1731 research outputs found
Sort by
Development of a Gamma-Ray Detector based on Silicon Photomultipliers for Prompt Gamma Imaging and High-Energy Spectroscopy
No full textProton therapy is a recent type of radiotherapy that uses high-energy proton beams, and more recently carbon ion beams, to benefit of their physical selectivity.
The energy deposited by these particle beams is inversely proportional to their velocity. Therefore they release most of the energy at the end of their path into the tissue. The energy is deposited in a few millimeters, in a zone called the Bragg peak. Before and after the Bragg peak the energy deposition is minimal. The depth and the width of the Bragg peak depends on the beam energy and on the density of tissues located along the beam path. By setting the beam energy, the Bragg peak can be positioned in the tumor site, avoiding the healthy tissues. Because of the sharpness of the Bragg peak zone, proton therapy is advantageous for tumors located near to important body part, such as the brain, spine, and neck. The drawback is that small uncertainties on particle range can have a serious impact on treatment and limit the efficiency of the proton therapy.
To obtain more effective treatments in proton therapy real-time range verifications are necessary to perform on-line corrections of the delivered treatment. Among different techniques presented in the literature, positron emission tomography (PET) and prompt gamma imaging (PGI) are the most promising methods for in vivo range verification. PET and PGI are indirect approaches to measure protons penetration depth inside patients because they aim to detect secondary particles resulting from the interaction between proton beams and tissue nuclei. PET imaging detects coincidence gamma rays due to the production of positron emitters and requires some minutes to achieve enough statistics to have a sufficient signal to noise ratio. PGI instead uses prompt gamma rays generated by de-excitation of target nuclei; the quantity of these rays and their temporal emission (few nanoseconds) allow to perform a range verification during treatment with the PGI. Several research groups are evaluating different approaches to realize a prompt gamma imaging system suitable for the use in clinical condition and the optimization of a gamma-ray detector for PGI is still ongoing.
The Gammarad project works in this direction and aims to develop an high-performance and solid-state gamma ray detection module (GDM) with a slit camera design. The project is based on a collaboration among Fondazione Bruno Kessler (FBK, Trento, Italy), Politecnico di Milano (Milano, Italy), the Trento Institute for Fundamental Physics and Applications (TIFPA, Trento, Italy ), and the Proton Therapy Center of Trento (Italy). The project is divided into two parts. The first part focuses on the technological development of a gamma-ray imaging module. This module is composed by a gamma-ray detector, based on a solid-state silicon sensor, and an integrated circuit. They are assembled into a compact module with data and control systems. The second part of the project will be dedicated to the experimental validation of the system both in laboratory with radioactive sources and in a real environment, that of proton therapy.
The most innovative part of the gamma-ray detector developed for the project is the photo-sensor used for the scintillation light readout. In traditional applications it is a photomultiplier tube (PMT). However, in recent years, Silicon Photomultiplier (SiPM) has become increasingly popular in a variety of applications for its promising characteristics. Among them, current-generation SiPMs offer high gain, high Photon Detection Efficiency (PDE), excellent timing performance, high count-rate capability and good radiation hardness. Due to these characteristics they are used as PMTs replacement in several applications, such as in nuclear medicine (PET), in high-energy physics (calorimeters), astrophysics (Cherenkov telescopes) and in others single-photon or few-photon applications.
For its characteristics, the SiPM is also very promising for the scintillator readout in prompt gamma imaging and in high energy gamma-ray spectroscopy. Detectors for these applications must be compact, robust, and insensitive to the magnetic field. They have to provide high performance in terms of spatial, temporal, and energy resolution. SiPMs can satisfy all these requirements but typically they have been used with relatively low energy gamma rays and low photon flux, so manufacturers have optimized them for these conditions. Because of the limited number of micro-cells in a standard SiPM, 625 cells/mm^2 with 40 µm cells, the detector response is non-linear in high energies condition. Increasing the cell density is extremely important to improve the linearity of the SiPM and to avoid the compression of the energy spectrum at high energies, which worsens the energy resolution and makes difficult the calibration of the detector. On the other hand, small cells provide a lower Photon Detection Efficiency (PDE) because of the lower Fill Factor (FF) and as a consequence a lower energy resolution. Summarizing, the energy resolution at high energies is a trade-off between the excess noise factor (ENF) caused by the non-linearity of the SiPM and the PDE of the detector. Moreover, the small cell size provides an ultra-fast recovery time, in the order of a few of nanosecond for the smallest cells. A short recovery time together with a fast scintillator such a LYSO, reduces pile-up in high-rate applications, such as PGI.
