[diss] Perustieteiden korkeakoulu / SCI

Permanent URI for this collectionhttps://aaltodoc.aalto.fi/handle/123456789/52

Browse

Recent Submissions

Now showing 1 - 20 of 1604
  • Neurophysiological methods for evaluation of infants’ sensory processing
    (2025) Ahtola, Eero
     School of Science | Doctoral dissertation (article-based)
    Early infancy is a critical period for brain development, during which medical adversities can have lifelong consequences. Neurodevelopmental disorders often first manifest as sensorineural defects. However, assessing these in infants is challenging, as tests designed for older children are not yet suitable for developing babies. Therefore, there is a pressing need for innovative diagnostic tools for objective evaluation of sensory processing, offering insights into developing brain function and facilitating early interventions and care. This Thesis presents new technical approaches to assess infants’ sensory systems by characterizing electroencephalography (EEG) responses to visual and somatosensory stimulation. Additionally, eye tracking was employed in a behavioral task to study the mechanisms of visual attention and face preference for various facial expressions. The research incorporated advanced computational techniques—such as source reconstruction and connectivity analysis—integrated with task designs that systematically targeted specific aspects of sensory processing. In Study I, healthy 5- and 7-month-old infants were examined in a face-distractor competition paradigm, utilizing newly developed dynamic eye tracking metrics. Results revealed age-specific differences in gaze disengagement from faces, especially those with fearful expressions, indicating sensitivity to developmental changes. Studies II and III assessed cortical visual processing in healthy 3–5-month-old infants using cyclic orientation reversal, global form, and global motion stimuli. Eye tracking integrated with the visual stimulus presentation enabled detection of occipital steady-state EEG responses in 92–100% of the recordings, substantially exceeding detection rates reported in the literature. Connectivity analysis of these responses revealed recruitment of widespread functional networks extending also beyond the traditional visual streams, with a global form-related network metric correlating positively with later cognitive performance. Study IV examined proprioceptive EEG responses in asphyxiated neonates during repetitive hand movement stimulation. All neonates produced contralateral responses with vast cortico-cortical recruitment, though the extent of activation was reduced in cases with hypoxic-ischemic encephalopathy. Overall, the studies demonstrate significant potential of the presented methods for early assessment of sensory processing and related neurocognitive development. However, high individual variability suggests that these tools may not yet be ready for clinical use. Further research is still needed with larger and more heterogeneous cohorts.
  • Toward an Integrated Leadership Model
    (2025) Äkräs, Juha
     School of Science | Doctoral dissertation (monograph)
    Leadership, an enduring subject of inquiry since antiquity, stands as a complex and multifaceted phenomenon in the current era. Although there are many theories and definitions, scholars critique the field for its ambiguity, suggesting that current conceptions of leadership can encompass either everything or nothing. Such critique highlights a need for new theorization that adopts collective and interaction-centric perspectives over traditional leader-centric models. Additionally, a shift toward systemic and integrated approaches has been advocated. Addressing these critiques, the aim of my study, titled “Toward an Integrated Leadership Model,” is to develop and test an integrated leadership model. My dissertation presents the development of the Integrated Leadership Model (ILM), a comprehensive conceptual framework synthesizing the antecedents of leadership (leader traits, leadership structure, and leadership context), leadership behaviors (leadership functions and styles), leadership mediators (psychological mechanisms), and leadership outcomes (task and socio-emotional effectiveness). The model emphasizes the interconnectedness of these concepts, thus reflecting the dynamic and contextual nature of leadership. To empirically assess the partial operationalization of ILM, I conducted a study with 883 knowledge workers and managers from seven diverse Finnish organizations, employing a measurement instrument designed for this purpose. Despite limitations such as geographic scope and the inherent challenges in quantifying complex leadership phenomena, the study underscores the ILM’s potential to refine our understanding of leadership, addressing the field’s prevailing issues of ambiguity and conceptual overlap.
  • Use cases of emerging digital technologies in operations management—exploring purposes and development approaches
    (2025) Jaribion, Alireza
     School of Science | Doctoral dissertation (article-based)
    Emerging digital technologies (EDTs) promise to transform operations management (OM) practices by offering new ways to analyze, automate, and optimize. However, assessing the opportunity to use EDTs remains challenging due to their ongoing development and yet-to-be-realized practical applications. The existing literature suggests that presenting use cases is crucial for both the adoption and development of EDTs. Despite the significance of use cases in information systems, the approach is underdeveloped in OM. Although some studies have highlighted the use case as a concept, presented methodological frameworks, and discussed theoretical perspectives, literature falls short in elaborating on the purposes and development of EDT use cases in OM. This doctoral dissertation, drawing on insights from five research articles, seeks to bridge this gap by examining why and how to develop use cases for EDTs in OM. The dissertation summary presents a framework for the development and assessment of EDT use cases, grounded in a review of the literature on the challenges in assessing the use of EDTs, and use case development of specific EDTs in OM presented in the research articles. The framework is applied to examine and analyze the use cases for three EDTs—digital twins, additive manufacturing (AM), and blockchain—presented in the research articles. We suggest that developing use cases alleviates specific challenges in assessing the value of EDTs in OM, including timing and rationale for adoption, contextual application, technological capability and requirements, regulatory considerations, and implications of use. Moreover, the analysis of the research articles highlights that use case development indicates the stage of the EDT hype cycle, facilitating the identification of the over-optimistic stage of “wide-ranging use”, the over-pessimistic “limited use”, and the more realistic “actual use”. Furthermore, in terms of specific methods for use case development, this dissertation summary reviews established practices such as experimentation, prototyping, and expert consultation, before introducing hackathons as an emerging and promising method for EDT use case development in OM, investigated and applied in two of the research articles. The research articles included in the dissertation make technology-specific contributions by introducing digital twin use cases for building lifecycle management and risk management in hydrogen storage and transportation, AM use cases for the emergency production of medical supplies during disruptive events such as pandemics, and a blockchain use case for protecting intellectual property rights (IPR) when sharing digital designs in AM. For operations managers, the framework proposed in this dissertation summary serves as a valuable tool for guiding the development and assessment of EDT use cases. Additionally, this dissertation introduces hackathons as a method to help operations managers develop use cases that are specifically tailored to their business needs and real-world problems, rather than merely following industry trends.
