No / RATIONALE: 5,6-Dimethylxanthenone-4-acetic acid (DMXAA) is a low molecular weight drug of the flavonoid group, which has an anti-vascular effect in tumours causing endothelial cell apoptosis and activation of cytokines. Flavonoid-based compounds have been reported to lead to an upregulation in the expression of lysophosphatidylcholines (LPC)-type lipids in solid tumours. A study employing TLC/MALDI-MS and MALDI-MS imaging to examine LS174T colorectal adenocarcinoma xenografts following administration of DMXAA has been conducted into this effect. METHODS: LS174T colorectal adenocarcinoma xenografts grown in male immune-deficient mice were treated with 27.5 mg/kg DMXAA. The control (before treatment) and 4 h and 24 h post-treatment tumours were excised and divided into two. MALDI-MS imaging experiments were carried out on 12 microm cryosections sections taken from one half of the tumours and from the other half the lipids were extracted and analysed by TLC/MALDI-MS. These experiments were carried out in triplicate. RESULTS: Statistical analysis of the MALDI-MS imaging data set indicated an increased amount of LPC in the 24 h post-treated sample and a decreased amount of PC in the 24 h post-treated sample, compared with the 4 h post-treated sample and the control. These effects were confirmed by the TLC/MALDI-MS data. The lipid extracts were separated into six spots on the TLC plate. These were identified as arising from different lipids classes, i.e. LPC, sphingomyelins (SM), phosphatidylcholines (PC) and phosphatidylethanolamines (PE). The TLC/MALDI-MS data indicated that LPC were highly expressed in the 4 h and 24 h post-treated tumour samples compared with the control. Examination of the mass spectrometric images confirms this increase and demonstrates additionally that the increase in the signals arising from LPC appears to be localised primarily within the central areas of the xenograft. CONCLUSIONS: An increase in expression of LPC lipids in solid tumours treated with DMXAA has been demonstrated and shown to be localised in the central area of the tumour.

Les modèles à deux processus de la mémoire de reconnaissance postulent l’intervention de deux mécanismes indépendants, la familiarité et la récollection, là où les modèles à un processus envisagent un continuum. Aujourd’hui, les théories à deux processus sont les plus largement acceptées grâce à la découverte de deux composantes électrophysiologiques qui reflèteraient les processus de familiarité et de récollection, respectivement la FN400 et la LPC. Cependant, l’indépendance de ces deux composantes est sujette à débat. L’objectif principal de cette thèse a donc été d’examiner la nature de ces processus et de leurs corrélats neuronaux afin de tester leur indépendance. Pour ce faire, nous avons d’étudier l’influence des informations perceptives (études 1 et 2), sémantiques (études 1, 2, 3 et 4), associatives (études 1 et 2), ainsi que la répétition à l’encodage (étude 3), et l’effet de l’âge (étude 4), sur les processus de récollection, de familiarité et leurs corrélats respectifs en couplant les données ROC et EEG. Les résultats vont dans le sens d’une conception de la familiarité reflétée par la FN400 comme étant multi-déterminée et permettent de préciser le caractère conditionné et hiérarchique dans le traitement des informations : perceptives, sémantiques puis associatives. Nos données indiquent en revanche que la récollection et la LPC serait principalement influencée par des informations sémantiques et associatives et ensuite, dans un second temps, par des informations perceptives. Dans leur ensemble, ces résultats sont en faveur du modèle de redondance, suggérant que la familiarité et la récollection dépendent d’un processus commun, multi-déterminé et dynamique. / Dual-process models of recognition postulate that two independent processes sustained recognition (familiarity and recollection) while single process models consider the familiarity and the recollection as the ends of a continuum reflecting a common process. For now, dual-process models are the most accepted, due to electrophysiological evidences of two distinct components, FN400 and LPC, which have been linked to familiarity and recollection, respectively. However, a debate is growing concerning the independence of these components. The main objective of this thesis was to investigate familiarity, recollection and their neural correlates’ nature. To this end, we studied the influence of perceptual (study 1 and 2), semantic (all study), associative information (study 1 and 2), the repetition (study 3) and the age-related effect (study 4) on familiarity, recollection and corresponding neural correlates using ROC and EEG data. Results support the view that familiarity, as reflected by FN400, is multi-determined but allow to specify the conditioned the hierarchy in the information’s processing: perceptual then semantic and then associative. Recollection, as reflected by the LPC, would also be multi-determined, but was mainly influenced by semantic and associative information and then, in a secondary way, by perceptual information. Taken together, present data support the redundancy model suggesting that familiarity and recollection depend on a common, multi-determined and dynamic process.

