10081 research outputs found
Sort by
Data driven regularization models of non-linear ill-posed inverse problems in imaging
Get PDFImaging technologies are widely used in application fields such as natural sciences, engineering, medicine, and life sciences. A broad class of imaging problems reduces to solve ill-posed inverse problems (IPs). Traditional strategies to solve these ill-posed IPs rely on variational regularization
methods, which are based on minimization of suitable energies, and make use of knowledge about the image formation model (forward operator) and prior knowledge on the solution, but lack in incorporating knowledge directly from data. On the other hand, the more recent learned approaches can easily learn the intricate statistics of images depending on a large set of data, but do not have a
systematic method for incorporating prior knowledge about the image formation model. The main purpose of this thesis is to discuss data-driven image reconstruction methods which combine the benefits of these two different reconstruction strategies for the solution of highly nonlinear ill-posed
inverse problems. Mathematical formulation and numerical approaches for image IPs, including linear
as well as strongly nonlinear problems are described. More specifically we address the Electrical
impedance Tomography (EIT) reconstruction problem by unrolling the regularized Gauss-Newton
method and integrating the regularization learned by a data-adaptive neural network. Furthermore
we investigate the solution of non-linear ill-posed IPs introducing a deep-PnP framework that integrates
the graph convolutional denoiser into the proximal Gauss-Newton method with a practical
application to the EIT, a recently introduced promising imaging technique. Efficient algorithms are
then applied to the solution of the limited electrods problem in EIT, combining compressive sensing
techniques and deep learning strategies. Finally, a transformer-based neural network architecture
is adapted to restore the noisy solution of the Computed Tomography problem recovered using the
filtered back-projection method
Optimizing AI at the Edge: from network topology design to MCU deployment
Get PDFThe first topic analyzed in the thesis will be Neural Architecture Search (NAS).
I will focus on two different tools that I developed, one to optimize the architecture of Temporal Convolutional Networks (TCNs), a convolutional model for time-series processing that has recently emerged, and one to optimize the data precision of tensors inside CNNs.
The first NAS proposed explicitly targets the optimization of the most peculiar architectural parameters of TCNs, namely dilation, receptive field, and the number of features in each layer. Note that this is the first NAS that explicitly targets these networks.
The second NAS proposed instead focuses on finding the most efficient data format for a target CNN, with the granularity of the layer filter. Note that applying these two NASes in sequence allows an "application designer" to minimize the structure of the neural network employed, minimizing the number of operations or the memory usage of the network.
After that, the second topic described is the optimization of neural network deployment on edge devices. Importantly, exploiting edge platforms' scarce resources is critical for NN efficient execution on MCUs.
To do so, I will introduce DORY (Deployment Oriented to memoRY) -- an automatic tool to deploy CNNs on low-cost MCUs.
DORY, in different steps, can manage different levels of memory inside the MCU automatically, offload the computation workload (i.e., the different layers of a neural network) to dedicated hardware accelerators, and automatically generates ANSI C code that orchestrates off- and on-chip transfers with the computation phases.
On top of this, I will introduce two optimized computation libraries that DORY can exploit to deploy TCNs and Transformers on edge efficiently.
I conclude the thesis with two different applications on bio-signal analysis, i.e., heart rate tracking and sEMG-based gesture recognition
New approaches to the treatment of aortic valvulopathy
Get PDFBackground
L’incidenza di malattie valvolari aortiche è in costante aumento. La terapia definitiva è chirurgica o interventistica, determinando un evidente miglioramento della qualità di vita, a fronte di un rischio operatorio ormai estremamente basso. Le linee guida internazionali più recenti pongono in classe I entrambe le procedure nella fascia di età fra 65 e 80 anni.
Materiali e metodi
È stata effettuata un’analisi retrospettiva dei pazienti di età compresa fra 65 e 80 anni, sottoposti a sostituzione valvolare aortica isolata chirurgica con bioprotesi sutureless (gruppo SU-AVR), oppure trans-catetere (gruppo TAVR), presso Maria Cecilia Hospital tra gennaio 2011 e dicembre 2021. Mediante propensity score matching sono stati analizzati, nei due gruppi risultanti, gli outcomes di mortalità e complicanze intraospedaliere, a 30 giorni, ad un anno e attuariale.