  • Real-space observation of non-covalent interactions in planar and non-planar molecules using scanning probe microscopy
    (2025) Choi, Shukning
     School of Science | Doctoral dissertation (article-based)
    With the rapid advancement of supramolecular materials, the study of non-covalent interactions (NCIs) has become increasingly critical. These interactions are fundamental to the forming, stability, and functionality of biological and synthetic materials. Scanning probe microscopy (SPM) offers an unparalleled platform for investigating NCIs within low-dimensional supramolecular self-assemblies on surfaces, allowing direct visualization and overcoming ensemble-averaging errors. The emergence of electrospray deposition (ESD) in SPM has significantly broadened its applicability to non-volatile, fragile biomolecules. However, accurately imaging and confirming the precise 3D structures of such molecules remains a considerable challenge. Addressing this difficulty requires the development of efficient structure search and reconstruction methods, which are still relatively underexplored. This thesis investigates the structural and functional implications of NCIs in low-dimensional supramolecular nanostructures on surfaces, progressing from planar molecules to non-planar small molecules and flexible biomolecules. This systematic approach demonstrates how to overcome the increasing complexity of the systems studied. The discussion begins with planar self-assemblies of DNA base molecules, which create dynamic confined environments that stabilize water dimers, revealing their influence in causing mismatched hydrogen bonding. It then advances to integrating machine learning (ML) to predict the structures of disordered water nanoclusters on metal surfaces, enabling the determination of complex hydrogen-bond patterns. Finally, the investigation focuses on glycosidic bond stereochemistry in carbohydrate self-assemblies, highlighting its critical role in regulating the charge distribution of groups attached to the non-anomeric carbon and governing chirality transfer in self-assemblies, using data-efficient, multi-fidelity structure search methods integrated with ML. Through these studies, this thesis provides insights into addressing the challenges posed by increasingly complex systems, paving the way for more efficient methods to study NCIs and their impact on supramolecular materials at atomic resolution.
  • Software engineering curriculum change in University of applied sciences: perspectives of teachers, students and graduates
    (2025) Vesikivi, Petri
     School of Science | Doctoral dissertation (article-based)
    The field of software engineering continually shapes university curricula to meet industry and societal demands. In response to the growing need for 21st century competences, a project-based curriculum is recommended as an effective way to develop these competences. Project-based learning has been shown to foster crucial professional competences and practical experiences, preparing students for the challenges of their profession. CDIO (Conceive, Design, Implement, Operate) is a notable approach that aims to develop technical knowledge and leadership abilities in engineering students. The significance of competences in communication and teamwork has increasingly become a priority as recruiting criteria for potential employers. Project-based learning serves as a primary method for enabling students to acquire these essential competences during their studies. Metropolia University of Applied Sciences offers education in fields such as Business, Technology, Arts and Health Care. In 2014, a new project-based curriculum was deployed in Metropolia with the aim of providing students opportunities to acquire not only professional knowledge, but also professional competences. Additionally, the target was to improve retention as the new funding model was tied to student success. In autumn 2014, the first IT student groups started their studies according to the new project-based curriculum. The new IT curriculum consisted of 15 ECTS modules taught by a multidisciplinary teacher team. This dissertation examines curriculum change by using the perspectives of students, teachers, and graduates. It comprises three journal papers focusing on student experiences, retention, graduate competence perceptions, and teachers’ views on team teaching, as well as two conference papers and one book chapter. The study results suggest that project-based learning significantly improved first-year retention rates. Graduates reported acquiring essential professional competences such as teamwork, communication, and project work. Furthermore, they felt they had acquired adequate technical knowledge and skills during their studies. Teachers expressed concerns about losing autonomy, especially those not involved in team teaching. The thesis concludes by offering ideas for universities planning to transition to a project-based curriculum based on the results and analysis.
