Approved for public release, distribution is unlimited / "This work accomplishes signal denoising using the Bootstrap method when the additive noise is Gaussian. The noisy signal is separated into frequency bands using the Fourier or Wavelet transform. Each frequency band is tested for Gaussianity by evaluating the kurtosis. The Bootstrap method is used to increase the reliability of the kurtosis estimate. Noise effects are minimized using a hard or soft thresholding scheme on the frequency bands that were estimated to be Gaussian. The recovered signal is obtained by applying the appropriate inverse transform to the modified frequency bands. The denoising scheme is tested using three test signals. Results show that FFT-based denoising schemes perform better than WT-based denoising schemes on the stationary sinusoidal signals, whereas WT-based schemes outperform FFT-based schemes on chirp type signals. Results also show that hard thresholding never outperforms soft thresholding, at best its performance is similar to soft thresholding."--p.i. / First Lieutenant, Turkish Army

Master-Arbeit Univ. St. Gallen, 2008.

Several methods have been proposed to estimate the kurtosis of a distribution. The three common estimators are: g2, G2 and b2. This thesis addressed the performance of these estimators by comparing them under the same simulation environments and conditions. The performance of these estimators are compared through confidence intervals by determining the average width and probabilities of capturing the kurtosis parameter of a distribution. We considered and compared classical and non-parametric methods in constructing these intervals. Classical method assumes normality to construct the confidence intervals while the non-parametric methods rely on bootstrap techniques. The bootstrap techniques used are: Bias-Corrected Standard Bootstrap, Efron’s Percentile Bootstrap, Hall’s Percentile Bootstrap and Bias-Corrected Percentile Bootstrap. We have found significant differences in the performance of classical and bootstrap estimators. We observed that the parametric method works well in terms of coverage probability when data come from a normal distribution, while the bootstrap intervals struggled in constantly reaching a 95% confidence level. When sample data are from a distribution with negative kurtosis, both parametric and bootstrap confidence intervals performed well, although we noticed that bootstrap methods tend to have smaller intervals. When it comes to positive kurtosis, bootstrap methods perform slightly better than classical methods in coverage probability. Among the three kurtosis estimators, G2 performed better. Among bootstrap techniques, Efron’s Percentile intervals had the best coverage.

No description available.

No description available.

L’accident vasculaire cérébral ischémique (AVCi) représente un véritable enjeu de santé publique avec des taux de mortalité et des coûts de prise en charge élevés. L’établissement d’un diagnostic rapide et précis basé sur les signes cliniques et les résultats de l’imagerie médicale est nécessaire pour la prise de décision thérapeutique (thrombolyse ou plus récemment la thrombectomie mécanique). Conformément aux recommandations actuelles, l’IRM est la modalité d’imagerie de première intention à réaliser pour confirmer la suspicion d’AVCi. La réalisation d’une thrombolyse est notamment guidée par l’existence d’une discordance (ou « mismatch ») diffusion-perfusion, déterminée à partir de deux séquences distinctes, associée à un meilleur pronostic pour le patient. La séquence IVIM (« Intravoxel Incoherent Motion») permet l’étude simultanée de la diffusion et de la microcirculation à partir de l’analyse biexponentielle du signal issu d’une unique séquence de diffusion sensibilisée à la perfusion. Par conséquent, cette séquence présente un fort potentiel pour le diagnostic de l’AVCi. Le travail méthodologique de cette thèse a consisté en l’optimisation des acquisitions et des traitements de données pour l’imagerie quantitative simultanée de la diffusion et de la perfusion cérébrale avec la méthode IVIM. Dans un premier temps, le modèle biophysique conventionnel et les acquisitions de la séquence IVIM ont été validés dans le cadre d’une étude sur sujet sain, en comparaison aux mesures de perfusion obtenues avec l’ASL (Arterial Spin Labeling). Dans un deuxième temps, le modèle conventionnel a été modifié pour la prise en compte du comportement non-gaussien de la diffusion dans le parenchyme cérébral à l’aide d’un paramètre quantitatif supplémentaire, le kurtosis (modèle DKI-IVIM). Ce modèle a été validé expérimentalement avec une étude sur sujet sain en comparaison au modèle standard biexponentiel IVIM. Enfin, la méthodologie développée a été mise en œuvre dans un cadre clinique sur cinq patients souffrant d’AVCi. Les résultats préliminaires obtenus démontrent l’efficacité de la méthode DKI-IVIM pour le diagnostic de l’AVCi en phase aigüe, en comparaison avec la méthode conventionnelle diffusion-perfusion ASL avec l’estimation du paramètre supplémentaire du kurtosis permettant une meilleure caractérisation de l’atteinte parenchymateuse. / Ischemic stroke is a serious neurological disease of public health concern that constitutes a major cause of death and high costs of medical care. A quick and accurate diagnosis based on both clinical signs and medical images is necessary for the therapeutic decision (thrombolysis or mechanical thrombectomy). In accordance with the current recommendations, MRI is the primary intention modality to perform in order to confirm the ischemic stroke suspicion. The choice of a thrombolysis treatment is guided by the presence of a diffusion-perfusion mismatch, determined with two different sequences, and is associated with a better life outcome for the patient. The IVIM (« Intravoxel Incoherent Motion ») MRI sequence allows for the simultaneous diffusion and microperfusion quantification with the biexponential analysis of the diffusion signal obtained by a diffusion sequence sensitized to perfusion. This sequence could be of great interest for the ischemic stroke diagnosis. The methodological aspects implemented during this thesis consisted of the optimization of the acquisitions and data processing of IVIM imaging for quantitative assesments of cerebral diffusion and perfusion. First of all, the conventional biophysical model and IVIM sequence acquisitions were implemented and validated with a study on healthy subjects, in comparison with perfusion assesments obtained using ASL (Arterial Spin Labeling). Secondly, the conventional model was modified to consider the non-gaussian diffusion behavior in tissues with an additional quantitative parameter estimation, the kurtosis (DKI-IVIM model). This new model was also validated with a study on healthy subjects in comparison with the standard biexponential IVIM model. Finally, this method was applied in a clinical setting on five stroke patients. The preliminary results demonstrated the DKI-IVIM method efficiency for the acute ischemic stroke diagnosis when compared with the conventional diffusion-ASL perfusion with the additional estimation of the kurtosis for a better lesion characterization.