Based on the above considerations, the aim of this thesis is to develop an optimized gamma-ray detector composed of SiPMs for high-dynamic-range application, such as the scintillation light readout in prompt gamma imaging and in high-energy gamma-ray spectroscopy. SiPMs evaluated for the detector are High-Density (HD) and Ultra-High-Density (UHD) SiPM technologies recently produced at Fondazione Bruno Kessler (FBK). Instead of standard SiPMs, HD and UHD SiPMs have a very small micro-cell pitch, from 30 µm down to 5 µm with a cell density from 1600 cells/mm^2 to 46190 cells/mm^2, respectively.
HD SiPMs are produced using a lithography technology with smaller critical dimensions and designed with trenches among SPADs. Small cells have a lower gain which helps to reduce correlated noise, such as After-Pulse and Cross-Talk. Trenches provide an optical and electrical cell isolation, and a smaller dead border around cells which increase the FF limiting PDE losses. UHD SiPMs push the limits of the HD technology even further, by reducing all the feature sizes, such as contacts, resistors and border region around cells. UHD SiPMs have hexagonal cells in a honeycomb configuration which generate a circular active area and a dead border around cells lower than 1 µm. The reduction of this dead boarder can improve the FF in smaller cells although it usually decrease with cell sizes. It is necessary understand how these significant layout changes affect the optical properties of SiPMs to evaluate which SiPM technology provides best performance in high-energy gamma-ray applications.
In the first part of the thesis, I presents the characterization of HD and UHD SiPM technologies in terms of PDE, gain, Dark Count Rate, and correlated noise for the cell sizes between 30 and 7.5 µm. The most important markers of SiPMs performance in gamma-ray spectroscopy are however the energy resolution and the linearity when coupled to the scintillator for the detection of high-energy gamma-rays. A typical characterization of the energy resolution of SiPMs, coupled to scintillator crystals, is performed with radioactive source up to 1.5 MeV. However, PGI features gamma ray-energies up to 15 MeV which are not easily provided by the usual laboratory calibration sources. Extrapolating the behaviour of the detector from the "low" energy data is not correct and leads to unreliable data for calibration and performance estimation. Therefore, I developed a novel setup that simulates the LYSO light emission in response to gamma photons up to 30 MeV. A LED (emitting at 420 nm) is driven by a pulse generator, emulating the light emitted by a LYSO scintillator when excited by gamma rays. The pulse generator parameters (amplitude, duration, rise and fall time constants) are adjusted so that the LED emitted photons match the intensity and time distribution of the LYSO emission. The photon number in each light pulse is calibrated from the measurements at 511 keV obtained with a ^(22)Na source and a LYSO crystal coupled to the SiPMs. Using this LED setup I characterized the energy resolution and non-linearity of HD and UHD SiPMs in high-energy gamma-ray conditions. The second part of the thesis provides a detailed description of the scintillator setup and of the setup for the simulation of high-energy gamma-ray response, followed by the results of the characterization performing with these setups.
Summarizing the results, the lowest non-linearity is provided by the technology with highest cell density, the RGB-UHD. For the 10 and 12.5 µm-cells we obtained values of 4.5% and 5% respectively at 5 MeV and 6 V over-voltage. On the other hand, we measured the best energy resolution of 2.6% and 2.3% at 5 MeV for the largest SiPM cells of 20 and 25 µm respectively, without the intrinsic term of the scintillator crystal and at 6 V over-voltage. This is due to the dependence of the energy resolution on the photon detection efficiency, which increases with the size of the SiPM cell. The optimal performance of the detector in high-dynamic-range applications, which depends on the several SiPM parameters, such as excess noise factor, photon detection efficiency, and cell sizes of the SiPM, is a trade off between non-linearity and energy resolution. At 5 MeV, the best trade-off for prompt gamma imaging application is reached by the 15 µm-cell. At 10 MeV the 12.5 µm-cell provides the best trade-off, because of the higher number of photons emitted by the scintillator. Furthermore, I distinguish the different components of the energy resolution (intrinsic, statistical, detector and electronic noise) as a function of cell sizes, over-voltage and energy, thanks to the combination of the scintillator and LED setups. The estimation of the intrinsic contribution of the scintillator crystal, coupled to the HD SiPMs, getting consistent results among the several cell sizes.
