform called bowl to the dCovICA method, we effectively separate multivariate sources in the presence of outliers. Finally, to bridge the gap between robust statistics, traditionally more established in R, and Python-based data analysis, we present RobPy, a comprehensive Python library for robust regression, PCA, covariance estimation and cellwise outlier detection. ​ ​

Pedro Miguel Alves Santos

• 20/06/2026
  
• 10 a.m.
• Venue: Campus Middelheim, G.010
• Supervisor: Sonja Hohloch
• Department of Mathematics

Abstract

The study and classification of integrable systems have been central problems in mathematics and physics for centuries. This thesis investigates two complementary directions in the theory of integrable systems.

First, we study the symplectic topology of the fibers of semitoric systems. We extend McDuff’s method of probes from toric to semitoric systems and apply it to analyze the displaceability and nondisplaceability properties of fibers in three distinct examples. These results provide new insight into the symplectic topology underlying semitoric systems beyond the toric setting. ​ ​

The second part of the thesis develops symplectic invariants for simple hypersemitoric systems, with the broader goal of extending the classification theory of semitoric systems to hypersemitoric systems. Hypersemitoric systems generalize semitoric systems by allowing for more complicated singularities, and a complete classification theory for them is not yet known. We introduce the affine invariant of simple hypersemitoric systems, generalizing both the Delzant polytope of toric systems and the polytope invariant of semitoric systems. We compute and visualize this invariant for a range of increasingly sophisticated examples, illustrating the richness of the hypersemitoric setting. ​ ​

Finally, we define the J-height invariant and the J-Taylor series invariants of hyperbolic-regular lines in two-degree-of-freedom integrable systems admitting a circle symmetry. Furthermore, we compute these invariants in two topologically distinct examples, providing explicit models that demonstrate their geometric and topological significance. ​ ​

Together, these results contribute new tools for the study and classification of integrable systems and open pathways toward a broader classification theory beyond the semitoric case.

Saja Tawalbeh

• 19/06/2026
  
• 3 p.m.
• Venue: Campus Middelheim, A.143
• ​
• Supervisor: José Antonio Oramas Mogrovejo
• Department of Computer Science

Abstract

This thesis is dedicated to the analysis of internal representations encoded within capsule neural networks (CapsNets), with particular emphasis on both interpretability and explainability tasks. In the first two chapters, we focus on examining interpretation methods and assessing the interpretable capabilities of CapsNets, including analyzing the extent to which part-whole relationships are encoded within the learned representations. We also study the explanation methods in the first part of the thesis and propose an explanation method specifically tailored for feed-forward CapsNets. Finally, we explore whether CapsNets encode semantic concepts within their internal representations.

Ioanna Gavriilidi

• 16/06/2026
  
• 4 p.m.
• Venue: Campus Drie Eiken, Building Q, Promotiezaal Q2
• ​
• Supervisors: Raoul Van Damme, Simon Baeckens & Panayiotis Pafilis
• Department of Biology

Abstract

This thesis investigates whether and how insularity shapes cognition and personality, using wall lizards (genus Podarcis) as model organisms. Cognitive and personality traits allow animals to respond to socio-ecological challenges, but their expression also entails energetic and ecological costs. Natural selection is therefore expected to favour behavioural phenotypes that balance these costs and benefits within specific environmental contexts. Islands provide “natural laboratories” for studying such evolutionary processes because they often share ecological characteristics that consistently differ from the mainland. These conditions have led many insular animals to evolve characteristic morphological, physiological, and life-history adaptations collectively known as the “island syndrome”. While these adaptations are well documented, the effects of insularity on cognition and personality remain poorly understood.

The first chapter reviews the existing literature on behavioural evolution on islands and argues that insularity is likely to shape cognitive and personality traits in multiple ways, although the direction of change is not always easy to predict. Apart from the well-documented phenomenon of “island tameness,” empirical evidence for behavioural components of the island syndrome remains limited and inconclusive. The second chapter compares cognitive and personality traits between mainland and insular lizards. While most traits showed no consistent differences in mean values, insular populations exhibited reduced aggressiveness, likely reflecting reduced territoriality at high population densities. Behavioural traits were also more repeatable in insular populations, suggesting greater behavioural specialization and reduced flexibility on islands.

The third chapter examines the effects of insularity on a social component of cognition by comparing lizards’ attraction to a foraging site by the presence of conspecifics. Both mainland and insular lizards modified their foraging behaviour in the presence of conspecifics, but insular individuals engaged more frequently in aggressive interactions, suggesting that the costs and benefits of social foraging differ between habitats.

The fourth chapter investigates the effects of insularity on the chemical discrimination abilities. Although both mainland and island lizards could discriminate between chemical cues, island individuals showed reduced tongue-flick rates, suggesting reduced chemosensory responsiveness.

The fifth chapter examined the neuroanatomical consequences of insularity. Insular lizards possessed relatively smaller brains and longer digestive tracts, consistent with the expensive tissue hypothesis under energetic constraints. Contrary to expectations, insular lizards had also relatively larger olfactory bulbs, indicating that chemoreception remains functionally important despite reduced tongue-flick rates.

Overall, the thesis demonstrates that insularity influences cognition and personality, although these effects are subtler and less consistent than morphological or life-history changes.

Julie Symons

• 05/06/2026
  
• 4 p.m.
• Venue: Campus Middelheim, A.143
• ​​
• Supervisors: Wendy Lowen & Michel Van den Bergh
• Department of Mathematics

Abstract

Driven by the classical interpretation of the lower Hochschild cohomology groups of an abelian category in terms of infinitesimal deformation theory, we seek tangible deformation-theoretic interpretations for all higher Hochschild cohomology groups. To achieve this, we make the transition to dg-categories, and more specifically to t-dg-categories, which provide a bridge between the abelian and dg-frameworks. ​ ​

We develop a deformation theory for such categories, leading to an interpretation of all their higher Hochschild groups in terms of so-called t-deformations. In addition, we address the question of whether the dg-enhancement that is part of the data of a t-dg-category is unique, or whether we are making a choice (that in turn influences the Hochschild complex). We show that uniqueness fails in general by constructing counterexamples.

Daniël Sebastiaan Verdoorn-Kański ​ ​

• 26/05/2026
  
• 1 p.m.
• Venue: University of Maastricht, Aula, kamernummer 0.023, Minderbroedersberg 4-6, 6211 LK Maastricht
• ​
• Supervisors: Bert Maes & Romano Orru
• Department of Chemistry

Abstract

This thesis investigates cobalt-catalyzed carbene and nitrene transfer reactions to isocyanides, with the aim of developing efficient and sustainable methodologies for the synthesis of heterocyclic compounds and heteroallenes. While noble metals have traditionally dominated group transfer chemistry, cobalt offers an attractive alternative due to its low cost and environmental compatibility. The work presented here demonstrates that cobalt catalysts enable versatile multicomponent transformations and provides mechanistic insight into their mode of action. ​ ​

A cobalt(II)-mediated three-component reaction was developed for the synthesis of 5-substituted N-sulfonyl-1,3,4-oxadiazol-2(3H)-imines from sulfonyl azides, N-isocyaniminotriphenylphosphorane (NIITP), and carboxylic acids. The reaction proceeds through a cascade involving nitrene transfer to the isocyanide, in situ formation of a carbodiimide intermediate, and intramolecular aza-Wittig cyclization. Reaction selectivity was found to depend strongly on both the steric profile of the carboxylic acid and the cobalt loading, providing control over product formation. This methodology enables rapid access to structurally diverse oxadiazole imines from readily available starting materials. ​ ​

In a complementary approach, a cobalt-catalyzed three-component synthesis of 2,5-substituted 1,3,4-oxadiazoles was established using electron-deficient diazo compounds, NIITP, and carboxylic acids. In this transformation, carbene transfer to the isocyanide generates a ketenimine intermediate, which undergoes nucleophilic addition followed by intramolecular aza-Wittig cyclization. The reaction shows broad substrate scope and is compatible with functionalized substrates, enabling late-stage derivatization of molecules of pharmaceutical relevance. ​ ​

Mechanistic investigations were extended to a cobalt-based metal–organic framework (ZIF-67) applied in nitrene transfer reactions to isocyanides. Although initially designed as a heterogeneous catalyst, kinetic studies revealed that trace cobalt leaching (down to 0.04 mol%) is sufficient to sustain catalysis. The active species are proposed to be homogeneous cobalt complexes stabilized by coordination to isocyanides, highlighting the non-innocent role of the isocyanide substrate. This slow-release system displays significantly enhanced catalytic efficiency compared to conventional homogeneous cobalt systems. ​ ​

Overall, this work establishes cobalt as a powerful catalyst for carbene and nitrene transfer to isocyanides, enabling efficient access to heterocyclic scaffolds while providing fundamental insight into catalyst speciation and reactivity. These findings contribute to the development of sustainable transition metal catalysis and highlight the importance of mechanistic understanding in catalyst design.