  • Different is more - Selected works in heterostructures of transition metal dichalcogenides
    (2025) Huang, Xin
     School of Science | Doctoral dissertation (article-based)
    Heterostructures are composite materials composed of different parts put together, and they are designed to bring artificial properties. Especially after the discovery of graphene in 2004 and later other two-dimensional layered materials, the idea of assembling heterostructures like toy bricks came true and became a daily routine in quantum material research. These studies continue as many heterostructures are waiting to be realized and direct synthesis and transfer technology are expecting optimization and upgrading. On the other hand, people can now expect to use simple materials and easy protocols to reproduce strongly correlated quantum phenomena shown before in complex compounds with multiple components, such as unconventional high-temperature superconductivity in cuprates, heavy fermions and quantum criticality in rare earth compounds. The simple material platform saves energy for reproducing and reduces the difficulty of analysis. The work here chooses transition metal dichalcogenides with a layered structure as the building blocks. Together with heterostructure technology, they bring new opportunities and new insights to these challenges. This booklet tries to answer the question of what more physics will emerge when assembling different materials together, by presenting several examples, as well as necessary experimental and theoretical tools. It first briefly introduces all necessary tools, e.g., the materials transition metal dichalcogenides, molecular beam epitaxy, scanning tunneling microscopy and spectroscopy tech, as well as theories such as the Hubbard model, the single impurity Anderson model and the Kondo model. Together with these techs in the odyssey of searching new quantum phases, some key findings are reported, e.g., a doped Mott insulator in the vertical heterostructure 1T/1H-NbSe2, where the 1T-phase gives Hubbard model physics and that 1H-phase acts as a charge transfer dopant; 1D interfaces in 2D lateral heterostructures 1T-VSe2—1H-NbSe2, with side-coupled Kondo resonance where VSe2 provides localized magnetic moments and NbSe2 provides conduction electrons; and finally other heterostructure applications. These compositions of different components effectively and seamlessly bring new properties which cannot be found in solo materials. These experiments successfully explore new synthesizing methods and discover new condensed matter platforms for many-body physics. These different materials, different Hamiltonian terms bring more phases and more physics.
  • Lattice models and conformal field theory
    (2025) Adame-Carrillo, David
     School of Science | Doctoral dissertation (article-based)
    Within the realm of Mathematical Physics, this thesis studies the connection between statistical mechanics and conformal field theory (CFT) in two dimensions. More precisely, we contribute to the mathematical understanding of the scalinglimit convergence of critical discrete models to conformal field theories. In two dimensions, the conformal symmetries of the theory are encoded into its space of local fields in an algebraic manner: it constitutes a representation of the Virasoro algebra. In this thesis, this algebraic feature is exploited to establish the emergence of conformal field theories in the scaling limit of critical models. The key insight that enables our results is that, in critical models with enough integrability, the relevant observables can be proven to carry the algebraic structure of a two-dimensional CFT by means of tools of discrete complex analysis. In Publication I, the discrete model of interest is the (double) dimer model. Within this model, we consider certain fermionic observables whose correlation functions present suitable discrete holomorphicity. The main result can then be stated as follows. The space of the fermionic observables carries a representation of the symplectic-fermions algebra and, furthermore, it constitutes a Virasoro representation with central charge −2.In Publication II, we establish, for the first time, the scaling-limit convergence of a discrete model to a fully-fledged CFT. The space of local observables of (the gradient of) the discrete gaussian free field (DGFF) is proven to be in one-toone correspondence with the space of local fields of the free boson CFT. Then, the (suitably renormalised) correlation functions of DGFF local observables are proven to converge to the CFT correlation functions of the corresponding local fields in the scaling limit. In Publication III, the symplectic fermions CFT in general domains of the complex plane is expounded in detail. The main motivation for considering this logarithmic FT is the scaling limit of the observables introduced in Publication I.
  • Deep learning methods for modeling of spatiotemporal dynamical systems governed by partial differential equations
    (2025) Iakovlev, Valerii
     School of Science | Doctoral dissertation (article-based)
    This dissertation focuses on data-driven modeling of spatiotemporal dynamical systems, using observational data to develop models that approximate the underlying dynamical processes. Spatiotemporal modeling has a rich history with numerous successful applications. It has been continually advanced by technological and methodological improvements, evolving from early qualitative approaches to modern sophisticated deep learning methods. Despite recent progress enabled by deep learning—which has shown promise in modeling complex systems like weather patterns, traffic dynamics, and crowd flows—current deep learning-based spatiotemporal models face significant limitations. These include restricted applicability due to simplifying assumptions (such as fully observed states on fixed grids), data inefficiency requiring large datasets for good generalization, and long training times coupled with instabilities arising from complex loss landscapes. This dissertation addresses these challenges by developing novel deep learning-based models and techniques that enhance the flexibility, data efficiency, and stability of spatiotemporal systems modeling. To extend applicability of deep learning-based spatiotemporal models, a graph-based continuoustime model inspired by the method of lines is introduced, enabling modeling on irregular spatiotemporal grids. This is further extended to a space-time continuous model operating in latent space, allowing for learning dynamics from partially observed and noisy states. Finally, a model incorporating a spatiotemporal point process is developed to learn system dynamics from unstructured observations made at random times and locations. To improve data efficiency, the models leverage the locality bias inherent in PDE systems, achieving remarkable data efficiency and requiring significantly fewer training trajectories to generalize compared to previous methods. To enhance training stability and speed, an amortized Bayesian multiple shooting technique is proposed, extending classical multiple shooting to the Bayesian setting and modern computational regimes. This method stabilizes training and reduces training time by up to an order of magnitude. Additionally, a latent space interpolation technique is introduced to further accelerate training without compromising predictive accuracy. Overall, this dissertation advances the field of data-driven spatiotemporal modeling by introducing deep learning methods and techniques that are more widely applicable, data-efficient, and computationally efficient. These developments enable the modeling of a broader spectrum of complex dynamical systems under more realistic conditions than was previously possible.