Risultati
Sono stati inclusi nello studio 638 pazienti, di cui 338 (52.98%) nel gruppo SU-AVR e 300 (47.02%) nel gruppo TAVR. Dopo propensity score matching, sono stati ottenuti due gruppi di pazienti (124 per gruppo) senza differenze statisticamente significative nelle comorbidità preoperatorie. La mortalità a 30 giorni è risultata sovrapponibile nei 2 gruppi. Il gruppo TAVR ha mostrato un’incidenza significativamente maggiore di impianto di pacemaker definitivo e di danni vascolari maggiori, mentre il gruppo SU-AVR ha mostrato una maggior incidenza di fibrillazione atriale, di trasfusioni e di insufficienza renale. La mortalità per tutte le cause a un anno è risultata significativamente maggiore per il gruppo TAVR e il divario continua ad aumentare con il tempo.
Conclusioni
La sostituzione valvolare aortica trans-catetere (TAVR) mostra risultati molto buoni nel breve termine nei pazienti fra 65 e 80 anni di età. Al follow-up a medio termine, tuttavia, i risultati preliminari mostrano un miglior outcome dei pazienti sottoposti a sostituzione valvolare chirurgica, sia in termini di mortalità per qualsiasi causa che di eventi cardiovascolari e cerebrovascolari maggiori.Background
The incidence of aortic valve disease is constantly increasing. The definitive therapy is surgical or interventional, resulting in a clear improvement in the quality of life, with an extremely low operative risk. The most recent international guidelines place both procedures in class I in the age group between 65 and 80 years.
Materials and methods
A retrospective analysis of patients aged 65 to 80 years undergoing surgical isolated aortic valve replacement with sutureless bioprostheses (SU-AVR group), or trans-catheter (TAVR group) was performed at Maria Cecilia Hospital between January 2011 and December 2021. Using propensity score matching, intrahospital, 30-day, one-year and actuarial mortality and complications outcomes were analyzed in the two resulting groups.
Results
A total of 638 patients were included in the study, counting 338 (52.98%) in the SU-AVR group and 300 (47.02%) in the TAVR group. After propensity score matching, two groups of patients (124 per group) were obtained without statistically significant differences in preoperative comorbidities. Mortality at 30 days was comparable in the 2 groups. The TAVR group showed a significantly higher incidence of definitive pacemaker implantation and major vascular damage, while the SU-AVR group showed a higher incidence of atrial fibrillation, transfusions and renal failure. One-year all-cause mortality was significantly higher for the TAVR group, and the gap continues to increase with time.
Conclusions
Transcatheter aortic valve replacement (TAVR) shows very good results in the short term in patients between 65 and 80 years of age. At medium-term follow-up, however, preliminary results show a better outcome of patients undergoing surgical valve replacement, both in terms of all-cause mortality and major cardiovascular and cerebrovascular events
PCM-based in-memory computing: architectures, circuits and applications
Get PDFAnalog In-memory Computing (AIMC) has been proposed in the context of Beyond Von Neumann architectures as a valid strategy to reduce internal data transfers energy consumption and latency, and to improve compute efficiency. The aim of AIMC is to perform computations within the memory unit, typically leveraging the physical features of memory devices. Among resistive Non-volatile Memories (NVMs), Phase-change Memory (PCM) has become a promising technology due to its intrinsic capability to store multilevel data. Hence, PCM technology is currently investigated to enhance the possibilities and the applications of AIMC. This thesis aims at exploring the potential of new PCM-based architectures as in-memory computational accelerators. In a first step, a preliminar experimental characterization of PCM devices has been carried out in an AIMC perspective. PCM cells non-idealities, such as time-drift, noise, and non-linearity have been studied to develop a dedicated multilevel programming algorithm. Measurement-based simulations have been then employed to evaluate the feasibility of PCM-based operations in the fields of Deep
Neural Networks (DNNs) and Structural Health Monitoring (SHM). Moreover, a first testchip has been designed and tested to evaluate the hardware implementation of Multiply-and-Accumulate (MAC) operations employing PCM cells. This prototype experimentally demonstrates the possibility to reach a 95% MAC accuracy with a circuit-level compensation of cells time drift and non-linearity. Finally, empirical
circuit behavior models have been included in simulations to assess the use of this technology in specific DNN applications, and to enhance the potentiality of this innovative computation approach
Trees: radiocarbon and dendrochronology - together for the Resolution project
Get PDFThe year 14,226 BP marks an important border in the actual radiocarbon (14C) calibration curve: the high resolution and precision characterising the first part (0 – 14,226 BP) of the curve are due to the potential represented by tree-ring datasets, which directly provide the atmospheric 14C content at the time of tree-rings formation with high resolution. They systematically decrease going back in time, where only a few floating tree-ring chronologies alternate to other low-resolution records.