Maria Gabriela Maridueña Zavala

• 30/04/2026
  
• 4 p.m.
• Venue: Campus Groenenborger, G.T.135
• Supervisors: Gerrit Beemster, Hamada AbdElgawad & Carlos Noceda
• Department of Biology

Abstract

Plant viruses pose significant constraints on crop productivity, with papaya exhibiting particular susceptibility due to the rapid impact of viral infections on fruit yield and quality. Systems-level approaches, including transcriptomics, metabolomics, and epigenomics, are crucial for elucidating the complex host responses that underlie susceptibility and tolerance. Babaco mosaic virus (BabMV) remains insufficiently characterized at the molecular level, while integrative multi-omics approaches are still rarely applied in plant virus research, together limiting the identification of effective ways to manage the disease and the selection of more resistant papaya plants. This thesis addresses this knowledge gap by presenting an integrated, cultivar-specific framework for understanding BabMV–papaya interactions. ​ ​

The main objective of this doctoral thesis was to investigate the temporal response of Carica papaya to Babaco mosaic virus (BabMV) through a multi-layered integration, supported by a novel genomic assembly. The analyses reveal that the papaya response constitutes a synchronized transition through distinct biological phases, rather than a static stress state. ​ ​

The results demonstrate that BabMV infection progresses through distinct early and late stages, as reflected in transcriptomic, metabolomic, and epigenetic data. The early stage is characterized by cell-wall reinforcement, and initial effects on photosynthesis, including chlorophyll decline, without complete metabolic disruption. Concurrently, metabolites associated with sulfur metabolism, phenolic compounds, and tricarboxylic acid (TCA) cycle organic acids increase, indicating early carbon reorganization and the initiation of secondary metabolism activation. ​ ​

In contrast, the late stage is characterized by the predominance of stress-response, redox, and antioxidant programs, consistent with sustained oxidative pressure and intensified metabolic reprogramming. Defense-related metabolites, particularly those associated with proline, polyamine, and phenylpropanoid pathways, become increasingly prominent, alongside elevated phenylalanine ammonia-lyase (PAL) activity and enhanced antioxidant buffering. These later responses are accompanied by broader DNA methylation remodeling, with genes involved in catalytic activity, endomembrane organization, sugar metabolism, and chloroplast redox regulation. Collectively, these findings indicate that BabMV infection represents a coordinated transition from early adjustment to a later defense- and stress-dominated state, rather than a simple on-off response. ​ ​

This thesis offers new insights into BabMV–papaya interactions by constructing a temporally resolved, multi-omics perspective of infection. Together, these findings provide a more complete framework for understanding disease progression and create a stronger basis for future studies aimed at improving papaya virus management. ​ ​

Ali Ghojavand

• 27/04/2026
  
• 4 p.m.
• Venue: Campus Groenenborger, US.025
• Supervisors: Milorad Milosevic, Cem Sevik & Eric Bousquet
• Department of Physics

Abstract

Magnetoelectric (ME) coupling in multiferroics has attracted considerable attention over the past decade due to its wide range of applications in spintronic devices, data storage technologies, and electrically tunable microwave components. Different ME coupling mechanisms underpin distinct device functionalities, and a fundamental understanding of their dynamics remains essential for advancing high-performance multiferroic materials. Recently, two-dimensional van der Waals magnets have emerged as a particularly promising platform for realizing and tuning such ME phenomena at the atomic scale. Since the discovery of antiferromagnetic order in FePS3 crystals, this rapidly growing family of materials has revealed a wealth of exotic magnetic phases and offers a versatile foundation for engineering heterostructures due to their weak interlayer interactions. Within this family, NiX2 (X = I, Br, Cl) monolayers stand out as experimentally realized systems, successfully fabricated in both few-layer and monolayer forms. ​ ​

In this thesis, we develop effective design strategies to tailor the ME properties of NiX2 monolayers. We first investigate their intrinsic magnetism, where the interplay between magnetic frustration and biquadratic exchange defines the ground-state properties. We then establish strain as a powerful control parameter: both biaxial and uniaxial strains prove highly effective in tuning critical temperatures, and stabilizing a variety of spin configurations, including ferromagnetic, spiral, and skyrmionic phases. ​ ​

Beyond magnetism, we demonstrate that strain directly controls the magnitude and direction of electric polarization by tuning the noncollinear spin textures and ME tensor components. This reveals a robust and highly tunable ME coupling in these materials. ​ ​

This comprehensive exploration is extended to evaluate the impact of point defects, revealing their critical role in pinning magnetic textures and subsequently tailoring the resulting spin-induced electric polarization. All the findings together establish a multi-faceted engineering strategy, combining strain and defect control, to master coupled magnetic and ferroelectric orders in a single 2D material, paving the way for innovative device concepts in nanoelectronics and spintronics.

Bereket Alemayehu Arba

• 23/04/2026
  
• 2.30 p.m.
• Venue: Campus Drie Eiken, Building O, Aula O.5
• ​​
• Supervisors: Herwig Leirs & Aberham Megaze
• Department of Biology

Abstract

Cutaneous leishmaniasis (CL) is a vector-borne disease transmitted by the bite of infected female sandflies. In Ethiopia, CL is primarily caused by Leishmania aethiopica and is considered a zoonosis involving sandfly vectors and hyrax reservoirs. Complex ecological and epidemiological interactions among hosts, vectors, and environmental conditions influence its transmission. Understanding these dynamics is essential for designing effective prevention and control strategies. This PhD study aimed to: (i) assess community awareness and epidemiology of CL, and (ii) investigate the ecology of CL, with emphasis on sandfly abundance and distribution. These aims were addressed in four components: (1) assessing community knowledge, attitudes, and practices (KAP) regarding the disease; (2) exploring the disease eco-epidemiology and confirming active transmission among humans, sandflies, and hyraxes; (3) determining sandfly abundance and distribution along altitudinal gradients, focusing on Phlebotomus pedifer; and (4) assessing the abundance of P. pedifer across habitat types. A community-based cross-sectional design was used for the KAP and epidemiological investigations. Structured questionnaires were administered to household heads. Suspected CL cases were sampled for laboratory confirmation. Sandflies were collected indoors and outdoors using CDC light and sticky traps, and hyraxes were captured using snare traps. Human, sandfly, and hyrax samples were screened for Leishmania infection, and blood-meal analysis was conducted to identify blood-meal origins for sandflies. Data were statistically analyzed, and appropriate tests were performed for associations. The results showed a low level of community knowledge of CL and its prevention. The overall CL prevalence (active lesions and scars combined) was 12.7%, with children being the most affected. L. aethiopica infection was confirmed in humans, sandflies, and hyraxes, demonstrating active local transmission. Sandflies from the genera Phlebotomus and Sergentomyia were identified. Humans were the predominant blood-meal source for P. pedifer. Sandfly abundance and distribution varied by altitude, habitat, and season. Phlebotomus pedifer density increased with altitude and was highest in highland areas, whereas P. alexandri, P. gibiensis, and Sergentomyia spp. were more common at lower elevations. In conclusion, the study confirms active CL transmission and highlights significant knowledge and prevention gaps within the study communities. The findings emphasize the need for targeted health education and vector control strategies, particularly around hyrax dwellings and human residences in the study areas.

Hanan Barich

• 21/04/2026
  
• 4 p.m.
• Venue: Campus Drie Eiken, Building O, room O.1
• Supervisor: Karolien De Wael
• Department of Bioscience Engineering

Abstract

Growing environmental concerns and stricter sustainability regulations highlight the urgent need for improved monitoring of chemical pollutants in water systems. Among these, phenolic compounds are particularly problematic due to their widespread use and associated risks, including toxicity, persistence, bioaccumulation and endocrine-disrupting effects. Their occurrence in industrial and environmental waters highlights the need for advanced detection strategies that are both selective and sensitive.

This PhD research addresses this challenge through the development of a combi-electrosensor, combining square voltammetric (SWV) and photoelectrochemical (PEC) sensing, for the selective and sensitive detection of phenolic compounds in industrial wastewater. The SWV sensor enables selective identification of phenolic compounds via electrochemical fingerprinting, while the PEC sensor allows sensitive quantification. The PEC sensor uses a type II photosensitiser (PS), a highly fluorinated zinc phthalocyanine derivative, which generates singlet oxygen (1O2) under illumination to oxidise phenolic compounds. The oxidised phenols are subsequently reduced at the electrode surface, establishing an electrocatalytic redox cycle that generates a measurable photocurrent, indicative for the total phenol content.