On the basis of previous characterization, HD SiPMs with dimensions of 4x4 mm^2 and 15 µm-cell were chosen to produce the photo-detector module of the gamma camera, optimized for an energy range between 2 and 8 MeV. This module is a 8x8 array of SiPMs which is called tile. The production of the tile requires research on packaging techniques to solve two main challenges: the maximization of the photo-sensitive area and the application of a protective resin, transparent in the near UV to maximize light collection from the LYSO. After some R&D on packaging, I obtained a fully functional tile with 64 SiPMs with a fill factor, ratio between the photo-sensitive area and the total area, of about 86%. This fill factor is comparable to the values obtained when a Through Silicon Vias (TSVs) technique is used to connect SiPMs but without the high production cost and the additional fabrication process complexity of the TSV. It should be highlighted that packaging operations is very critical because it is necessary to produce a tile with all working SiPMs, since defective items can not be replaced in the tile. The last part of the thesis presents the packaging procedure that I have defined to produce photo-detector modules and the characterization of the photo-detector array in terms of energy resolution, position sensitive and non-linearity. The measurements on the tile were carried out jointly with the Gammarad partner of Politecnico di Milano, which provided the ASIC and DAQ for the readout.
In conclusion, the R&D activity carried out during this thesis has provided to Gammarad project the final photo-detection module with state of the art performance for high-energy gamma-ray spectroscopy. The characterization of the module shows also a position sensitivity that matches with the SiPM dimensions, and a proper acquisition of high-energy gamma-ray events from 800 keV to 13 MeV. This module will be tested on beam in an experimental treatment room at the Proton therapy facility in Trento by the Gammarad project partners
Investigating and modelling the interaction among vegetation, hydrodynamics and morphology
Get PDFThe dissertation presented in this manuscript contributes to river science by providing a detailed overview on the state of the art on the interaction between riparian vegetation and hydrogeomorphological processes, by devising a novel model encompassing most of such processes and by proposing a field methodology aimed at providing means for improving the modelling of such
interactions. The state of the art is summarized in an extensive review describing riparian vegetation
and hydrogeomorphological processes mutual feedbacks. Such review did not simply seek to describe these feedbacks but, compiling from a large array of results from field, laboratory and modelling studies, provides a set of physical thresholds that trigger system changes. Therefore, processes are not only described terms but also explained with a quantitative approach. Processes description provided the conceptual foundation for the development of the novel simulation model while model parameterization was based on the quantitative information collected in the review. Such novel model, encompasses the main relationships entwining riparian woody vegetation and hydrogeomorphological processes and is able of replicating long term riparian landscape dynamics considering disturbance events, environmental stressor and riparian woody vegetation establishment from seeds and large wood. The manuscript presents the model structure and its conceptual validation by means of hydrological scenarios aimed at testing the coherence of the simulation results with expected system behaviour.
Examples of such coherences are vegetation growth rate in response to hydrological regime, entrainment and establishment of large wood in an unconfined river system and vegetation effect on erosion and deposition patterns.
Analysis of sedimentation patterns from the modelled results suggested that vegetation flow resistance
should be modelled with greater detail. These conclusions pointed the dissertation research towards the testing of a novel class of vegetation flow resistance equations, proposed by different authors, able of describing woody vegetation flow resistance on a physical basis. These equations have the advantage of considering flow stage, plants foliation level and species-specific flexibility. However, the use of such equations is limited by the difficulty of measuring the vegetation properties required as equation-inputs. In order to test if these equations could effectively improve sediment dynamics
predictions, a field method was formulated and tested. The field method allows to sample vegetation properties that can be used with these novel class of flow resistance equations. In the manuscript, such method is applied and the resulting vegetation properties used in several modelling scenarios. Such scenario proved that hydraulic variables modelled with these novel flow resistance approaches are more realistic and thus that the model developed during the dissertation could benefit from inclusion of such flow resistance equations in its source code
Deep Learning for Distant Speech Recognition
No full textDeep learning is an emerging technology that is considered one of the most promising directions for reaching higher levels of artificial intelligence.