The lack of resolution in the dating procedure before 14,226 years BP leads to significant issues in the interpretation and untangling of tricky facts of our past, in the field of Human Evolution.
Research on sub-fossil trees and the construction of new Glacial tree-ring chronologies can significantly improve the radiocarbon dating in terms of temporal resolution and precision until 55,000 years BP to clear puzzles in the Human Evolution history.
In this thesis, the dendrochronological study, the radiocarbon dating and the extrapolation of environmental and climate information from sub-fossil trees found on the Portugal foreshore, remnants of a Glacial lagoonal forest, are presented.
The careful sampling, the dendrochronological measurements and cross-dating, the application of the most suitable cellulose extraction protocol and the most advanced technologies of the MICADAS system at ETH-Zurich, led to the construction of a new 220-years long tree-ring site chronology and to high resolution, highly reliable and with a tight error range radiocarbon ages.
At the moment, it results impossible to absolutely date this radiocarbon sequence by the comparison of Δ14C of the trees and 10 Be fluctuations from the ice-cores. For this reason, tree growth analysis, comparisons with a living pine stand and forest-fires history reconstruction have made it possible to hypothesize site and climate characteristics useful to constrain the positioning in time of the obtained radiocarbon sequence
The European public intervention for the territory between globalisation and the crisis of the state. Theoretical instruments for reducing inequalities.
Get PDFIl presente elaborato indaga l’evoluzione del rapporto tra ordinamento giuridico
e territorio, alla luce dei processi di globalizzazione e integrazione europea.
Nel secondo dopoguerra, infatti, si è assistito ad un’evoluzione dello Stato
moderno in quello che è stato definito come lo Stato keynesiano, nel quale
convivevano una forte presenza pubblica nell’economia interna e un basso livello di
internazionalizzazione del commercio mondiale.
La crisi di tale modello è la crisi dello Stato territoriale, che viene attraversato
da nuovi flussi economici che premiano città e regioni, facendo perdere di importanza
alla dimensione nazionale. Ciò è avvolorato dal processo di integrazione europea che
dagli anni ’80 in poi trova nuovo vigore e comincia a limitare l’intervento pubblico
nell’economia agendo sia sul piano degli aiuti di Stato, che sui bilanci nazionali.
Tali dinamiche producono la crisi dell’unità dello Stato sotto un aspetto
giuridico-territoriale, per via del crescente ruolo tanto delle istituzioni europee, da una
parte, e di città e Regioni, dall’altra, le quali diventano nuove dimensioni normative
che sfidano la sovranità statale. Allo stesso tempo, si indebolisce anche l’unità sociale,
con divari territoriali crescenti, sia a livello inter- che intra- regionali.
In questo contesto, si approfondirà come l’intervento pubblico europeo sia sul
piano della coesione, che su quello relativo agli aiuti di Stato non solo tenda ad una
riduzione dei divari, ma anche ad una riconfigurazione del territorio europeo. Infatti,
grazie ai suoi strumenti, città e regioni hanno la possibilità di superare i propri confini
amministrativi al fine di creare nuove forme di cooperazione territoriale.