This thesis includes the optimisation and validation of the SWV sensor for selective phenol detection, followed by its integration with the PEC sensor to create a complementary system capable of both identification and quantification. The resulting combi-electrosensor demonstrated strong performance in industrial wastewater samples, outperforming commercial test kits in terms of analysis time, selectivity (via SWV) and sensitivity (via PEC), with detection limits reaching the nanomolar range. To enhance sustainability and accessibility, an alternative commercially available PS, pheophorbide a, was evaluated, reducing reliance on fluorinated compounds. In parallel, graphite-based screen-printed electrodes were developed in-house, enabling reproducible, cost-effective and scalable sensor fabrication with performance comparable to commercial electrodes. Finally, the integration of the combi-electrosensor into a flow system highlighted the adaptability and versatility of this approach, enabling real-time, continuous monitoring of phenolic compounds. This dual-mode in-flow sensing approach enhances sensitivity, while offering additional benefits such as SWV sensor reusability, faster analysis, automated sample injection, and real-time detection of low-concentration phenols. Altogether, it provides a robust and efficient strategy that outperforms commercial test kits and supports sustainable, cost-effective water quality monitoring.

Jenevieve Hara

• 21/04/2026
  
• 2 p.m.
• Venue: Campus Drie Eiken, Building R, Aula R3
• ​
• Supervisors: Raewyn Town & Ronny Blust (University of Antwerp) - Jana Asselman & Colin Janssen (University of Ghent)
• Department of Biology

Abstract

Micro- and nanoplastics (MNPs) are pervasive in marine environments, posing significant ecological risks due to their persistence and physicochemical heterogeneity. Environmental aging alters particle properties, influencing interactions with marine organisms. However, mechanistic understanding of their biological effects under environmentally realistic conditions remains limited. This thesis addresses these gaps by elucidating the biological reactivity of aged MNPs, focusing on cellular, tissue, and organismal responses under environmentally relevant exposure scenarios. ​ ​

A systematic assessment of MNP toxicity in marine invertebrates was conducted using chronic exposure data from the Toxicity of Microplastic Explorer (ToMEx) for Aquatic Organisms database (version v2.0). Species sensitivity distributions were constructed based on fitness-related endpoints, accounting for two distinct toxicity mechanisms. Derived HC₅ values were used to propose provisional thresholds for protecting marine invertebrate communities. ​ ​

Complementary 𝘪𝘯 𝘷𝘪𝘷𝘰 and 𝘪𝘯 𝘷𝘪𝘵𝘳𝘰 experiments provided integrative evidence that aged polyethylene terephthalate (PET) MPs (~1.9 µm) and NPs (~0.68 µm) elicit cellular- and tissue-level effects in mussels (𝘔𝘺𝘵𝘪𝘭𝘶𝘴 𝘦𝘥𝘶𝘭𝘪𝘴) under environmentally relevant exposure conditions. Results revealed both immediate cellular perturbations and recovery potential. Histopathological analyses demonstrated structural alterations in key organs, particularly the gills, indicating potential impairment of essential physiological functions. Although no mortality or growth effects were observed, exposure induced systemic sublethal responses that may compromise organismal fitness under chronic or combined stress conditions. ​ ​

Through label-free 3D live-cell imaging by holotomography, unlabeled aged PET NPs were successfully resolved within living hemocytes, revealing particle internalization and localization patterns, as well as associated morphological changes. ​ ​

Overall, this thesis advances mechanistic understanding, spanning from community-specific hazard thresholds to the intricate cellular mechanisms underlying particle-induced toxicity. ​ ​

Benjamin Vandersmissen

• 20/04/2026
  
• 5 p.m.
• Venue: Stads91�Թ���, S.R.219
• ​​
• Supervisor: José Antonio Oramas Mogrovejo
• Department of Computer Science

Abstract

After the discovery that the training of neural network can be vastly accelerated by exploiting the parallelization capacities of commercially available GPUs, deep learning models have been used to tackle increasingly complex problems. Together with the complexity of these problems, the number of trainable parameters in a neural network has also seen a drastic increase. While additional parameters are often necessary to tackle more complex problems, most models have vastly more parameters than strictly required for the specific task. This has the effect that computational costs in both the training and inference phase are significantly higher than they could be. ​ ​

This overparameterization is often deemed necessary for good convergence behaviour on the problem setting. Indeed, larger models often generalize better than smaller models. To reduce computational cost, techniques have been introduced that shrink the model while retaining the same generalization. Because naively training a smaller network results in worse generalization, typical approaches start from a large dense network that is fully trained, and condense their encoded representations into a smaller network. In this thesis we explore two different techniques to achieve this, namely output matching, and weight pruning. ​ ​

First we introduce Twin Network Augmentation (TNA) which is a methodology that minimizes the distance between the outputs of two small neural networks during training. We show that this method results in better generalization than training a single bigger neural network, while requiring the same computational resources during training, and being more efficient during inference. We furthermore extend this formulation by introducing techniques promoting smarter matching of the outputs, and generalize the formulation to include an arbitrary number of neural networks. ​&�Բ�����;

Next, we investigate the Lottery Ticket Hypothesis (LTH). This hypothesis states that within any dense overparameterized network there exists a sparse network that can be trained to the same generalization as the dense network. Specifically we focus on different aspects of the common algorithm – Iterative Magnitude Pruning (IMP) – to generate these sparse networks, namely the pruning criterion determining which connections should be removed from the network, and analyze the impact of different hyperparameters on generalization. ​ ​

Finally, we integrate Twin Network Augmentation (TNA) with the Iterative Magnitude Pruning (IMP) algorithm and employ it to significantly reduce the computational cost associated with finding sparse trainable neural networks. Our contributions in this work thus serve to push the boundaries of parameter-efficient neural networks. ​ ​

Bharadwaj Mysore Ramesha

• 02/04/2026
  
• 1 p.m.
• ​
• Supervisor: Vera Meynen
• Department of Chemistry

Abstract

Porous materials have revolutionized the world of chemistry by the role they play in catalysis, separation, gas storage, carbon capture, waste management and many other (industrial) applications. Hybrid porous materials are a subclass of porous materials wherein inorganic and organic precursors are combined synergistically to obtain porous frameworks whose surface chemistry and pore architectures are adjusted to fit the application. However, the limited solubility of Ti(IV) ions and the high reactivity of titania precursors often leads to materials that are poorly organized, lacking controlled porosity. This is further complicated by the addition of organodiphosphonic precursors whose chelating ability is dependent on the phosphonic functionality (ester versus acid). In this PhD, insights into the formation of titanium phosphonates by combining titanium alkoxides with organodiphosphonic precursors were achieved by identifying the role of key parameters in the synthesis protocol, such as the choice of organodiphosphonic precursor, solvent, and amount of acid on the formation of hybrid titanium phosphonate is studied. The reaction between titanium alkoxides and diphosphonic acids result in rapid precipitation with limited control over the structure of the titanium phosphonate formed. In contrast, the reaction between titanium alkoxides and diphosphonic esters results in the formation of TiO2 clusters that are too weakly bonded with the phosphonic moiety. As phosphonic esters and acids present two extremes in reactivity, the influence of varying HCl content, enabling (partial) in-situ hydrolysis of the phosphonic ester precursor, was evaluated while all other ingredients were kept constant. The fate of the organodiphosphonic ester (TEPD) during the formation of hybrid titanium dioxide phosphonates was followed by 1H and 31P NMR spectroscopy. It revealed that varying the HCl amounts mainly affects the degree of hydrolysis of TEPD. The change in hydrolysis degree with HCl content leads to changes in the surface bonding dynamics resulting in the variation in particle aggregation. Moreover, when altering the carbon content of the alcohol used in the synthesis, the hydrolysis degree of TEPD could be broadly varied. As the carbon content of alcohol increases the degree of hydrolysis of TEPD increases in the order of n-butanol > isopropanol > ethanol. This adds knowledge to the field of hybrid titanium phosphonates synthesis, that in a next step will be used to systematically alter materials with the aim to tune their performance in applications such as catalysis, separation and proton conduction.

Philip Ley

• 02/04/2026
  
• 5 p.m.
• Venue: UCSIA - Manresazaal, Koningstraat 2, 2000 Antwerpen
• Supervisors: Filip Meysman & Jeanine Geelhoed
• Department of Biology

Abstract

In 2012, a novel group of long, filamentous bacteria was discovered in the seafloor. It was shown that these so-called “cable bacteria” facilitate electrical currents over centimeter distances. Somehow, living organisms are capable of efficiently conducting electricity. Later on, it was found that the cell envelope of these cable bacteria harbors specialized conductive wires. The conductivity of these wires is extremely high, and supersedes that of any other biomaterial. This discovery holds great promise for the development of new conductive bio-based electronic materials.