  • Managing Complex Patient Journeys: A Healthcare Operations Management Perspective
    (2025) Vesinurm, Märt
     School of Science | Doctoral dissertation (article-based)
    The single most important challenge of modern healthcare operations management is that of the complex patient journey. Modern healthcare systems observe an increasing trend of healthcare use and costs being accumulated from a small group of multimorbid, long-term – complex patients, whose patient journeys are riddled with discontinuities, disruptions, and a maze of service providers who do not communicate with each other. With the development of new technologies, it is possible to treat more conditions with better outcomes, with a subsequent higher price tag. Despite ongoing efforts to address this issue, it remains a persistent and pressing problem that many healthcare systems are unable to keep up with. Framing this challenge through the lens of healthcare operations management raises the question, "How can we deliver personalized healthcare services with the efficiency of mass production?" Theories in healthcare operations management suggest that addressing complex patient journeys involves segmenting demand, standardizing modular service components, reducing or streamlining non-value-adding activities such as handovers and setups, and maximizing the use of repeatable processes. It also calls for delegating decision-making and judgment to the lowest hierarchical level where the necessary information and expertise are available. However, there are still many unanswered questions in the current literature, such as how to understand variation and its causes in the high-variability context of complex patients, how existing tools like patient pathways can be used to reduce variation while allowing flexibility where appropriate, and how these tools can promote value co-creation and bring decision-making and judgment closer to the patient. This thesis employed a mixed methods approach within the Finnish healthcare system to address patient journey disruptions. Causes of discontinuities were explored through focus groups with private-sector healthcare professionals and a survey of abdominal patients. Conceptual analysis, supported by expert input and interviews with healthcare professionals and home care patients, led to the development of a generalizable patient journey disruption framework. Guidelines for patient pathway use were derived from an exploratory survey of public-sector healthcare professionals, while the impact of a digital care pathway on value co-creation for people with multiplesclerosis was assessed via statistical analysis of electronic health records. The results of this thesis highlight various ways to structure management problems in healthcare. Key findings include insights into healthcare professionals' experiences with patient pathway use, guidelines for optimizing these pathways, and evidence that digital care pathways can enhance co-creation for people with multiple sclerosis. The primary theoretical contribution is the conceptualization of patient journey disruption as an agency-based event linked to negative outcomes in care pathways relative to reasonable expectations. This concept emphasizes identifying, analyzing, and addressing PJDs to optimize complex patient journeys.
  • Detection of Quantum Phase Transitions with a Lee-Yang Formalism and Many-Body Algorithms
    (2025) Vecsei, Pascal Marc
     School of Science | Doctoral dissertation (article-based)
    Phase transitions in many-body quantum systems arise from the collective behaviour of many degrees of freedom. For many interacting quantum many-body systems, it remains challenging to determine their phase diagram due to the exponential growth in the Hilbert space size and the difficulty of existing numerical techniques to tackle generic interacting quantum many-body systems. In this thesis, we combine state-of-the-art numerical techniques with an approach to symmetrybreaking phase transitions inspired by the works of Lee and Yang on the zeros of partition functions in the complex plane. Specifically, we study the behaviour and distribution of zeros of the momentgenerating function of the relevant order parameters to locate the phase transitions. We refer to these zeros as Lee-Yang zeros. Our approach involves using tensor networks and neural quantum states as variational states to describe the ground states or finite temperature density matrices of the systems studied. While tensor networks allow for a direct evaluation of the moment-generating function, and therefore a direct determination of the position of these Lee-Yang zeros, this is not possible for neural quantum states. Therefore, we also present a method that uses high cumulants of the order parameter combined with knowledge about the symmetries of the Lee-Yang zeros to estimate their locations. By extrapolating the distance from the origin to these zeros to the thermodynamic limit, the presence of a phase transition can be determined. Using this Lee-Yang formalism, we map out the phase diagram of the transverse field Ising model and a fermionic chain, thereby showing its practical applicability to specific quantum many-body models. Also using neural quantum states and tensor networks, we determine the phase diagram of a tetramerized antiferromagnetic spin-1/2 J₁-J₂ Heisenberg model on the square lattice. Without relying on our Lee-Yang formalism, we are able to trace out the phase diagram using conventional means, such as studying the susceptibility, spin structure factor and the many-body gap. This model, which has been recently realized in experiment, exhibits an intriguing competition between conventional magnetically ordered phases and a higher-order symmetry protected topological phase. By mapping out its phase diagram, we contribute to guiding experiments towards the parameter regimes of interest.
  • Machine Learning for Precision Medicine
    (2025) Julkunen, Heli
     School of Science | Doctoral dissertation (article-based)
    Precision medicine is an emerging approach to healthcare that tailors prevention and treatment strategies by accounting for individual patient variability. Its implementation is becoming more feasible due to advances in the scalability and cost-effectiveness of various molecular profiling technologies, such as genomics, transcriptomics, proteomics, and metabolomics. These advances have expanded not only the amount of molecular data measurable from individuals but also the availability of large-scale datasets for research, creating opportunities to discover more effective treatments, identify disease biomarkers, and develop models for predicting disease risk. However, the sheer volume and complexity of these data necessitate advanced computational methods to extract meaningful and actionable insights for precision medicine. This dissertation develops and applies computational frameworks to address various aspects of precision medicine, including predicting the effects of drug combination treatments, utilizing metabolomic biomarkers in disease risk assessment, and improving the methodological aspects of disease risk prediction. The first publication presents a machine learning framework designed to predict the effects of drug combinations across varying doses, providing an improvement over existing methods by enabling precise dose-specific predictions. This method achieved highly accurate predictions and identified novel drug combination synergies, which were subsequently experimentally validated. This framework provides an efficient tool for systematic pre-screening of drug combinations, particularly to advance cancer treatments. The second set of publications expands the current understanding of blood biomarkers in disease risk prediction through the analysis of population-scale metabolomic data. These studies identified novel metabolomic biomarkers and highlighted their potential in predicting the risks of various diseases, including diseases where metabolomics had not previously been studied at scale. The final publication proposes a machine learning method aimed at improving time-toevent disease risk prediction by incorporating comprehensive interaction effects among predictor variables. This method demonstrated improved accuracy in risk prediction compared to standard methods across multiple diseases and different data sources, thereby supporting the development of more accurate tools for risk assessment. Taken together, the novel methods and biological insights presented in this dissertation advance the translation of molecular data into prevention and treatment strategies in precision medicine.