Ziel dieser Arbeit war es unter Einsatz einer Auswahl an Algorithmen die Sprachqualität von CASA separierten Sprachsignalen zu verbessern. Der Ursprung dieses Problems liegt in der von CASA verwendeten Zeit-Frequenz-Maskierung. Diese kann zu einer fehlerbehafteten Ausgabe führen, wenn zur gleichen Zeit zwei Sprecher den selben Frequenzbereich mit ihren Stimmen abdecken. Der beschriebene Bereich kann folglich nur als Ganzes einem Sprecher zugeordnet werden, sodass fehlerbehaftete Signale entstehen. Zur Verbesserung dieser wurde der Einsatz eines LPC-Vocoders, die Reduktion der Abtastrate der Signale sowie eine Kombination beider als Reparaturverfahren eingesetzt und überprüft.

Denna rapport behandlar restspänning i stålet 16MnCrS5 som uppkommer efter sätthärdning med traditionell uppkolning med släckning i olja och lågtrycksuppkolning med släckning i gas. Rapporten har ett fokus på restspänningar men behandlar även materialegenskaper generellt. I rapporten utförs en karaktärisering av materialet där författarna tittar på, kolhalt, kolprofil, ythårdhet (Rockwell C), kärnhårdhet (Vickers10kg) och hårdhetsprofil (Vickers 1kg) samt analys av mikrostruktur. Även resultat för FWHM kommer tas upp som visat sig kunna relateras till hårdheten hos materialet. Diskussionen handlar om att förklara restspänningen i materialet med hjälp av resultatet från karaktäriseringen. Lågtrycksuppkolning har visat sig inte ge inre oxidation vid sätthärdning, medan provbitar som sätthärdats med traditionell sätthärdning har uppvisat inre oxidation. Den inre oxidationen kan bidra till tryckrestspänningar vid ytan på materialet. Hög andel restaustenit har generellt visat sig vara en bidragande faktor till dragrestspänningar främst vid ytan men även inne i sätthärdningsskiktet. Resultatet har också visat att prover som värmebehandlats med lågtrycksuppkolning och har hög halt restaustenit vid ytan kan också ha hög nivå av dragrestspänning vid ytan. / This report addresses residual stress in the steel 16MnCrS5 that occurs after carburization with traditional carburization with quenching in oil and low-pressure carburization with quenching in gas. The report focuses on residual stresses but also deals with material properties in general. The report characterizes the material by looking at the, carbon content, carbon profile, surface hardness (Rockwell C), core hardness (Vickers10kg) and hardness profile (Vickers 1kg) as well as analysis of the microstructure. Low pressure carburization has been shown not to produce internal oxidation, but samples heat treated with traditional carburization has exhibited internal oxidation. High proportion of residual austenite has also been shown to be a contributing factor to tensile residual stresses, mainly at the surface but also within the curing layer. The result have also shown that residual austenite at the surface have been associated with high levels of tensile residual stress at the surface.

The objective of this thesis is to reduce the training size for the Mandarin address inputting system and the Mandarin phrase recognition system. A set of two-word Mandarin phrases is developed by the balanced sieving and mixture training techniques. This greatly reduces the training data size for the systems. Hidden Markov model using both MFCC and LPCC features is proposed in this thesis. Speech-text alignment, frame overlapping and tone recognition are incorporated to increase the correct recognition rates. For the speaker-dependent case, any phrase in these two speech systems can be recognized within one second.

This thesis compares the speech quality of Code Excited Linear Predictor (CELP, Federal Standard 1016) speech coder with a new wavelet method to compress speech. The performances of both are compared by performing subjective listening tests. The test signals used are clean signals (i.e. with no background noise), speech signals with room noise and speech signals with artificial noise added. Results indicate that for clean signals and signals with predominantly voiced components the CELP standard performs better than the wavelet method but for signals with room noise the wavelet method performs much better than the CELP. For signals with artificial noise added, the results are mixed depending on the level of artificial noise added with CELP performing better for low level noise added signals and the wavelet method performing better for higher noise levels.

The diploma thesis Detection of Alcohol in Speech Signal first describes the effect of alcohol on the human body. The second part deals with ways to obtain parameters that describe the speech signal. The third part provides a brief overview of previous case studies and patents focused on the detection of alcohol in the speech signal. The fourth part presents the collected own database of voice recordings and developed software application for the analysis of intoxicated speech. The final part describes the measured changes in speech signal parameters that indicate alcohol intoxication.