In this PhD project, we investigated different aspects of the phylogenetic and morphological diversity of cable bacteria. Currently, there are two established genera of cable bacteria: Candidatus Electrothrix, mainly inhabiting saltwater sediments, and Candidatus Electronema, predominantly living in freshwater sediments. However, the overall genetic diversity of cable bacteria is likely underestimated. We expanded the current phylogenetic tree of the cable bacteria clade by sequencing the 16S rRNA gene of cable bacteria from various unstudied locations and compiled public sequencing data. In total, we identified more than 90 potential species-level clades across six genus-level clusters, including a distinct cluster of potential deep sea-inhabiting species.

Besides the large phylogenetic diversity, different cable bacteria species are also highly variable in their morphological traits. We conducted a systematic morphological investigation of the cable bacteria clade using high-resolution microscopy images. While filaments of the same strain show little morphological variation, different species can vary significantly in their filament diameters and the number of ridges. Filaments with larger diameters show a larger capacity to transport electrons, as characterized by the larger number of conductive wires and the area of the wires.

Among the newly identified cable bacteria in this project, one species stood out due to its distinct phylogenetic placement between the two established cable bacteria genera. After developing a single-species culture, we characterized its morphology and genomic potential. It showed distinct, broad ridges and produced tube-shaped, extracellular sheaths. Interestingly, this species shares metabolic pathways and genes with both established genera. We proposed the name Candidatus Electrothrix yaqonensis YB6. As little is known about the formation of extracellular sheaths in cable bacteria, we selected sheath-forming cable bacteria and characterized the prevalence, detailed structure and chemical composition of these sheaths using different microscopy and spectroscopic methods. Sheaths consist of highly organized, parallel nanofibrils which are composed of anionic polysaccharides. Potential functions are the protection from grazers or aid in motility.

Lucilla Boito

• 02/04/2026
  
• 4 p.m.
• Venue: Stads91�Թ���, SR 004
• ​ 
• Supervisor: Sara Vicca
• Department of Bioscience Engineering

Abstract

Climate change constitutes one of the most urgent global challenges, driven predominantly by rising atmospheric concentrations of greenhouse gases. To remain within the temperature limits stipulated by the Paris Agreement, scenario analyses increasingly indicate that substantial atmospheric carbon dioxide removal will be required alongside rapid emission reductions. Within this context, enhanced weathering (EW) of silicate minerals has emerged as a promising carbon dioxide removal (CDR) strategy. By accelerating natural silicate dissolution, EW removes atmospheric CO₂. However, despite accelerating interest in EW, significant uncertainties persist regarding its interactions with the complex biogeochemical processes of agricultural soils, most notably soil organic carbon (SOC) dynamics, nitrogen cycling (with particular attention to nitrous oxide (N₂O) emissions), and crop performance. ​ ​

This dissertation addresses these uncertainties through controlled mesocosm experiments and a field trial in Westmalle, Belgium, employing both natural silicates (basalt) and steel slags, derived from industrial steel production. Across all experiments, EW consistently elevated soil pH and base cation availability, confirming its capacity to counteract soil acidification. In contrast, crop biomass, nutrient uptake and broader agronomic performance exhibited only minimal responses, suggesting that agronomic co benefits are likely limited in already relatively fertile agricultural systems. This thesis also shows that EW affect SOC cycling, with responses strongly shaped by biotic interactions: basalt reduced soil organic matter decomposition only in plant free mesocosms, indicating SOC stabilization. In planted treatments, basalt increased rhizosphere activity, underscoring the central role of biological processes in governing EW outcomes. ​ ​

EW also influenced nitrogen cycling, with steel slags treatments increasing N₂O emissions, thereby revealing a risk associated with EW deployment. Heavy metal mobilisation due to silicate application remained generally low, yet measurable increases in nickel highlight the need for continued environmental monitoring of EW to ensure its safe and sustainable application. Across the thesis, mineralogical differences among feedstocks drove variability in agronomic responses and greenhouse gas fluxes, emphasizing the critical importance of feedstock selection. ​&�Բ�����;

Overall, the findings demonstrate that EW can contribute to SOC stabilization and mitigate soil acidification, yet its net climate effect depends on interactions among feedstock mineralogy, soil properties and biological processes. These results highlight the necessity for monitoring, reporting and verification frameworks to account not only for inorganic carbon sequestration but also for organic carbon cycling and emissions of non-CO2 greenhouse gases such as N2O. This dissertation provides a more comprehensive understanding of EW and a foundation for its responsible implementation in agricultural landscapes.

Koray L. Iroc

• 30/03/2026
  
• 4 p.m.
• Venue: Campus Groenenborger, US.025
• ​​
• Supervisors: Joke Hadermann & Nick Schryvers
• Department of Physics

Abstract

Nuclear fusion is regarded as a promising long-term solution for meeting the rising global energy demand, while offering a potential for abundant, intrinsically safe and low-carbon emission electricity. However, achieving fusion energy on industrial scale requires overcoming several critical engineering and scientific challenges, including development of radiation-resistant plasma-facing component (PFC) materials to withstand fusion conditions. Tungsten (W) has emerged as the leading candidate for divertor and first-wall components in next generation fusion reactors due to its exceptional thermal properties. Nevertheless, under fusion-relevant operational conditions, PFCs undergo significant microstructural degradation, including defect formation, accumulation and transmutation-induced precipitation, all of which impact the mechanical stability and eventually lifetime of the component. A comprehensive understanding of these irradiation-driven mechanisms is essential to design, simulate and optimize durable fusion materials.

Transmission electron microscopy (TEM) is a powerful technique to visualize the microstructure of the inspected material. For nuclear science, it enables to reveal microscopic features, their characteristics and morphologies, induced by irradiation. Furthermore, combining with in-situ heating enables information about high-temperature stability of the radiation-induced defects. Despite extensive research efforts, critical gaps remain in literature regarding the microstructural evolution of PFCs under different irradiation conditions. The response of the material strongly depends on irradiation type (neutron, ion, plasma etc.), dose (dpa), temperature, initial microstructural state and alloy composition, yet systematic comparisons across irradiation conditions are limited. This thesis aims to address some of these gaps by developing a comprehensive dataset describing the defect evolution of PFCs under both neutron and heavy-ion irradiation across a range of irradiation parameters. The work focuses on the experimental investigation of the fundamental microstructural mechanisms such as defect formation, aggregation and recovery as well as understanding radiation-induced hardening and the role of transmutation products.

Jonas Lescroart

• 30/03/2026
  
• 5.30 p.m.
• Venue: Stads91�Թ���, Building R, room S.R.201
• Registration for physical access to public defence:
• Registration for livestream/online access to public defence:
• ​
• Supervisors: Hannes Svardal & Eduardo Eizirik
• Department of Biology

Abstract

This PhD project was a coordinated effort to scale up genomic research on the small cat species of the Neotropics (genus Leopardus). With the aid of many, we sourced DNA samples across 10 range countries in the Americas, sometimes resorting to museum specimens, and realized the novel sequencing of 45 complete genomes. For 12 taxa, whole-genome sequencing data simply did not exist at the time this project took off. This data is now accessible to the community in public repositories, hopefully to be reused many times over. ​ ​

Our time-scaled phylogenomic results reveal the sequence and timing of speciation events in genus Leopardus. Current diversity in the genus arose with two pulses of speciation, one in the Early Pliocene and another in the Early Pleistocene. We raise serious doubts regarding the margay-ocelot (L. wiedii/pardalis) sister species relationship and lay bare a deep paraphyly and multiple cryptic lineages within the tiger cat species complex. These results lead us to designate three subgenera, support species status for tiger cat populations in the northern Andes (L. pardinoides) and Caatinga ecoregion (L. emiliae) and suggest the existence of an additional species in the tiger cat complex. Furthermore, preliminary data supports multiple species-level taxa within the pampas cat species complex. ​ ​

We find that phylogenetic signal in Leopardus is highly discordant, that is, genetic similarities between species wax and wane depending on the precise location in the genome under investigation. We use this information to discern traces of introgressive hybridization from the echoes of incomplete lineage sorting. Hybridisation is detected between ocelot and the ancestral lineage of subgenus Oncifelis, between Geoffroy’s cat (L. geoffroyi) and the southern tiger cat (L. guttulus) and between Peruvian and Bolivian populations of tiger cat (L. tigrinus). Moreover, introgressed topologies are significantly associated with locally elevated rates of recombination across the genome, emphasizing the role of local recombination in modulating the effects of interspecific gene flow. ​ ​