This article proposes a simple method to visualize peak and tails in continuous distributions with finite variance. The excess peak and tails areas in unimodal symmetric and nonsymmetric distributions, and the missing area in U-shaped distributions, are identified by comparing the distribution under consideration with the uniform distribution with equal center and variability. Agreement with kurtosis orderings based on the CDFs, and a strong correlation between the total peak and tails area with quantile kurtosis, were found for the distributions examined. The visualization of tails and peak could be used to introduce the notion of kurtosis in undergraduate statistics courses.

Over the past decade, there has been an increasing interest in the real-time monitoring of ambulatory vital signs such as heart rate (HR) and arterial blood oxygen saturation (SpO2) using wearable medical sensors during field operations. These measurements can convey valuable information regarding the state of health and allow first responders and front-line medics to better monitor and prioritize medical intervention of military combatants, firefighters, miners and mountaineers in case of medical emergencies. However, the primary challenge encountered when using these sensors in a non-clinical environment has been the presence of persistent motion artifacts (MA) embedded in the acquired physiological signal. These artifacts are caused by the random displacement of the sensor from the skin and lead to erroneous output readings. Several signal processing techniques, such as time and frequency domain segmentation, signal reconstruction techniques and adaptive noise cancellation (ANC), have been previously developed in an offline environment to address MA in photoplethysmography (PPG) with varying degrees of success. However, the performance of these algorithms in a spasmodic noise environment usually associated with basic day to day ambulatory activities has still not been fully investigated. Therefore, the focus of this research has been to develop novel MA algorithms to combat the effects of these artifacts. The specific aim of this thesis was to design two novel motion artifact (MA) algorithms using a combination of higher order statistical tools namely Kurtosis (K) for classifying 10 s PPG data segments, as either ‘clean’ or ‘corrupt’ and then extracting the aforementioned vital parameters. To overcome the effects of MA, the first algorithm (termed ‘MNA’) processes these ‘corrupt’ PPG data segments by identifying abnormal amplitudes changes. The second algorithm (termed ‘MNAC’), filters these ‘corrupt’ data segments using a 16th order normalized least mean square (NLMS) ANC filter and then extracts HR and SpO2.

Over the past decade, there has been an increasing interest in the real-time monitoring of ambulatory vital signs such as heart rate (HR) and arterial blood oxygen saturation (SpO2) using wearable medical sensors during field operations. These measurements can convey valuable information regarding the state of health and allow first responders and front-line medics to better monitor and prioritize medical intervention of military combatants, firefighters, miners and mountaineers in case of medical emergencies. However, the primary challenge encountered when using these sensors in a non-clinical environment has been the presence of persistent motion artifacts (MA) embedded in the acquired physiological signal. These artifacts are caused by the random displacement of the sensor from the skin and lead to erroneous output readings. Several signal processing techniques, such as time and frequency domain segmentation, signal reconstruction techniques and adaptive noise cancellation (ANC), have been previously developed in an offline environment to address MA in photoplethysmography (PPG) with varying degrees of success. However, the performance of these algorithms in a spasmodic noise environment usually associated with basic day to day ambulatory activities has still not been fully investigated. Therefore, the focus of this research has been to develop novel MA algorithms to combat the effects of these artifacts. The specific aim of this thesis was to design two novel motion artifact (MA) algorithms using a combination of higher order statistical tools namely Kurtosis (K) for classifying 10 s PPG data segments, as either ‘clean’ or ‘corrupt’ and then extracting the aforementioned vital parameters. To overcome the effects of MA, the first algorithm (termed ‘MNA’) processes these ‘corrupt’ PPG data segments by identifying abnormal amplitudes changes. The second algorithm (termed ‘MNAC’), filters these ‘corrupt’ data segments using a 16th order normalized least mean square (NLMS) ANC filter and then extracts HR and SpO2.

Recent research documents that voluntary disclosure—in particular, managerial forecast guidance—lowers uncertainty levels, as proxied by option implied variances. In this study I explore the effect of such voluntary disclosure on other dimensions of uncertainty. In particular, I investigate the effect of managerial guidance on the variance risk premium (VRP). Prior research predicts and provides empirical evidence of the VRP, which reflects that implied variances (on average) exceed actual variances, and exists to compensate traders, who sell variance protection for equity options. First, I confirm previous findings that implied variances are lower when firms issue management guidance. Second and more importantly, I document that the VRP is higher when firms provide guidance. I reconcile these seemingly contradictory results by (i) confirming that a significant portion of the increase in VRP is attributable to uncertainty specific to the impending earnings announcement, consistent with the primary role played by the voluntary management disclosure; and (ii) documenting that a higher moment of uncertainty—implied kurtosis levels (i.e., price jump risk)—is higher with managerial guidance. Additional tests examining characteristics of managerial guidance reveal these findings are strongest for firms issuing sporadic guidance, guidance issued close to earnings announcements, and those exhibiting the largest surprise. Overall, the evidence suggests that voluntary disclosure such as management guidance can reduce expected variance, but simultaneously increase higher order moments of uncertainty such as expected price jumps.