In questo scenario, si proporrà una riflessione sulla possibilità di un rinnovato
principio di sussidiarietà, che tenga conto della struttura reticolare dell’attuale contesto
territoriale europeo, così come degli attuali rapporti tra la dimensione del mercato e
quella sociale, al fine di meglio descrivere un ordinamento europeo in senso materiale.This doctoral thesis investigates the evolution of the relationship between the legal system and the territory, in the light of the processes of globalisation and European integration. The post-World War II period witnessed the evolution of the modern state into what has been defined as the Keynesian state, in which a strong public presence in the domestic economy and a low level of internationalisation of world trade coexisted. The crisis of this model is the crisis of the territorial state, which is crossed by new economic flows that reward cities and regions, making the national dimension lose importance. These dynamics produce a crisis of the unity of the state from a legal-territorial point of view, due to the growing role of both the European institutions on the one hand and cities and regions on the other, which become new normative dimensions that challenge state sovereignty. At the same time, social unity is also weakening, with growing territorial divides, both inter- and intra-regional. In this context, we will explore how European public intervention, both in terms of cohesion policy and state aid, tends not only to reduce the gaps, but also to reconfigure the European territory. In fact, thanks to its instruments, cities and regions have the possibility to go beyond their administrative borders in order to create new forms of territorial cooperation. In this scenario, a reflection will be proposed on the possibility of a renewed principle of subsidiarity, which takes into account the network structure of the current European territorial context, as well as the current relations between the market and the social dimension, in order to better describe a European order in the material sense
Urine-derived Renal Epithelial Cells from kidney transplanted patients: phenotype, immunomodulatory properties, and effect of the exposition to NGAL
Get PDFDuring kidney transplant procedure transplanted organs can undergo ischaemia reperfusion phenomena, often associated with the onset of acute kidney damage, loss of kidney function and rejection. These events promote cell turnover to replace damaged cells and preserve kidney function, thus cells deriving from nephrons structures are highly voided in urine. Urine derived cells represents a promising cell source since they can be easily isolated and cultured. The aim of this project was to characterise Urine-derived Renal Epithelial Cells (URECs) from transplanted kidney and to evaluate how these cells react to the co-culture with immune cells. URECs expressed typical markers of kidney tubule epithelial cells (Cytokeratin and CD13), and a subpopulation of these cells expressed CD24 and CD133, which are markers of kidney epithelial progenitor cells. The expression of immunosuppressive molecules as HLA-G and CD73 was also observed. As matter of fact, during the co-culture with PBMCs, UREC suppressed the proliferation of CD4 and CD8 Lymphocytes and reduce the T helper 1 subset, while increasing the T regulatory counterpart. Also, preliminary data observed in this study indicated that the exposition to kidney damage associated molecule, such as NGAL, could significantly affect UREC viability and immunomodulatory capacity. These results add new information about the phenotype of urine cells obtained after kidney transplant and reveal that these cells show promising immunomodulatory properties, suggesting their potential application in personalized cell therapy approaches
“Clinical and biological role of adjuvant dendritic cells vaccination in newly glioblastoma patients”
Get PDFBackground. Glioblastoma (GBM) is the most common primary tumor of central nervous system and it has a poor prognosis. Standard first line treatment, which includes surgery followed by adjuvant radio-chemotherapy,produces only modest benefits to survival. The interest for immunotherapy in
this field derives from the development of new drugs and effective therapies as immune-check points inhibitors, adoptive T-cell approaches or dendritic cell (DC) based vaccines or a combinations of these. GBM is described as a
typical “immune-deserted” cancer exhibiting a number of systemic and environmental immunosuppressive factors. Considering the role of microenvironment, and above all the lower tumor load and depletion of immunosuppressive cells in GBM, our hypothesis is that DC vaccine may induce an immune response.