  • Dispersion interactions in machine learning potentials for large-scale atomistic simulations
    (2025) Muhli, Heikki
     School of Science | G5 Artikkeliväitöskirja
    Dispersion interactions are intermolecular interactions that are present in all materials. Due to their cumulative and long-range nature, and the fact that these interactions originate from electron correlation, they are not properly captured by most exchange-correlation functionals used in density-functional theory (DFT). Instead, some kind of dispersion correction is often added to these calculations to account for the missing dispersion interactions. In many systems, accurate modeling of dispersion interactions requires many-body dispersion (MBD) to be taken into account, but these interactions significantly increase the computational resources required for simulations compared to the often used pairwise additive models, making them intractable for large-scale atomistic simulations. With the recent advent of machine learning (ML) in materials science, the problem with the computational cost of the DFT calculations can be mitigated by training ML potentials for various materials using DFT data as the reference database. These potentials are able to interpolate new results on the potential energy surfaces of the systems without further DFT calculations. However, because these potentials are trained on the DFT data based on descriptors of local atomic environments, they also lack proper dispersion interactions. The non-local nature of these interactions and the locality requirement of the ML potentials pose a challenge to the inclusion of these interactions in ML frameworks. In this thesis, we show how dispersion interactions can be included in ML potentials using local parametrization of the atomic environment. We first modify a pairwise-additive dispersion model for the ML framework such that it produces atom-centered contributions to the dispersion energies from local parameters. Next, the method is generalized to MBD by deriving the atom-centered dispersion from the global MBD energy such that the computational scaling of the final model is linear with respect to the number of atoms. Since the MBD implementation inevitably incurs a significant additional computational cost to the computationally efficient ML potentials, we also design a method to reparametrize the pairwise-additive model with periodic corrections from the MBD model on the fly during molecular dynamics simulations. With this method we achieve a significant speed-up for the simulations with an acceptable error in the results. As linear-scaling and readily parallelizable by design, our methodology allows one to run extremely large-scale molecular dynamics in a fraction of the computational time required for the corresponding DFT calculations, while producing quantum-mechanically accurate results that include MBD interactions.
  • Ultrasound-enhanced fine-needle aspiration for biopsy: From device development to in vivo human validation
    (2025) Le Bourlout, Yohann
     School of Science | Doctoral dissertation (article-based)
    A hypodermic needle is a widely used medical instrument. With more than 16 billion needles used annually, it is an essential instrument of modern medicine. Medical needles are primarily used to deliver or draw material to or from a patient, playing a crucial role in various medical procedures. Despite the broad use of needles, limited technological improvements have been made to them over the last decades, leading to urgent unmet needs in applications such as biopsy. Biopsy is a medical procedure that aims to collect tissue from a suspected lesion for diagnosis. The analysis is intended to define what pathology is present in the target tissue and to decide on the potential treatment. While diagnostic techniques have improved, the quantity of tissue required has only grown due to the increasing range of assessment, each of them requiring part of the sample. Moreover, the tissue quantity and representativeness of the lesion is limited by the current techniques. To address this unmet clinical need, an Ultrasound-enhanced Fine-Needle Aspiration Biopsy (USeFNAB) has been proposed. USeFNAB utilizes ultrasonic actuation at the needle tip (peak-to-peak displacement < 200 μm) to detach the cells and tissue construct from the target. Combined with low pressure, USeFNAB aims to collect more tissue than the state-of-the-art methods to address the demand for increased tissue required for additional pathological analysis. To date, USeFNAB instrument has not been optimized in terms of power efficiency, ergonomics and size, and validation in human ex vivo and in vivo tissue has been missing. The aim of this Thesis was to develop a novel USeFNAB device capable of providing ultrasonic movement at the needle tip efficiently. This would allow miniaturization, battery integration, and improvement of the safety of the device. Once the device configurations were optimized in silico and prototyped, a validation of USeFNAB in tissue was conducted. USeFNAB was first tested on benign ex vivo human tonsils, and then on ex vivo human neo-plastic tissues (pleomorphic adenomas and head and neck cancers). Finally, the benefits of USeFNAB were demonstrated in a clinical trial on in vivo human pleomorphic adenomas. The results showed that the USeFNAB was capable to provide a flexural actuation at the needle tip with up to 73% electrical-to-acoustical power efficiency. It also demonstrated an increase in terms of mass collection in ex vivo human tissue by 2-7× compared to state-of-the-art needle biopsies without compromising the diagnosis outcome. Finally, in the in vivo clinical setting, USeFNAB showed a tissue area increase on the histological slide by 1.7-3.4× without compromising the diagnosis or increasing complications during the procedure as compared to state-of-the-art needle biopsies. The findings in this Thesis show promise for the USeFNAB to be the next-generation tool for cancer diagnosis. They also open avenues for applying ultrasound-enhanced medical needles to other applications, such as therapy.