Lastly, we present estimates of genomic diversity metrics, noticing that genetic diversity varies highly between species. Recomputing diversity across 39 felid species, we register the highest diversity (0.32%) in ocelot and the second to lowest (0.015%) in Andean cat (L. jacobita). Comparing current effective population sizes against historic sizes highlights sharp population declines especially in guigna (L. guigna) and clouded tiger cat (L. pardinoides), with alarming signs of a fragmented population in the eastern tiger cat (L. emiliae). ​ ​

Kumar Akhil

• 27/03/2026
  
• 2 p.m.
• ​
• Supervisor: Hans Van Haevermaet
• Department of Physics

Abstract

Since the first direct detection of gravitational waves in September 2015, gravitational wave astronomy has transformed from a theoretical pursuit into an observational science. Landmark events such as GW170817, the first binary neutron star merger observed in 2017, opened the era of multi-messenger astronomy and underscored the need for more sensitive detectors. To advance beyond the capabilities of Advanced LIGO and Virgo, third-generation observatories such as the Einstein Telescope (ET) are being designed. Achieving their target sensitivity requires extending the observation band to lower frequencies, which in turn demands unprecedented levels of seismic isolation, pushing residual ground motion at the mirrors to below the 10^{-20}m/sqrt{Hz} level. This is realised through multi-stage suspensions and it is the active control that ultimately stabilises the mirrors at the required level. At the heart of this controls are the precise sensors that measure vibrations, and actuators that apply corrective forces. The reliability of the entire isolation chain therefore depends critically on the design, performance, and integration of these sensing and actuation elements. This thesis focuses on these components in the context of the Einstein Telescope pathfinder (ETpf), a research facility in Maastricht to test the technologies for ET. A universal simulation pipeline was developed to model these sensors and actuators. Building on this modelling, complete production procedures for sensors were established. From design and winding to cleaning for ultra-high vacuum compatibility and final validation, a standard operating protocol was defined. A dedicated test bench was constructed at 91�Թ��� to characterise the transfer functions, noise spectra, and non-linear behaviour of the sensors, enabling direct comparison with model predictions. These studies demonstrate that the sensors achieve nanometre sensitivity at 10Hz. The validated sensors were subsequently integrated into the ETpf suspensions, where they form the backbone of the active control system. Complementary filter strategies were applied to blend the sensors, ensuring broadband control signals. Noise propagation studies confirmed that the intrinsic sensor noise is within the required limits. To push performance further, optimisation studies co-designed sensors and actuators, achieving improvements of more than 250\%. Novel concepts were explored, in particular a so-called reversed LVDT design. Both simulations and experimental measurements confirmed equivalent performance to the conventional design, with the potential advantage of simplified electronics through unified sensing and actuation.

Leonore Vander Donck

• 24/03/2026
  
• 5 p.m.
• Venue: Campus Drie Eiken, O.5
• ​​
• Supervisors: Sarah Lebeer & Veronique Verhoeven
• Department of Bioscience Engineering

Abstract

Women’s intimate health has long been constrained by scientific neglect, fragmented methodologies, and a limited understanding of the vaginal ecosystem. This dissertation addresses these gaps by reframing the vaginal microbiome as a dynamic, interconnected system shaped by both host biology and external factors. The vaginal microbiome, typically dominated by Lactobacillus species in health, plays a crucial role in preventing infections, supporting fertility, and promoting healthy pregnancies. Disturbances to this ecosystem remain an under‑recognized burden in women’s health worldwide. This PhD advances an ecological and integrative perspective on intimate microbial health. First, it demonstrates that the vaginal microbiome cannot be studied in isolation. Using cross‑body site analyses, the work reveals that vaginal lactobacilli disperse across the groin and breast skin in women, challenging traditional assumptions about body‑site specificity. This finding supports the existence of a connected female intimate microbiome shaped by hormones, hygiene practices, sexual activity, and physical proximity. Second, the dissertation shows that the stability of Lactobacillus crispatus‑dominated communities is driven primarily by ecological synergy and metabolic cooperation among microbes, rather than by host factors alone. Synthetic ecology approaches provided a powerful and reproducible platform to dissect these interactions and to evaluate how external influences, such as menstrual products, affect microbial dynamics. Finally, these ecological insights were translated into real‑world context through the Luna intervention study, one of the first to investigate how menstrual products shape vaginal microbiome trajectories. The study found that menstrual cups attenuate menstruation‑associated increases in microbial diversity, whereas external absorbent products are associated with shifts toward taxa linked to inflammation. These results highlight menstrual products as determinants of intimate microbial ecology rather than neutral consumer goods. Building on this, the dissertation proposes foundational steps toward standardized, microbiome‑relevant safety frameworks for menstrual products and intimate hygiene items. Together, this work contributes to a paradigm shift in women’s health research: from viewing the vaginal microbiome as a static niche to understanding it as a dynamic, interconnected ecosystem influenced by daily practices and structural inequities. The findings support more informed menstrual health guidance, the development of microbiome‑based interventions, and regulatory practices aligned with the Sustainable Development Goals for 2030, while helping to destigmatize menstruation and center women’s lived experiences in scientific inquiry.

Malaurie Hons

• 18/03/2026
  
• 4 p.m.
• Venue: Campus Drie Eiken, Q.002
• Supervisor: Jonas Schoelynck
• Department of Biology

Abstract

River restoration is a necessity for reversing biodiversity loss and securing flood safety in a climate with intensifying extremes. However, because ecological outcomes are often unpredictable and heavily influenced by catchment-scale stressors or extreme weather, simply altering a riverbank is rarely a guaranteed fix. Truly understanding the complex, non-linear responses of these ecosystems is the only way to move beyond localized "quick fixes" and build the resilient, self-sustaining river systems required for long-term environmental health. This thesis applied an integrative, process-based approach to evaluate short-term, multimetric ecological effects of reach-scale restoration in the Demer River (Belgium). Measures included meander reconnection, restoration of lateral connectivity, and in-stream structures such as bed sills and deadwood. A defining factor of the study was a severe summer flash flood that occurred shortly after restoration, serving as a critical "stress test" that influenced ecological trajectories and occasionally masked restoration-related signals.

Restoration rapidly altered flow and sediment dynamics, but outcomes were strongly design-dependent. Lateral connectivity proved critical: unfixed banks enabled sediment redistribution and habitat diversification, whereas bank fixation constrained channel development. Rather than resetting historical conditions, restoration initiated new, trajectory-dependent dynamics, also strongly impacted by hydrodynamics. Water quality improvements were modest and mostly transient. Aquatic communities showed no consistent river-wide recovery in multimetric indices, but did show local improvements based on age-class distributions and reproductive guilds, especially in the meander with additional in-stream measures. Fish recovered relatively quickly after flood-induced mortality, while macroinvertebrates and macrophytes exhibited prolonged declines. In contrast, riparian vegetation and carabid beetles responded more consistently, with increased species richness and shifts toward flood-tolerant specialists at sites were lateral connectivity was restored.

The findings demonstrate that restoration outcomes are shaped by design, hydrodynamic variability, and spatial scale, while an appropriate monitoring framework is essential for accurately interpreting ecological responses. Reach-scale interventions can effectively restore local hydro-geomorphic processes and habitat heterogeneity, particularly benefiting riparian biota. However, achieving sustained ecological recovery requires integration with catchment-scale management.

Caroline Dricot

• 12/03/2026
  
• 5 p.m.
• Venue: Campus Drie Eiken, O1
• ​​
• Supervisors: Sarah Lebeer & Irina Spacova
• Department of Bioscience Engineering

Abstract

Lactobacilli are well recognized as key members of healthy vaginal ecosystems, where they prevent pathogen colonization and engage in symbiotic interactions with both commensals and the host. Yet, the precise metabolic and immunological mechanisms underlying their beneficial functions remain incompletely understood. Traditionally, the protective effects of vaginal lactobacilli have been attributed to lactic acid-mediated pathogen exclusion; however, growing evidence suggests that many additional microbial metabolites play critical roles in shaping the vaginal niche. In other mucosal environments, including the gut and respiratory tract, microbial B vitamin metabolism, has emerged as an important mediator of microbiome stability, epithelial physiology, and immune signaling.