Among the other achievements, building computers that understand speech represents a crucial leap towards intelligent machines. Despite the great efforts of the past decades, however, a natural and robust human-machine speech interaction still appears to be out of reach, especially when users interact with a distant microphone in noisy and reverberant environments. The latter disturbances severely hamper the intelligibility of a speech signal, making Distant Speech Recognition (DSR) one of the major open challenges in the field.
This thesis addresses the latter scenario and proposes some novel techniques, architectures, and algorithms to improve the robustness of distant-talking acoustic models. We first elaborate on methodologies for realistic data contamination, with a particular emphasis on DNN training with simulated data. We then investigate on approaches for better exploiting speech contexts, proposing some original methodologies for both feed-forward and recurrent neural networks. Lastly, inspired by the idea that cooperation across different DNNs could be the key for counteracting the harmful effects of noise and reverberation, we propose a novel deep learning paradigm called “network of deep neural networks”.
The analysis of the original concepts were based on extensive experimental validations conducted on both real and simulated data, considering different corpora, microphone configurations, environments, noisy conditions, and ASR tasks
Persistence and Adaptation of Pseudomonas Aeruginosa in cystic Fibrosis Airway
Get PDFBackground. Infections caused by Pseudomonas aeruginosa are the main cause of morbidity and mortality in Cystic Fibrosis (CF) patients and occur via primary colonisation
of the airway followed by the accumulation of pathoadaptive mutations in the bacterial genome which increase fitness in the lung environment and result in chronicization. A better understanding of i) the evolutionary dynamics occurring during chronic airway
infections in CF patients and ii) the genetic adaptation of strains to the CF lung environment, might give further clues for preventive measures or novel therapies to
control CF infections in the future. In this work, we obtained genomic sequences of 40 P. aeruginosa isolates from a single CF patient collected over an eight-year period (2007-2014) and analysed the population in terms of clonality of the isolates, phylogenetic relationships, and presence of polymorphisms and variants between the strains. Population structure and microevolution. In silico Multilocus Sequence Typing (MLST) analysis revealed a characteristic single clonal population dominated by a previously characterized sequence type (ST390) and a small number of new, closely related
ST variants (ST1863, ST1864, ST1923). EBURST analysis of the sequence types revealed that all members of this population belong to the same clonal lineage and likely evolved from a single ancestral colonizing strain. Furthermore, the phylogenetic analysis based on
SNPs also divided the population into two subpopulations derived from the evolution of the first infecting strain.
The annotation of SNPs allowed us to identify mutations with moderate or high impact. Genes with high impact variants encoded respiratory nitrate reductase subunit
gamma nail, polyprotein signal peptidase lspA, the ABC transporter-binding protein aaltP, the copper resistance protein A precursor pcoAin, and four hypothetical proteins. The evolution of strains in the CF airway is characterized by the loss of many virulence traits, including motility and protease secretion, along with the acquisition of multidrug resistance. Functional phenotypic assays of the collection, including motility and secretion of proteases, showed a decrease over time in the persistent isolates. We also determined
the antibiotic susceptibility profile of the collection; while early isolates were found to be susceptible to almost all these antibiotics, resistant phenotypes dramatically increased over time in the population.