  • Gas-phase Synthesis of Semiconducting Single-walled Carbon Nanotubes for Advanced Electronics
    (2024) Liu, Peng
     School of Science | Doctoral dissertation (article-based)
    The advancement of electronics towards portable, high-performance, and flexible technologies faces significant challenges as device sizes continue to shrink. This has driven the exploration of advanced nanomaterials, with single-walled carbon nanotubes (SWCNTs), standing out due to their unique one-dimensional structure and exceptional properties, including superior conductivity, transparency, flexibility, and stability. However, scalable and controllable synthesis of SWCNT remains a critical challenge for their large-scale integration into nanoelectronics.  This dissertation focuses on the synthesis of semiconducting SWCNTs (s-SWCNTs) and their application in electronics, with the prime goal of achieving high-purity s-SWCNTs and effectively integrating them into electronic devices. The study employs a floating catalyst chemical vapor deposition (FCCVD) method to enable continuous large-scale synthesis. By optimizing growth parameters, such as temperature, gas composition, feeding rate, and carbon sources, we achieved a controllable synthesis of s-SWCNTs with a purity as high as 94%, one of the highest purities attained through direct synthesis. Water and carbon dioxide are believed to act as oxidizing agents, selectively etching metallic SWCNTs and enhancing the yield of s-SWCNTs. The resulting nanotubes have a mean diameter of approximately 1 nm and a mean length of 6.38 μm, significantly longer than those produced via solution-sorting methods. Furthermore, these synthesized s-SWCNTs demonstrated a mean mobility of 376 cm2V-1s-1 and an on-off ratio of up to 8.33×106 in individual-SWCNT field effect transistors (FETs). The carrier mobility in the optimal FETs approaches the theoretical limit for 1 nm SWCNT on a SiO2 substrate at room temperature.  Importantly, this dissertation introduces a novel, lithography-free fabrication technique for creating flexible, wafer-scale, all-CNT device arrays, addressing common issues of contamination and damage that typically arise during traditional wet processing. The resulting wafer-scale all-CNT photodetector arrays exhibit excellent uniformity, wearability, environmental stability, and notable broadband photoresponse. These photodetectors achieve a high responsivity of 44 A/W, significantly outperforming similar CNT photodetectors fabricated using photolithography and solution sorting, even with purities of up to 99% s-SWCNT. Furthermore, the dry transfer manufacturing process was employed to fabricate transparent flexible electrodes, successfully transferring SWCNT film onto MoS₂, thereby optimizing the photodetectors' light absorption and carrier separation, leading to improved responsivity. 
  • Modeling of Charged Polymers in Aqueous Solutions in the Presence of Salt Ions and Substrates
    (2024) Vahid, Hossein Dastjerdi
     School of Science | Doctoral dissertation (article-based)
    Polyelectrolytes (PEs) are charged macromolecules. They interact with surrounding ions in aqueous solutions, playing a pivotal role in various scientific and industrial applications. Examples of PEs include biologically relevant molecules such as DNA, RNA, and proteins, as well as synthetic PEs like poly(styrene sulfonate) and poly(diallyl dimethyl ammonium chloride), which are widely used in industrial applications. The interactions of PEs with ions and substrates are governed by a complex interplay of electrostatic, steric, and dielectric effects, making it essential to develop a comprehensive understanding of these processes at the molecular level. This thesis aims to enhance our understanding of PE interactions through theoretical models and molecular dynamics simulations. It explores the influence of ion characteristics and investigates how they affect PE behavior in bulk solutions and at interfaces. To this end, a soft-potential-enhanced Poisson-Boltzmann model is developed and optimized using coarse-grained molecular dynamics (CGMD) simulations and atomistic models. This model captures the influence of ion size in monovalent salt solutions, providing accurate predictions of ion distributions around rodlike PEs. Beyond mean-field theory, CGMD simulations are used to study the effect of correlations arising from multivalent ions and ions with varying shapes on PE interactions. The results highlight the critical role of ion shape and concentration in controlling the range of PE interactions, offering valuable insights for designing PE-based systems, with potential applications in drug delivery, material science, and beyond.In addition, the research highlights the role of ion valency on the electrophoretic mobility (EM) of PEs. Contrary to the conventional assumption that EM is controlled by counterion valency, we show that coion valency can significantly influence EM. Specifically, we find that increasing the coion valency can reverse EM in systems with high salt concentrations, revealing a complex, non-monotonic relationship between ion valency and EM. Moreover, we examine how dielectric discontinuities between the solvent and solid substrates influence PE adsorption. The findings demonstrate that the dielectric mismatch, as well as salt concentration, ion valency, and PE charge, play a crucial role in determining whether PEs are attracted to or repelled from the substrate. This understanding provides practical guidelines for optimizing PE adsorption and monolayer formation on neutral, polarizable substrates, which is essential for applications in material science, biotechnology, and beyond. Overall, this thesis contributes to a deeper understanding of the fundamental mechanisms governing PE interactions in complex environments, offering strategies for controlling and optimizing PE behavior in various scientific and industrial applications, paving the way for the development of more efficient and targeted PE-based systems.