Riboflavin (vitamin B2) is of particular interest due to its antioxidative and anti inflammatory properties, its essential function as a metabolic cofactor in central metabolism, and the ability of its biosynthetic intermediates to activate mucosal associated invariant T (MAIT) cells via MR1. Moreover, (hidden) riboflavin deficiency disproportionately affects women due to fluctuating physiological demands linked to urogenital and reproductive health, the menstrual cycle, pregnancy, and breastfeeding. Despite of this, the role of microbial riboflavin metabolism in the vaginal environment had not been systematically explored. ​&�Բ�����;

This PhD thesis elucidates how riboflavin producing vaginal lactobacilli shape the immunometabolic landscape of the vagina, with emphasis on epithelial-immune crosstalk. Population scale multi omics analyses from the Isala citizen science project revealed that vaginal riboflavin levels are strongly associated with Lactobacillus crispatus abundance. Meta omic datasets further showed enriched expression of riboflavin biosynthesis genes in health associated, L. crispatus dominated profiles, with transcripts primarily derived from L. crispatus and L. jensenii. In vitro assays confirmed that riboflavin production is a widespread, ecologically relevant (though typically low level) trait among vaginal Lactobacillaceae. Genetic and metabolic characterization of Limosilactobacillus reuteri isolates identified a single nucleotide polymorphism in the RFN regulatory element that disrupts FMN mediated attenuation of the riboflavin operon, resulting in riboflavin overproduction and transient accumulation of MAIT antigens. Using a physiologically relevant 3D vaginal epithelial model we demonstrated that riboflavin is bioavailable to epithelial cells, basolaterally transported, and capable of modulating epithelial redox metabolism. Moreover, riboflavin producing lactobacilli seemed to affect epithelial MR1-MAIT signaling and trigger secretion of cytokines and tissue repair factors. Finally, the VIAB2L human intervention study-featuring oral and/or colon targeted delivery of riboflavin and riboflavin producing lactobacilli validates some of these in vitro findings.

Collectively, this work establishes microbial riboflavin metabolism as a conserved and functionally significant feature of vaginal lactobacilli, providing mechanistic insight into host-microbe symbiosis and laying the foundation for targeted, metabolism driven microbiome interventions.

Ben Huyge

• 26/02/2026
  
• 5 p.m.
• Venue: Campus Drie Eiken, O.01
• ​​
• Supervisors: Jan Sijbers & Jan De Beenhouwer
• Department of Physics

Abstract

Edge illumination (EI) is an X-ray phase contrast imaging technique with which three complementary image contrasts can be measured: attenuation, phase and dark field contrast. Unfortunately, extracting all three contrasts requires the acquisition of multiple images, causing EI scans to be several times slower compared to conventional X-ray imaging. The main goal of this thesis is to accelerate EI scanning via continuous acquisition, an approach where images are acquired continuously to avoid idle times of the scanning system.

To achieve this, a novel reconstruction technique for continuous acquisition is presented that can model any type of object motion. This technique is not immediately applicable to EI, because it only models object motion, while EI also has a moving mask. Next, continuous acquisition is introduced to EI, focusing on continuous mask motion with a stationary object. It is theoretically proven and experimentally validated that continuous mask motion does not influence the different contrasts. This allows for faster scanning without the need to change the model. Finally, the application potential of EI is demonstrated by extracting fiber orientations from fiber reinforced materials with spherical deconvolution. This allows to extract the orientations without needing to resolve the individual fibers. ​ ​

Miroslav Yosifov

• 23/02/2026
  
• 4 p.m.
• Venue: Campus Drie Eiken, N.008
• ​
• Supervisors: Jan Sijbers, Jan De Beenhouwer & Christoph Heinzl
• Department of Physics

Abstract

X-ray Computed Tomography (XCT) is an increasingly important tool for nondestructive testing, dimensional measurement, and inspection in industrial and scientific applications. Despite its widespread use, significant challenges remain in the reliable quantification of defect detectability and measurement uncertainty, particularly when inspections are complex, costly, or difficult to repeat experimentally. This thesis addresses these challenges by advancing probability of detection (POD) methodologies through simulation-driven approaches, enabling a more rigorous, systematic, and efficient framework for XCT-based inspection. The research is grounded in a detailed examination of X-ray imaging physics, system operation, and radiographic modeling using the X-ray simulation tool SimCT. By analyzing the influence of physical effects such as noise, acquisition geometry, focal spot blur, detector response, and modulation transfer function, the thesis establishes a comprehensive understanding of the factors that constrain image quality, defect detectability, and dimensional accuracy. This foundation is extended by a detailed consideration of POD theory, including POD curves, statistical modeling frameworks, and factors influencing detectability. Together, these methodologies enable an X-ray simulation-based approach to analyze both virtual and real XCT data.

The main contributions of the thesis focus on defect detectability, POD evaluation, measurement uncertainty, and XCT reliability. A simulation-based framework for applying POD to X-ray inspection is developed, demonstrating how virtual experiments can replace or complement costly test specimens. Using X-ray simulation, a controlled environment with artificial defects is generated, enabling systematic analysis of detectability under varied physical influences such as geometry, noise, and contrast. Building on this framework, traditional image segmentation approaches are compared with deep learning-based methods. Results obtained from simulated XCT datasets show that neural networks such as 3D U-Net and V-Net outperform classical algorithms in defect detectability and segmentation accuracy, supporting further automation in nondestructive testing.

Additional contributions include the application of simulation-empowered AI models to agricultural products, where XCT reconstructions of rice grains with controlled defects are analyzed through shape variation analysis to detect early degradation. Furthermore, a method for superimposing synthetic defects into measured XCT volumes is introduced, strengthening the link between simulated and real XCT data and enabling segmentation-based POD evaluation in a realistic yet controlled environment. Finally, the thesis investigates the influence of individual physical effects on dimensional measurement deviations, identifying dominant sources of dimensional error in XCT metrology.

Overall, the thesis demonstrates that X-ray simulations, deep learning, uncertainty analysis, and statistical methods can be integrated to advance POD methodologies for defect detection across multiple domains. ​ ​

Sarah Hautekiet

• 19/02/2026
  
• 4 p.m.
• Venue: Campus Drie Eiken, O.05
• ​​
• Supervisors: Stijn Temmerman & Maarten Kleinhans
• Department of Biology

Abstract

Tidal marshes are highly dynamic coastal ecosystems that provide essential ecosystem services such as shoreline protection and carbon sequestration. Hence, they belong among the most valuable ecosystems on our planet. However, centuries of anthropogenic disturbance, combined with ongoing climate change pressures, have led to widespread degradation of tidal marshes. This has created a growing interest for tidal marsh restoration. The aim of such practices is to create an intertidal ecosystem that displays equivalent characteristics and ecosystem functions as natural reference systems. However, restored tidal marshes often differ from their natural counterparts in this regard. As such, a thorough understanding of the mechanisms driving tidal marsh development is needed to improve outcomes in future restoration projects. The aim of this thesis was to advance our understanding of the abiotic and biotic mechanisms affecting tidal marsh development. Specifically, we focused on the development of two key features of tidal marshes (i.e. tidal creek networks and pioneer marsh vegetation) by investigating (1) the relative roles of abiotic and biotic processes in tidal creek network development, and (2) the mechanisms restricting or promoting pioneer vegetation establishment. In order to achieve this, we conducted scaled laboratory experiments in the Metronome tidal flume facility, a mesocosm experiment in the lab, field monitoring, field experiments and a field mesocosm experiment, and geospatial analyses of marsh development in restoration sites along the Scheldt estuary. Results demonstrate that while vegetation can influence tidal creek network density through bio-geomorphic feedbacks, abiotic drivers are the dominant controls on creek incision. Vegetation primarily modifies creek networks during the transition from bare mudflat to vegetated marsh, and the rate of plant colonization exerts a stronger influence on network development than spatial colonization patterns. Vegetation establishment on bare tidal mudflats was found to be strongly constrained by low intertidal elevation, rapid sediment accretion, and poor sediment drainage - conditions typical of recently restored sites. Well-drained areas, such as tidal creek edges, promote erosion-resistance, creating favorable conditions for seed retention and seedling survival. In addition, three-way interactions between filamentous algae, macrobenthos, and pioneer plants were shown to play a crucial role in plant establishment. Together, these findings highlight that tidal marsh development emerges from tightly coupled abiotic and biotic processes. Improving drainage, enhancing topographic diversity, and leveraging facilitative biotic interactions offer promising pathways to overcome key bottlenecks in marsh restoration and accelerate the transition from bare mudflat to resilient vegetated marsh.