Functional studies on specific strains. To identify additional functional variations related to pathoadaptive mutations occurring in the course of chronic infection in CF, we then selected three isolates for further characterization: one early CF isolate (TNCF_23
isolated in 2007); one clonal late CF isolate (TNCF_175 isolated in 2014); one clinical isolate (VrPa97) from a non-CF patient belonging to the same sequence type (ST390) as the former isolates. With this approach, we aimed to identify additional phenotypic and
functional variations between isolates with a very homogeneous genomic background, in an attempt to find out new pathoadaptive mutations occurring in the course of chronic infection in CF. Specifically, the following traits were investigated: killing of C. elegans and G. mellonella (in vivo virulence); immunomodulatory properties (IL-8 ELISA assay); competitive growth
in Artificial Sputum Medium (ASM); functionality of Type Six Secretion System (T6SS). Despite their close genetic relatedness, considerable variations were observed between the three isolates, among which the late isolate TNCF_175 showed several alterations 7 putatively resulting from the adaptation process to the CF lung. TNCF_175 presented a mutation in tssK3, part of H3-T6SS; this mutation (C958T) was therefore introduced in
the reference strains PAO1 and PA14, and mutated strains were subsequently complemented; killing rate on C. elegans and growth rate in ASM in mutant and complemented strains were evaluated. Conclusions. A rare feature of this strain collection is the consistent number of clonal
isolates obtained from a single patient over a rather long period of 8 eight years, thus providing a model to look at microevolutionary trends within a highly homogenous bacterial population, and avoiding potential biases due to the host genetic background and
clinical history. In spite of the close genomic relatedness of all isolates, a surprisingly high
diversity was observed for the majority of tested phenotypes. Investigating the competitive
ability of early versus late strains we propose a role for T6SS in the adaptation process to the CF lung environment. Our data suggest that once persistence has been established, a strain no longer requires its T6SS, allowing loss of function mutations to occur.
Conversely, acute and early CF strains still carry a number of virulence factors, including T6SS that potentially provide an advantage in outcompeting other microorganisms in the initial stage of CF infection
Digital Physics Education - Personal devices, mediated reality, serious gaming
Get PDFIn this work we discuss some of the most interesting new doors that the diffusion of low-cost, high-tech devices is opening. It is divided in two main sections.
In the first part, Chapter I, the general research framework in which we are working is presented. Then, in the second, third and fourth chapters several examples of practical applications regarding the presented topics, all designed and realized by the author (within his research group) are discussed. In particular: a workshop about scientific visualization held at the University of Trento in Chapter II; a teaching/learning sequence based on a low-cost spectrometer in Chapter III; a mediated reality setup for physics education in Chapter IV
Advanced regression and detection methods for remote sensing data analysis
Get PDFNowadays the analysis of remote sensing data for environmental monitoring is fundamental to understand the local and global Earth dynamics. In this context, the main goal of this thesis is to present novel signal processing methods for the estimation of biophysical
parameters and for the analysis icy terrain with active sensors. The thesis presents three main contributions.
In the context of biophysical parameters estimation we focus on regression methods.
According to the analysis of the literature, most of the regression techniques require a relevant number of reference samples to model a robust regression function. However, in real-word applications the ground truth observations are limited as their collection leads to high operational cost. Moreover, the availability of biased samples may result in low estimation accuracy. To address these issues, in this thesis we propose two novel contributions.
The first contribution is a method for the estimation of biophysical parameters that integrates theoretical models with empirical observations associated to a small number of in-situ reference samples. The proposed method computes and correct deviations between estimates obtained through the inversion of theoretical models and empirical observations.
The second contribution is a semisupervised learning (SSL) method for regression defined in the context of the ε-insensitive SVR. The proposed SSL method aims to mitigate the problems of small-sized biased training sets by injecting priors information in the initial learning of the SVR function, and jointly exploiting labeled and unlabeled samples in the learning phase of the SVR.
The third contribution of this dissertation addresses the clutter detection problem in radar sounder (RS) data. The capability to detect clutter is fundamental for the interpretation of subsurface features in the radargram. In the state of the art, techniques that require accurate information on the surface topography or approaches that exploit complex multi-channel radar sounder systems have been presented. In this thesis, we propose a novel method for clutter detection that is independent from ancillary information and limits the hardware complexity of the radar system. The method relies on the interferometric
analysis of two-channel RS data and discriminates the clutter and subsurface echoes by modeling the theoretical phase difference between the cross-track antennas of the RS. This allows the comparison of the phase difference distributions of real and simulated data. Qualitative and quantitative experimental results obtained on real airborne SAR and RS data confirm the effectiveness of the proposed methods
Integration in times of crisis. Testing Neofunctional hypotheses: a political economy investigation of crisis-led integration.