  • Deep Learning Methods for Point Matching, Visual Localization and 3D Reconstruction
    (2024) Wang, Shuzhe
     School of Science | Doctoral dissertation (article-based)
    This doctoral thesis explores advanced deep learning methods for three pivotal tasks in 3D computer vision: point matching, visual localization, and 3D reconstruction. These tasks are crucial for enabling machines to perceive and understand 3D environments, which is essential for applications ranging from virtual reality to robotics and autonomous driving. In point matching, this thesis investigates learning-based, visual descriptor-free matching pipelines that leverage geometric and color cues in conjunction with Graph Neural Networks. This approach significantly enhances the accuracy of 2D-3D keypoint matching while reducing memory usage and improving data privacy. For visual localization, the thesis introduces two hierarchical scene coordinate network architectures to establish dense 2D-3D matches for accurate 6-DoF camera pose estimation. These architectures incorporate conditioning mechanism and transformer to encode global context to local patches, overcoming scene ambiguities. Additionally, a novel few-shot learning setting is proposed, which reduces the training load for scene coordinate regression and improves training efficiency from days to minutes. A significant contribution of the thesis is the development of an end-to-end dense unconstrained 3D reconstruction pipeline based on Vision Transformers. This pipeline directly regresses point coordinates from image pairs without relying on camera parameters, simplifying traditional 3D reconstruction methods and introducing a unified framework for monocular and binocular reconstruction tasks. The thesis also explores methods to improve local feature matching by calculating the curvature of local 3D surface patches for detected points, enhancing matching accuracy with off-the-shelf learned matchers. Furthermore, it addresses the challenge of continual learning for visual localization by proposing an experience-replay-based baseline to prevent catastrophic forgetting and reduce computational and storage costs. Throughout the thesis, the importance of end-to-end learning is emphasized, where models are trained to directly produce desired outputs from raw input data. This paradigm shift can simplify the development pipelines and enhance the adaptability and scalability of 3D computer vision systems. By integrating deep learning with traditional geometric principles, this research provides a comprehensive framework for addressing key challenges in point matching, visual localization, and 3D reconstruction.
  • Temporal Coordination and Criticality in Human Neural Dynamics - Bridging Insights from rhythmicity, synchronization, and computational modeling
    (2024) Myrov, Vladislav
     School of Science | Doctoral dissertation (article-based)
    The neural processing of exogenous and endogenous information is organized into highly specialized anatomically distinct neural populations. These populations undergo rhythmic fluctuations in excitability, commonly referred to as neural oscillations. Neural oscillations have been studied for decades and are thought to be fundamental to the temporal coordination of information flow throughout the cortical processing hierarchy. However, after decades of research, data analysis methods for neural oscillations are still under active development. In the first study of this thesis, we extended the operationalization of oscillations beyond the power-based lens and developed a novel method that quantifies rhythmicity directly. The results showed that the actual rhythmicity of cortical oscillations is well delineated in both anatomical and spectral properties, with the mesoscale architecture organized into small frequency islands. However, no neuron operates independently of others. In the second study, we investigated long-range phase synchronization between high gamma frequencies in humans. We demonstrated that this synchronization has a modular architecture, segregates into healthy and epileptic zones, is characterized by laminar profiles, and is transiently enhanced and suppressed in separate frequency bands during a response inhibition task.Observables of neural dynamics vary between individuals and cognitive states. The critical brain hypothesis offers an explanation for this, and in Study III, we tested whether the level of variability is associated with the critical state of a neural system. We argue that in the human brain in vivo, a hypothetical critical point is stretched into a wider regime of critical-like dynamics—Griffith's phase—and that healthy brain regions operate on the subcritical and critical sides of this extended critical regime, while epileptogenic areas are located on the critical-supercritical side. Computational models have shown great utility in discovering the mechanistic principles of observed phenomena. In the fourth study, we developed a hierarchical model of critical-like oscillatory activity based on the Kuramoto model. We found that the model observables are physiologically plausible and most similar to real human recordings in the subcritical-critical range, demonstrating that the model can be fitted to reconstruct individual behavior with a high degree of similarity. Taken together, these studies provide new insights into the analysis of neural oscillations through complementary approaches, including connectivity in vivo, computational studies, phase-synchronization networks, and the criticality properties of individual nodes. In addition to extending the boundaries of known neural dynamics observables, we delved deeper into the Terra Incognita and found a way to directly quantify the rhythmicity of oscillations in alignment with the brain clock's timing. This may provide new insights into the analysis of information processing in the brain.
  • Non-diffractive structured light fields - Generation, properties, and applications
    (2024) Hildén, Panu
     School of Science | Doctoral dissertation (article-based)
    Propagation of light can be controlled by structuring the propagation medium or the light itself. Methods for controlling light are continuously improved and renewed to meet the requirements of modern optical and photonic devices. The work presented in this thesis advances these developments, being focused on propagation-invariant optical fields and their applications in laser beam engineering, optical imaging, and photonic chip technology.In the thesis, it is shown that diffraction-free optical beams can be structured, both spatially and spectrally, to propagate at an arbitrary group velocity in free space. Through a theoretical analysis, we find that in addition, the group velocity can be made to vary as a function of both space and time. As an example, a two-frequency Bessel beam with a spatially varying group velocity that reaches subluminal, superluminal, and negative values has been demonstrated in one of our experiments. This work also presents a design of a compound prism with an extraordinarily high angular dispersion, i.e., the ability to split the incident light into its constituent frequency components. The prism is designed for generating non-diverging two-dimensional light sheets. A common problem with such light sheets is the presence of multiple interference fringes in the beam profile, which the high angular dispersion of the prism helps remove.Non-diverging Bessel beams are also employed in this work to extend the depth of field of an optical imaging system. The system comprises two lenses and a ring-shaped aperture placed at their common focal plane. This makes the images consist of nearly diffraction-free Bessel-like beams. Consequently, the object can be displaced from the front focal plane of the system without blurring the image. The performance of the system is assessed with both coherent and incoherent light. The thesis also describes an analytical method for modeling discrete diffraction in waveguide arrays. This type of diffraction appears as a result of light leakage between neighboring waveguides in densely packed waveguide arrays. The phenomenon, known as crosstalk, limits the miniaturization of photonic integrated circuits. The presented model offers new insight into the problem of crosstalk suppression and shows that the problem can be solved by bending the waveguides. The results suggest a way to further reduce the footprint of photonic chips without hindering their performance.