Main aims and study design. The main aim of this project is to study the role of immune system in GBM, including identification of potential prognostic and predictive markers of outcome and response to dendritic cell vaccine.
Firstly, we performed a retrospective analysis on blood samples. Then, we analyzed the immuno-component in tissues samples of enrolled patients; and compared that with blood results. Then, the last part of the project is based on a prospective clinical trial on patients enrolled in DC-based vaccination produced at IRST Cell Factory and actually used for patients with melanoma and other tumors. The enrollment is still ongoing.
Expected results. The project will i) develop an immune-panel of prognostic and predictive markers to help clinicians to improve the therapeutic strategy for GBM patients; ii) provide preliminary results on the effectiveness of immunotherapy on GBM patients
Excitonic properties of transition metal oxide perovskites and workflow automatization of GW schemes
Get PDFThe Many-Body-Perturbation Theory approach is among the most successful theoretical frameworks for the study of excited state properties. It allows to describe the excitonic interactions, which play a fundamental role in the optical response of insulators and semiconductors.
The first part of the thesis focuses on the study of the quasiparticle, optical and excitonic properties of \textit{bulk} Transition Metal Oxide (TMO) perovskites using a GW+Bethe Salpeter Equation (BSE) approach. A representative set of 14 compounds has been selected, including 3d, 4d and 5d perovskites. An approximation of the BSE scheme, based on an analytic diagonal expression for the inverse dielectric function, is used to compute the exciton binding energies and is carefully bench-marked against the standard BSE results.
In 2019 an important breakthrough has been achieved with the synthesis of ultrathin SrTiO3 films down to the monolayer limit. This allows us to explore how the quasiparticle and optical properties of SrTiO3 evolve from the bulk to the two-dimensional limit.
The electronic structure is computed with G0W0 approach: we prove that the inclusion of the off-diagonal self-energy terms is required to avoid non-physical band dispersions. The excitonic properties are investigated beyond the optical limit at finite momenta.
Lastly a study of the under pressure optical response of the topological nodal line semimetal ZrSiS is presented, in conjunction with the experimental results from the group of Prof. Dr. Kuntscher of the Augsburg University.
The second part of the thesis discusses the implementation of a workflow to automate GW and BSE calculations with the VASP software.
The workflow adopts a convergence scheme based on an explicit basis-extrapolation approach [J. Klimeš \textit{et al.}, Phys. Rev.B 90, 075125 (2014)] which allows to reduce the number of intermediate calculations required to reach convergence and to explicit estimate the error associated to the basis-set truncation
Analysis of urban infrastructure for sustainable mobility through instrumented bicycles for students
Get PDFIn Europe almost 80% of the continent's population lives in cities. It is estimated that by 2030 most regions in Europe which contain major cities will have even more inhabitants on 35–60% more than now. This process generates a consequent elevate human pressure on the natural environment, especially around large urban agglomerations. Cities could be seen as an ecosystem, represented by the dominance of humans that re-distribute organisms and fluxes and represent the result of co-evolving human and natural systems, emerging from the interactions between humans, natural and infrastructures. Roads have a relevant role in building links between urban components, creating the basis on which it is founded the urban ecosystem itself. This thesis is focused on the research for a comprehensive model, framed in European urban health & wellbeing programme, aimed to evaluate the determinants of health in urban populations. Through bicycles, GPS and sensor kits, specially developed and produced by University of Bologna for this purpose, it has been possible to conduct on Bologna different direct observations that oriented the novelty of the research: the categorization of university students cyclists, connection among environmental data awareness and level of cycling, and an early identification of urban attributes able to impact on road air quality and level of cycling. The categorization of university students’ cyclist has been defined through GPS analysis and focused survey, that both permit to identify behavioural and technical variables and attitudes towards urban cycling. The statistic relationship between level of cycling, seen as number of bicycles passages per lane and pollutants level, has been investigated through an inverse regression model, defined and tested through SPSS software on the basis of the data harvest. The research project that represents a sort of dynamic mobility laboratory on two wheels, that permits to harvest and study detected parameters