No full textFor more than 10 years, from the launch of the Single Currency to the global financial crisis, the process of deepening European integration stagnated, while emphasis was on widening the borders of the Union. In the wake of the Eurocrisis, however, two phenomena have captured the attention of political economists interested in European integration: the outstanding amount of new institutions, policies, and legislation which have been set in place to counter the crisis, and the rising popular rejection of the very concept of European unity.
The old functionalist adage that “integration advances through crises” appears to be, prima facie, corroborated; nevertheless, the dynamics of (political) fragmentation seems to follow a similar pattern, as Postfunctionalists would expect. Two interrelated questions emerge: is the Eurocrisis a true functional crisis? Did the Eurocrisis trigger a new, “transformative” cycle of integration, embodied in its Postfunctional dynamics? In the attempt to address this research puzzle, this doctoral dissertation attempts to operationalize the research problem through six standalone papers clustered in two parts. In addition, Chapter 1 reconstructs the key elements of systemic functionalism, guiding the reader through the theoretical pillars of this work presenting its overall logic – questions, methodologies, and chapter connections. Part One of the dissertation (Chapters 1-4) deals with the fundamental question concerning the
Neofunctional nature of the Eurocrisis, attempting to clarify to what extent the Eurocrisis can be really qualified as a “functional crisis”. Chapter 2 provides a comprehensive historical account of functional crises through the first 60 years of European Integration; Chapter 3 provides an econometric test of the endogenous nature of the Eurocrisis; Chapter 4 provides a qualitative assessment of the functional nature of the institutions introduced during the crisis.
Part Two of the dissertation (Chapters 5-7) dives into the possible Postfunctional implications of the Eurocrisis, analyzing to what extent the crisis is contributing to create a mobilized European public sphere. In particular, Chapter 5 provides a theoretical analysis of how the crisis is changing the legitimacy of the EU; Chapter 6 provides an econometric assessment of the crisis’ impact on citizens’ preferences for further economic integration; finally, Chapter 7 provides an analysis of the crisis’ impact on the performance of extreme Eurosceptic parties
Il dilemma dell'eroe. Modelli evoluzionistici di analisi narrativa
Get PDFDopo una panoramica sull’analisi narrativa nel suo sviluppo storico, a partire dall’interpretazione allegorica greca, passando per l’esegesi cristiana, fino allo strutturalismo, l’ermeneutica, la psicanalisi, il femminismo e la decostruzione, il primo capitolo si avvale del contributo di David Bordwell (Making Meaning, 1989), per mostrare il limite del meccanismo cognitivo comune a tutti i modelli ermeneutici: la mancanza di un fondamento epistemologico solido e di un metodo condiviso, che permetterebbe di non costruire ogni volta un edificio concettuale diverso e intercambiabile, ma di far sì che una nuova teoria narrativa abbia un potere esplicativo superiore e non semplicemente alternativo rispetto a quello delle teorie che l’hanno preceduta. Tale fondamento e metodo vengono individuati nel secondo capitolo grazie al ricorso alle scienze evoluzionistiche, soprattutto biologia e neurobiologia. Come sintetizzato da Brian Boyd (On the Origin of Stories: Evolution, Cognition and Fiction, 2009) l’Homo sapiens non si è trovato capace e bisognoso di raccontare e ascoltare storie dall’oggi al domani. La funzione narrativa si è evoluta come qualsiasi altra caratteristica della nostra specie in funzione del mantenimento e della trasmissione della vita, probabilmente in quanto gioco cognitivo, meccanismo neuronale che permette al cervello umano di orientarsi nel tempo dando ordine e senso agli eventi. Adottandone il metodo e basandosi su questo e molti altri risultati delle scienze evoluzionistiche, l’analisi narrativa può finalmente cominciare a costruire su un fondamento solido, sottoponendo i propri risultati teorici al vaglio della verifica quantitativa. A questo è dedicato il terzo capitolo, aperto su una nuova prospettiva di ricerca, che attraverso l’applicazione dei modelli matematici della teoria dei giochi permette di individuare un metodo per verificare l'ipotesi evoluzionistica che la cultura umana, e le storie che nel corso del tempo questa ha raccontato, si siano evolute in una direzione ben precisa, quella della cooperazione (Robert Axelrod, The Evolution of Cooperation, 1984). La teoria dei giochi è infatti nata come studio delle strategie di comportamento economico attraverso l'elaborazione di modelli matematici, ma si presta (ed è stata con successo applicata) a qualunque tipo di interazione umana e non, economica, biologica, sociale, culturale, etica e in questo caso narrativa