Wouter Van Hoey

• 09/02/2026
  
• 5 p.m.
• Venue: Campus Middelheim, G.010
• ​
• Supervisor: Pegie Cool
• Department of Chemistry

Abstract

At present, the majority of all intricate global challenges are related to sustainability. Therefore, the urge for sustainable development is bigger than ever. To date, the chemical sector is already a major driving force for innovation and essential for sustainable growth across all sectors. However, much more scientific contributions will be needed to tackle the difficult task ahead, which is to create a sustainable world whereby the needs of the present are met without compromising the ability of future generations to meet their own needs. Hence, the aim of this thesis is to make a positive contribution to sustainability by applying heterogeneous catalysis in various green chemistry related applications. The research in this PhD thesis is mainly focussed on the development of innovative CeO2-based catalysts and can be divided into four different topics: catalytic combustion of VOCs, plasma-catalytic conversion of CO2, photo-induced reduction of nitrobenzene to aniline, and the selective hydrodeoxygenation of aromatic carbonates. In more detail, chapters 2, 3 and 4 focus on catalyst optimisation to improve the catalytic combustion of toluene, which is a model compound for aromatic VOCs. Based on the obtained results it can be stated that combining noble and transition metals offers a unique approach to limit the costs and consumption of noble metals for industrial large-scale combustion of VOCs. Unfortunately, while intriguing results are obtained regarding the catalytic combustion of VOCs, it cannot be ignored that combustion of carbonaceous compounds results in the production of CO2. Therefore, the fifth chapter tackles the plasma-catalytic conversion of CO2 using CeO2. More specifically, it is investigated if increasing the amount of oxygen vacancies at the surface of CeO2 enhances the dissociation of the stable CO2 molecule, which facilitates the conversion of CO2 into value-added products. Next, in chapter 6 the development of a more sustainable photochemical process to convert nitroarenes into anilines at room temperature and in open atmosphere without the addition of hydrogen is discussed. Additionally, it is examined if the presence of a heterogeneous catalyst can further improve the obtained results. Finally, in chapter 7 a thorough characterisation is provided to gain a better understanding on the selective hydrodeoxygenation of aromatic carbonates when commercially available heterogeneous Ni-SiO2 catalysts are applied. Unravelling the reason behind the remarkable activity is trivial as the successful hydrodeoxygenation provides a next step in the development of important bio-based compounds from wood.

Burak Karaduman

• 05/02/2026
  
• 10 a.m.
• Venue: Campus Middelheim, G.006
• Supervisor: Moharram 91�Թ���r
• Department of Computer Science

Abstract

As embedded system-originated paradigms grow in complexity and lack of autonomy, traditional control architectures struggle to manage the uncertainty and dynamism of real-world environments. Intelligent agent architectures, particularly those grounded in the Belief–Desire–Intention (BDI) model, offer promising cognitive capabilities for Cyber–Physical System (CPS). However, their integration into resource-constrained and (soft) real-time embedded platforms remains challenging. To address this, eventually, we propose a Model-driven Engineering (MDE) approach for deploying fuzzy-BDI agents into CPS. This approach combines fuzzy logic with BDI reasoning to manage imprecision and context shifts, while also enabling platform-independent system design through model-based abstraction. Despite the growing relevance of such hybrid architectures, existing literature lacks comprehensive methods that link cognitive agent behaviours to deployable embedded code in complex systems settings. This thesis fills this gap by introducing an integrated approach that bridges high-level agent modelling and low-level system implementation. The results demonstrate that fuzzy-BDI agents, supported by model-based analysis and engineering, can enhance adaptability, reduce development complexity, and improve robustness in uncertain environments. The central challenge addressed in this thesis lies in bridging the gap between traditional embedded control and intelligent decision-making within CPS. While intelligent agent architectures, particularly those grounded in the BDI model, provide a promising cognitive framework for autonomous reasoning, their practical integration with enhanced capabilities into resourceconstrained embedded platforms remains an unresolved problem. The BDI model enables agents to reason about their environment, goals, and actions proactively, maintaining beliefs about the world, forming desires representing objectives, and committing to intentions that guide action. However, classical BDI systems are based on crisp logic, which severely limits their applicability under uncertainty and continuous change. Without mechanisms to interpret ambiguous sensory data or adapt to fluctuating contexts, BDI agents fall short of the robustness required in CPS deployments. To address these shortcomings, this thesis introduces an integrated fuzzy-BDI and MDE framework for the development and deployment of intelligent CPS. The core idea is to combine fuzzy logic, which allows reasoning with degrees of truth, with the structured deliberation of BDI agents, thereby enabling systems to handle uncertainty natively at every stage from perception and planning to action execution. Complementing this reasoning innovation, the research employs MDE to elevate the abstraction level of system development. The proposed framework bridges high-level agent design models with low-level embedded implementations, achieving explainability, reusability, and platform-independent deployment. This dual integration of fuzzy and BDI reasoning with model-driven design forms the foundation of a new engineering methodology for building adaptive and smart CPS.

Arno Raes

• 02/02/2026
  
• 4 p.m.
• Venue: Stads91�Թ���, Klooster van de Grauwzusters, Promotiezaal
• Supervisor: Sammy Verbruggen
• Department of Bioscience Engineering

Abstract

Titanium dioxide is widely used in photocatalysis because it is abundant, stable, and inexpensive, yet pristine anatase and rutile absorb mainly ultraviolet light. One strategy to extend activity into the visible region is the introduction of sub-gap states associated with oxygen vacancies and Ti³⁺ species. These are typically generated through hydrogenation, high-temperature reduction, plasma treatment, or strong chemical reductants, routes that often raise energy concerns. This thesis therefore addresses a practical question: can TiO2 be reduced by mild and potentially scalable methods that still deliver useful functionality under realistic conditions, and where do such approaches succeed or fail?

High-intensity ultrasound was first investigated as a potential reduction route. Using calorimetrically calibrated power delivery and systematic probe integrity checks, any apparent darkening of TiO2 during aqueous sonication was traced to metallic debris from horn erosion rather than defect formation. UV–Vis diffuse reflectance showed no band-edge shift or sub-gap absorption, and no evidence for stable oxygen vacancies or Ti3+ species was found. Under these conditions, cavitation hot spots do not provide a sustained reducing environment, and any transient defects are rapidly re-oxidised by radical species and dissolved oxygen. Ultrasound-induced “blackening” is therefore an artefact rather than a viable reduction strategy.

Despite this negative outcome, ultrasound proved valuable as a synthesis and processing tool. During sol–gel synthesis it yielded high-surface-area, open-mesoporous amorphous TiO2 with interconnected nanoparticulate networks. Controlled ultrasonic crystallisation produced predominantly anatase while retaining about 50% of the surface area, compared to the 10% of conventional calcination. These materials showed rapid uptake of volatile organic compounds, consistent with their preserved mesostructure.

Vacuum annealing was then explored as a hydrogen-free route to defective TiO2. Mild vacuum treatment of P25 darkened the powder, increased visible-light absorption without shifting the band edge, and shifted the Ti:O ratio away from perfect stoichiometry. Functionally, this resulted in an approximately 25% increase in methane formation during CO2 photoreduction compared with pristine P25. Applying the same treatment to Au@P25 preserved plasmonic properties but provided only marginal additional activity beyond that already conferred by gold.

Overall, this work shows that mild routes give mild but reliable outcomes. They are effective for preserving structure, adsorption capacity, and stability, and can deliver meaningful functional gains in selected cases.

Arthur Gestels

• 02/02/2026
  
• 2 p.m.
• Venue: Campus Middelheim, A.143
• ​
• Supervisors: Koen Janssens & Gunther Steenackers
• Department of Physics

Abstract

Scientific research in cultural heritage has increasingly shifted toward non-invasive imaging methods that can reveal material information across entire artworks. Because paintings often exhibit complex stratigraphies and heterogeneous compositions, traditional point-based techniques and small-scale sampling frequently fail to provide representative insight into the full material structure of an artwork. In this context, hyperspectral imaging techniques have become essential tools for conservators, researchers, and art historians.

Macroscopic X-ray powder diffraction (MAXRPD) is one such hyperspectral technique. It enables direct identification of crystalline compounds, making it highly valuable for the study of pigments and degradation products in painted artworks. Despite its high chemical specificity, MAXRPD is limited by low spatial resolution and long acquisition times, restricting its applicability for large-scale investigations. To address these limitations, this thesis investigates the combination of MAXRPD with faster hyperspectral imaging modalities, including reflectance imaging spectroscopy (RIS) and macroscopic X-ray fluorescence (MAXRF), using machine learning approaches to reduce acquisition time and improve efficiency. ​ ​

The research begins with a case study on Rembrandt’s The Night Watch, where MAXRPD was used to assess the material impact of an acid vandalism attack. The results revealed the formation of anglesite as a crystalline degradation product in the upper paint layers, demonstrating the value of compound-specific imaging for conservation assessment and treatment planning.

Next, a proof-of-concept study on corroded metal plates showed that quantitative MAXRPD data could be combined with RIS data to train machine learning models capable of predicting material distributions with reasonable accuracy. This demonstrated the feasibility of extrapolating material-specific information from faster imaging techniques.

The methodology was then adapted to cultural heritage objects, enabling the generation of high-resolution, material-specific images of artworks. Case studies on an illuminated manuscript and oil paintings showed that fusing MAXRPD with RIS or MAXRF data produces reliable predictions. Finally, the approach was applied to more complex historical paintings to assess its generalizability, demonstrating that accurate compound mapping is possible even in unscanned areas when representative training data are available. Together, these results establish a scalable framework for accelerating MA-XRPD-based imaging while preserving its analytical power.