Nous proposons d'explorer électrophysiologiquement la question de l'accès au sens. Pour cela, nous nous intéressons particulièrement à deux ERPs. La composante N400 est classiquement associée à la prise en compte d'un contexte sémantique. La composante P600, moins explorée, est retrouvée lors de tâches syntaxiques, sémantiques ou non spécifiquement linguistiques. Nous avons décrit trois études électrophysiologiques en conditions naturelles de langage. La première étude nous a permis de valider l'usage de ce matériel original dans une population contrôle. Puis nous avons proposé le transfert clinique de ce protocole en conditions naturelles de langage auprès de patients souffrant de TAB (deuxième étude) et de schizophrénie (troisième étude, en cours). Nous avons retrouvé une composante N400 préservée dans les TAB, alors que les patients souffrant de schizophrénie présentent un profil ERP altéré, avec une compréhension perturbée des mots comme congruents au contexte sémantique. Après nous être interrogés sur la notion de transfert clinique, et la prudence nécessaire à l'interprétation de ces résultats en pratique clinique quotidienne, nous avons envisagé deux autres approches relatives à l'accès au sens. D'une part, nous avons repris un modèle linguistique issu des perspectives phénoménologique, structuraliste et morphodynamique, pour détailler un protocole expérimental en cours de réalisation, ainsi que des premiers résultats exploratoires obtenus en population contrôle. D'autre part, les données en première personne après entretien d'explicitation apportent une illustration subjective des phénomènes d'accès au sens mis en jeu dans les deux protocoles utilisés. / The aim of our projectwas to examine semantic access from an electrophysiological view. We specifically explored two Event-related Potentials (ERPs): N400, associated to semantic processing, and P600, associated to syntactic, semantic and non linguistic and general processes. In the first part of our project, we have explored these ERPs among healthy participants and psychiatric patients. People with schizophrenia exhibit thought and language disorders. In spite of common clinical manifestations with schizophrenia, patients with bipolar disorders (BDs) have rarely been assessed using neurolinguistic ERPs. We have conducted three electrophysiological studies in natural speech conditions. In the first study, we validated our original linguistic material among healthy participants. Then, in a clinical transfer perspective, we assessed manic patients suffering from BDs (second study) and patients with schizophrenia (third study, in preparation) using the same linguistic material. Our results support the hypothesis of a preserved N400 component in patients with BDs. On the contrary, preliminary results showed a disturbed N400 associated to congruent sentences among patients with schizophrenia.After discussing the limitations of the notion of clinical transfer and of our experimental studies, we also presented two alternative perspectives on speech comprehension. One the one hand, we described a linguistic model inherited from phenomenology, structuralism and morphodynamics. On the other hand, we reported first person data obtained through elicitation techniques. These experiential data enable us to better grasp semantic phenomena involved in our both experiments.

In the past years, several network reconstruction methods modeled as Gaussian Graphical Model in high dimensional settings where proposed. In this work we will analyze three different methods, the Graphical Lasso (GLasso), Graphical Ridge (GGMridge) and a novel method called LPC, or Local Partial Correlation. The evaluation will be performed in high dimensional data generated from different simulated random graph structures (Erdos-Renyi, Barabasi-Albert, Watts-Strogatz ), using Receiver Operating Characteristic or ROC curve. We will also apply the methods in the reconstruction of genetic co-expression network for the differentially expressed genes in cervical cancer tumors. / Vários métodos tem sido propostos para a reconstrução de redes em alta dimensão, que e tratada como um Modelo Gráfico Gaussiano. Neste trabalho vamos analisar três métodos diferentes, o método Graphical Lasso (GLasso), Graphical Ridge (GGMridge) e um novo método chamado LPC, ou Correlação Parcial Local. A avaliação será realizada em dados de alta dimensão, gerados a partir de grafos aleatórios (Erdos-Renyi, Barabasi-Albert, Watts-Strogatz ), usando Receptor de Operação Característica, ou curva ROC. Aplicaremos também os metidos apresentados, na reconstrução da rede de co-expressão gênica para tumores de câncer cervical.

In the past years, several network reconstruction methods modeled as Gaussian Graphical Model in high dimensional settings where proposed. In this work we will analyze three different methods, the Graphical Lasso (GLasso), Graphical Ridge (GGMridge) and a novel method called LPC, or Local Partial Correlation. The evaluation will be performed in high dimensional data generated from different simulated random graph structures (Erdos-Renyi, Barabasi-Albert, Watts-Strogatz ), using Receiver Operating Characteristic or ROC curve. We will also apply the methods in the reconstruction of genetic co-expression network for the differentially expressed genes in cervical cancer tumors. / Vários métodos tem sido propostos para a reconstrução de redes em alta dimensão, que e tratada como um Modelo Gráfico Gaussiano. Neste trabalho vamos analisar três métodos diferentes, o método Graphical Lasso (GLasso), Graphical Ridge (GGMridge) e um novo método chamado LPC, ou Correlação Parcial Local. A avaliação será realizada em dados de alta dimensão, gerados a partir de grafos aleatórios (Erdos-Renyi, Barabasi-Albert, Watts-Strogatz ), usando Receptor de Operação Característica, ou curva ROC. Aplicaremos também os metidos apresentados, na reconstrução da rede de co-expressão gênica para tumores de câncer cervical.