Optimal Codes and Entropy Extractors
Get PDFIn this work we deal with both Coding Theory and Entropy Extraction for Random Number Generators to be used for cryptographic purposes. We start from a thorough analysis of known bounds on code parameters and a study of the properties of Hadamard codes. We find of particular interest the Griesmer bound, which is a strong result known to be true only for linear codes. We try to extend it to all codes, and we can determine many parameters for which the Griesmer bound is true also for nonlinear codes. In case of systematic codes, a class of codes including linear codes, we can derive stronger results on the relationship between the Griesmer bound and optimal codes. We also construct a family of optimal binary systematic codes contradicting the Griesmer bound. Finally, we obtain new bounds on the size of optimal codes. Regarding the study of random number generation, we analyse linear extractors and their connection with linear codes. The main result on
this topic is a link between code parameters and the entropy rate obtained by a processed random number generator. More precisely, to any linear extractor we can associate the generator matrix of a linear code. Then, we link the total variation distance between the uniform distribution and the probability mass function of a random number generator with the weight distribution of the linear code associated to
the linear extractor. Finally, we present a collection of results derived while pursuing a way to classify optimal codes, such as a probabilistic algorithm to compute
the weight distribution of linear codes and a new bound on the size of codes
Gray matter covariance networks in the mouse brain
No full textThe presence of networks of correlation between gray matter volumes of brain regions - as measured across subjects in a group of individuals - has been consistently described in several human studies, an approach termed structural covariance MRI (scMRI). Complementary to prevalent brain connectivity modalities like functional and diffusion-weighted imaging, this approach can provide valuable insight into the mutual influence of regional trophic and plastic processes occurring between brain regions.
Previous investigations highlighted coordinated growth of these regions within specific structural networks in healthy populations and described their derangement in pathological states. However, a number of fundamental questions about the origin and significance of these couplings remains open and the mechanisms behind the formation of scMRI networks are still poorly understood. To investigate whether analogous scMRI networks are present in lower mammal species amenable to genetic and experimental manipulation such as the laboratory mouse, I coupled high resolution morpho-anatomical MRI with network-based approaches on a large cohort of genetically-homogeneous wild-type mice (C57Bl6/J). To this purpose, I first developed a semi-automated pipeline enabling reliable Voxel Based Morphometry (VBM) of gray matter volumes in the mouse. To validate this approach and its ability to detect plastic changes in brain structures, I applied it to a cohort of aged mice treated with omega-3 polyunsaturated fatty acids (n3-PUFA). This study revealed that treatment with n3PUFA, but not isocaloric olive oil preserved gray matter volume of the hippocampus and frontal cortices, an effect coincident with amelioration of hippocampal-based spatial memory functions. I next employed VBM to investigate scMRI networks in inbred mice using a seed-based approach. In striking resemblance with human findings, I observed the presence of homotopic (i.e. bilateral) architecture in several scMRI cortical and subcortical networks, a finding corroborated by Independent Component Analyses. Subcortical structures also showed highly symmetric inter-hemispheric correlations, with evidence of distributed antero-posterior networks in diencephalic regions of the thalamus and hypothalamus. Hierarchical cluster analysis revealed six identifiable clusters of cortical and sub-cortical regions corresponding to previously described neuroanatomical systems. This work documents for the first time the presence of homotopic cortical and subcortical scMRI networks in the mouse brain, and is poised to pave the way to translational use of this species to investigate the elusive biological and neuroanatomical underpinnings of scMRI network development and its derangement in neuropathological states