Donja Baetens

• 30/01/2026
  
• 2 p.m.
• Venue: Campus Drie Eiken, O.5
• Supervisor: Siegfried Denys
• Department of Bioscience Engineering

Abstract

Indoor air quality (IAQ) is increasingly recognised as a major public health concern, as people spend most of their time indoors and are therefore continuously exposed to pollutants present in indoor environments. Among indoor contaminants, particulate matter (PM) and volatile organic compounds (VOCs) are of particular concern due to their health impacts and prevalence in buildings. While air purification technologies can reduce exposure, most systems target only one type of pollutant, i.e. particulate or gaseous. This dissertation explores the integration of electrostatic precipitation (ESP) and photocatalytic oxidation (PCO) into a single device to enable simultaneous removal of PM and VOCs. It is hypothesised that incorporating PCO into the collector section of a two-stage ESP reactor enables simultaneous removal and may reduce the need for cleaning the ESP by degrading deposited particles.

A combined ESP-PCO reactor was developed with a photocatalytic coating immobilised on the collector plates and UV lamps positioned between them, alongside multiphysics models describing the underlying processes. First, the standalone ESP functionality was investigated to determine total and fractional particle collection efficiencies, the influence of operating conditions, and ozone emission. Next, the photocatalytic coating was examined independently for photocatalytic VOC (acetaldehyde) removal, comparing metallic substrates and studying the effects of coating layers, inlet concentration, and relative humidity. Photocatalytic soot degradation experiments were also performed to evaluate removal of deposited PM.

Subsequently, acetaldehyde removal was studied in the full two-stage ESP-PCO reactor to assess PCO and ESP functionality in terms of clean air delivery rate (CADR) and single-pass removal efficiency (SPRE). Activation of the ionisation section resulted in acetaldehyde removal comparable to photocatalytic removal, while combining ESP and PCO yielded the highest removal observed, although synergistic effects were not clearly identified.

Finally, multiphysics models for ESP and PCO were developed to describe air flow, electric field, ion transport, particle charging and motion, acetaldehyde transport, radiation distribution, and photocatalytic kinetics. The models were verified against experiments and provide insight into the influence of plate spacing, irradiance distribution, and electric field effects while identifying opportunities for performance optimisation.

Overall, this dissertation demonstrates the feasibility of integrating photocatalytic oxidation into a two-stage electrostatic precipitator for indoor air purification, showing that the combined system can remove both PM and VOCs and has the potential to reduce particle accumulation. However, the results highlight design, material, and modelling challenges requiring further development.

Jan Morez

• 29/01/2026
  
• 4 p.m.
• Venue: Campus Drie Eiken, O.005
• ​​
• Supervisors: Jan Sijbers & Ben Jeurissen
• Department of Physics

Abstract

Diffusion magnetic resonance imaging (dMRI) offers a unique way of imaging the human brain noninvasively. By carefully controlling various acquisition parameters, the dMRI signal can be used to probe the movement of water molecules inside biological tissues, revealing important information about tissue structure. However, this signal is subject to noise and other undesired electromagnetic effects that reduce the signal-to-noise ratio. Efficient acquisition schemes and robust estimators are required to obtain accurate and precise tissue maps of the human brain. In this thesis, we have aimed our efforts towards improving the accuracy, precision, generalizability and applicability of several dMRI analysis methods. ​ ​

Constrained spherical deconvolution (CSD) is a popular dMRI technique that can be used to infer the local tissue densities and their orientations in the human brain, which consists of cerebrospinal fluid, anisotropic white matter and isotropic gray matter. This is achieved by densely sampling q-space, the space of q-vectors representing the direction and strength of diffusion weighting. After sampling q-space in multiple shells, tissue densities and orientations can be estimated by deconvolving this multi-shell dMRI signal with the respective response functions for white matter, gray matter and cerebrospinal fluid. These response functions are typically represented using spherical harmonics (SH) basis functions. However, when sampling is nonspherical, either due to inhomogeneous gradients or by design such as Cartesian sampling, the estimated tissue maps suffer from biases. To counter this issue, we adopted a compact response function model that accounts for nonspherical sampling. On multi-shell data, our approach provides fiber orientation density functions and tissue densities indistinguishable from those estimated using SH. On Cartesian data, estimates are on par with those obtained from shell-wise data, significantly broadening the range of data sets analyzable using CSD. In addition, inhomogeneous gradients can be accounted for, resulting in more accurate apparent tissue densities and connectivity metrics. ​ ​

Q-space trajectory imaging is a dMRI technique that uses time-varying q-vectors to sensitize the dMRI signal to microscopic variations in heterogeneous tissues. By modelling the dMRI signal with a diffusion tensor distribution (DTD), this approach allows teasing apart variations in diffusivity from microscopic anisotropy, orientation dispersion, and mixtures of multiple isotropic diffusivities. To improve the estimation of the DTD parameters, we propose an efficient acquisition scheme optimized for the most used QTI-derived microstructural parameters. A constrained iteratively reweighted least squares estimator is used to further improve the bias and precision of the DTD parameters.

Jonathan Sanctorum

• 13/01/2026
  
• 4 p.m.
• Venue: Campus Drie Eiken, Q.001
• ​
• Supervisors: Jan Sijbers & Jan De Beenhouwer
• Department of Physics

Abstract

X-ray phase contrast imaging is known for its potential to yield high contrast in soft tissue and light materials compared to conventional transmission contrast, making it a valuable tool for (bio)medical applications and non-destructive testing. In addition to transmission and phase contrast, dark field contrast is a third contrast type that has received increased interest recently due to its unique ability to detect the presence of unresolved microstructures. Phase and dark field contrast are both related to the material’s X-ray refractive index, but as opposed to phase contrast, dark field contrast results from refractive index fluctuations that cannot be resolved by the imaging system. Remarkably, all three contrast types can be measured in a single experiment using dedicated X-ray phase contrast imaging systems. The introduction of compact lab-based systems for phase sensitive X-ray imaging has resulted in a growing number of applications. Optimization of conventional X-ray imaging setups often relies on Monte Carlo simulations, where stochastic models are used to simulate the physics processes. In this work, a collection of tools is presented that aim to meet the requirements associated with Monte Carlo simulations for the design of X-ray phase contrast imaging systems. Here, two X-ray phase contrast imaging methods are mainly of interest: grating-based interferometry and edge illumination, each of which relies on the use of gratings. First, the realization of X-ray phase contrast simulations for grating-based interferometry is demonstrated using the GATE Monte Carlo framework, hereby relying on a hybrid simulation approach combining Monte Carlo simulations with wave optics calculations. Although this simulation framework is suitable for edge illumination simulations as well, Monte Carlo simulations are known to be very time consuming, certainly for parameter studies. To address this limitation, the concept of virtual gratings is introduced. By replacing the gratings in the simulation with virtual gratings, the parameters of the gratings can be changed after the simulation, thereby significantly reducing the overall simulation time. This concept is subsequently used for the design of edge illumination gratings for the augmentation of the FleXCT micro-CT scanner to a phase sensitive X-ray imaging system. Finally, the benchmarking of multi-contrast X-ray imaging simulations is addressed. The reference values required for benchmarking are particularly difficult to determine for the dark field contrast. Here, a practical method based on the virtual grating approach is presented to directly estimate reference values for all three contrast types from the simulated X-ray trajectories, allowing for efficient benchmarking.

Salma Haidar

• 13/01/2026
  
• 4 p.m.
• Venue: Campus Middelheim, A.143
• ​
• Supervisor: José Oramas Mogrovejo
• Department of Computer Science

Abstract

Hyperspectral imaging (HSI) combines digital imaging with spectroscopy, capturing hundreds of contiguous wavelength bands per pixel. This produces a three-dimensional “hypercube” that integrates spatial and spectral information, offering powerful opportunities for material analysis, environmental monitoring, and beyond. Yet, HSI also poses challenges: high dimensionality, limited annotated data, and heavy computational demands.

My thesis addresses these challenges through three complementary approaches. First, I develop a deep learning framework for multi-label classification that demonstrates strong performance in complex, mixed-material scenes and highlights limitations in common annotation practices. Second, I explore self-supervised contrastive learning to enhance classification accuracy under limited supervision, demonstrating substantial gains across diverse datasets. Third, I apply explainability-driven dimensionality reduction to identify the most informative spectral bands, reducing redundancy while maintaining or even improving accuracy.

Together, these contributions demonstrate that tailored representation learning strategies and explainability-driven dimensionality reduction can deliver hyperspectral classifiers that are accurate, computationally efficient, and adaptable to challenging data conditions. The results highlight practical pathways for improving hyperspectral image analysis and open opportunities for further exploration across diverse application domains.