  • Control of Spin-Wave Propagation in Multiferroic Heterostructures: Magnetic Domain and Domain-Wall Driven Spin-Wave Guiding
    (2024) Zhu, Weijia
     School of Science | Doctoral dissertation (article-based)
    In today's digital age, modern information computing has become integral to our daily lives, facilitated by rapid technological advancements in transistors within central processing units (CPUs) that provide powerful computational capabilities. As the demand for more efficient and powerful processing continues to rise, the increasing density of transistors in circuits faces significant challenges, particularly in heat generation. To address these challenges, magnonics is being investigated as a promising alternative for future logic circuits. Unlike traditional electronics, which rely on electron charge and current, magnonics conveys information through the amplitude and phase of spin waves. This enables data transport via the transmission of spin angular momentum without Joule heating. The development of efficient magnonic circuits necessitates effective manipulation of spin waves. In this thesis, I present a study on the active control of spin-wave propagation in strain-coupled multiferroic heterostructures. These structures consist of a ferromagnetic film deposited on a ferroelectric BaTiO3 substrate that exhibits alternating ferroelectric polarization domains. Strain coupling at the bilayer interface transfers the ferroelectric domain pattern onto the ferromagnetic film, resulting in regular magnetic anisotropy patterns through inverse magnetostriction. The regular alternation of magnetic anisotropy firmly pins the magnetic domain walls to the ferroelectric domain walls. In a Fe/ BaTiO3 multiferroic heterostructure, the abrupt change in the spin-wave dispersion relation and phase velocity at the magnetic anisotropy boundary results in spin-wave refraction. I experimentally demonstrate zero-field routing of spin waves across these anisotropy boundaries. This refraction effect is further confirmed by micromagnetic simulations and calculations based on a modified Snell's law for magnonics. The routing of spin waves in this system can be efficiently controlled by adjusting parameters such as the frequency, incident angle, and anisotropy strength. In a CoFeB/ BaTiO3 multiferroic heterostructure, I utilized the strongly pinned magnetic domain walls as nanoscopic channels for spin-wave propagation. The robust pinning of domain walls enables the initialization of head-to-head and head-to-tail domain wall configurations. Each type of magnetic domain wall exhibits distinct magnetization rotation and width, resulting in switchable spin-wave dispersion relations that allow for fully reversible control of propagating spin-wave modes. I showed that an external magnetic field can tune the domain wall characteristics, shifting the dispersion relation without altering its position, thereby providing continuous control over the wavelength of localized spin waves. These findings, combined with the potential for electric-field-driven domain wall motion in multiferroic heterostructures, pave the way to low-power, reconfigurable magnonic devices.
  • Manipulating Quantum States in 2D Ferroics
    (2024) Amini, Mohammad
     School of Science | Doctoral dissertation (article-based)
    In the past 20 years, experimental condensed matter physics has experienced a drastic shift. The discovery of graphene has revolutionized the field and brought huge attention to studying 2D materials. Since then, many van der Waals (vdW) 2D materials with different properties, such as ferroelectricity, ferromagnetism, and multiferroicity, have been discovered. In this thesis, I will focus on creating and studying these 2D systems. I will use molecular beam epitaxy (MBE) technique to design and fabricate van der Waals (vdW) heterostructures and scanning tunneling microscope (STM) to study their properties at cryogenic temperatures. I will showcase how we can engineer different materials with novel properties, which are at the core of condensed matter physics. The first topic concerns artificial multiferroics, which we achieve by growing a magnetic layer of iron phthalocyanine (FePc) on a monolayer ferroelectric tin telluride (SnTe). I will show how we can control the orbital ordering of the molecule using the substrate's electric polarization. Secondly, I will focus on inelastic magnon excitations in monolayer ferromagnet chromium(III) bromide (CrBr3). We will show how their energy can be tuned by the moiré pattern between the CrBr3 and the highly ordered pyrolytic graphite (HOPG) substrate. These elementary excitations, which are the property of ferromagnets, are probed for the first time at the atomic scale using STM.  Finally, I will showcase our atomic scale study of the first known 2D multiferroic NiI2. I will show the existence of multiferroic properties in this system for the first time using STM. Our experiments allow atomic scale understanding of the mechanism of the multiferroicity in this system. By manipulating the multiferroic domains using external magnetic and electric fields, I will confirm the magnetoelectric coupling in this system. This thesis will provide an overview of 2D systems with novel ferroic properties and paves the way for further advancements in condensed matter physics and quantum materials research.