Micromachined biosensors are used in chemical and biological applications. A biosensor which uses mass based transduction is called a mass sensor. Mass sensors are used to detect extremely small mass of biomolecules such as proteins, viruses or even parts of DNA in the range of femtograms (10-15 gm) to zeptograms (10−21 gm). Highly effective and reliable microcantilevers are used for detecting the mass of biomolecules using either static deflection or dynamic resonant peak shifts. The main objective of our work is to investigate the effect of elasticity of the attached mass on the shift in the resonant frequency and examine the validity of the rigid mass assumption used in the literature. The natural frequencies of a resonator are either found by solving the governing differential equation or approximately using Rayleigh-Ritz method. The mass of a body, attached to a resonator beam is determined using resonant frequency shift method. In our study, we derive an analytical expression for ‘δm’ based on the shift in frequency ‘δf’ that accounts for the elasticity of the added mass and the location of the mass on the beam. We study the simplest model to incorporate these effects where the added mass is itself modeled as a single degree of freedom spring-mass system. The entire system is represented as a 2-DOF lumped model of cantilever and the attached elastic mass. The natural frequencies are obtained using eigenvalue analysis. We study the mass estimation of Escherichia Coli (E. Coli), a food borne pathogen, using experimental results reported in the literature. We treat E.Coli as an elastic mass and model it as a single degree of freedom system to account for its elasticity. We use the elastic model as well as the rigid mass model to check the results available in the literature and point out the difference that results in mass estimation using the two models. To demonstrate the effect of elasticity on mass sensing using the resonant peak shift technique, we conduct mesoscale experiments. Since the fundamental principle does not depend on any phenomenon exclusively dependent on micro scales, the mesoscale experiments are justified. For this purpose, an experimental set-up with metallic cantilevers and flexible rubber strands as attached masses are used. We also use our experimental set-up to study the effect of positional inaccuracy of the added mass (rigid) in the computation of its mass from the shift in the resonance frequency. The results obtained show that elasticity of the added mass as well as its position on the resonator affect the computed mass but this effect is dependent on the relative stiffness and mass of the resonator and the added mass. We also observe the limitations of the experiments in carrying out studies over the desired range of parameters. We also create a computational model of the system and carry out simulations to explore a larger range of parameter values. In particular, we create an FEM model of our system in ANSYS, and carry out modal analysis for the cantilever beam resonator with and without the added mass, varying the relative stiffness and mass of the two components (the cantilever beam and the added mass). We compare the results of shift in the resonant frequency with those obtained from the rigid mass model. The results show the effect of elasticity clearly in certain ranges of relative stiffness and mass.

Micro-mechanical sensors are typically fabricated both in large numbers and economically using the photolithographic processes that were originally developed in the integrated circuit industry. The magnitude of a change in resonant frequency of a micro-me chanical structure can be used to quantify a change in mass of such a device. Hence, when packaged with integrated measurement, actuation and control electronics, it is possible to deliver a low-cost and small system in a package using fabrication techniq ues that are both mature and widely available. A micro-mechanical resonator has been designed for this project and samples of the prototype resonator were used to investigate various methods for detecting a change in resonant frequency using discrete elec tronic components. The system that has been designed can eventually be integrated with a small micro-mechanical structure to create a mass sensor. Resonators have been fabricated at QinetiQ as part of the Europractice Foundry Access Program and characteri sation of typical devices is described in this thesis. A popular method for controlling the behaviour of resonant micro-mechanical sensors is a force feedback technique designed to increase the effective quality factor of the resonant system. In this thesis, an increase in the effective quality factor of the prototype system has been demonstrated. When the resonator operates in air at atmospheric pressure, an improvement in the effective quality factor of two orders of magnitude was achievable. This meant that it was possible to assess the potential benefits offered by the force feedback technique by testing the various detection schemes that have been implemented at the natural quality factor and also at a high effective quality factor. A prototype control system has been built using simple digital electroni cs, a key component of which is a direct digital frequency synthesis chip used to provide a stable and accurate driving frequency. Methods for determining a change in the resonant frequency of a micro-mechanical resonator using this control system have be en investigated. A method has been developed for determining the magnitude of a shift in resonance when the frequency of the excitation force is fixed. This thesis contains a description of the technique and also results demonstrating the corresponding de tection capability of the prototype sensor.

No description available.

Breast cancer is the most commonly diagnosed cancer among Canadian women; x-ray mammography is the leading screening technique for early detection. This work introduces a semi-automated technique for analyzing mammographic x-ray images to measure their degree of suspiciousness for containing abnormalities. The designed system applies the discrete wavelet transform to parse the images and extracts statistical features that characterize an images content, such as the mean intensity and the skewness of the intensity. A naïve Bayesian classifier uses these features to classify the images, achieving sensitivities as high as 99.5% for a data set containing 1714 images. To generate confidence levels, multiple classifiers are combined in three possible ways: a sequential series of classifiers, a vote-taking scheme of classifiers, and a network of classifiers tuned to detect particular types of abnormalities. The third method offers sensitivities of 99.85% or higher with specificities above 60%, making it an ideal candidate for pre-screening images. Two confidence level measures are developed: first, a real confidence level measures the true probability that an image was suspicious; and second, a normalized confidence level assumes that normal and suspicious images were equally likely to occur. The second confidence measure allows for more flexibility and could be combined with other factors, such as patient age and family history, to give a better true confidence level than assuming a uniform incidence rate. The system achieves sensitivities exceeding those in other current approaches while maintaining reasonable specificity, especially for the sequential series of classifiers and for the network of tuned classifiers.

Breast cancer is the most commonly diagnosed cancer among Canadian women; x-ray mammography is the leading screening technique for early detection. This work introduces a semi-automated technique for analyzing mammographic x-ray images to measure their degree of suspiciousness for containing abnormalities. The designed system applies the discrete wavelet transform to parse the images and extracts statistical features that characterize an images content, such as the mean intensity and the skewness of the intensity. A naïve Bayesian classifier uses these features to classify the images, achieving sensitivities as high as 99.5% for a data set containing 1714 images. To generate confidence levels, multiple classifiers are combined in three possible ways: a sequential series of classifiers, a vote-taking scheme of classifiers, and a network of classifiers tuned to detect particular types of abnormalities. The third method offers sensitivities of 99.85% or higher with specificities above 60%, making it an ideal candidate for pre-screening images. Two confidence level measures are developed: first, a real confidence level measures the true probability that an image was suspicious; and second, a normalized confidence level assumes that normal and suspicious images were equally likely to occur. The second confidence measure allows for more flexibility and could be combined with other factors, such as patient age and family history, to give a better true confidence level than assuming a uniform incidence rate. The system achieves sensitivities exceeding those in other current approaches while maintaining reasonable specificity, especially for the sequential series of classifiers and for the network of tuned classifiers.

No description available.

Un des buts de cette thèse est d’approfondir la compréhension de la dynamique non-linéaire, notamment celle des MEMS, en proposant de nouvelles méthodes d’analyse paramétrique et de calcul de modes normaux non-linéaires. Dans une première partie, les méthodes de détection, de localisation et de suivi de points de bifurcation selon un unique paramètre sont rappelées. Ensuite, une nouvelle méthode d’analyse multiparamétrique basée sur la continuation récursive d’extremums est présentée. Cette méthode est ensuite appliquée à un absorbeur de vibration non-linéaire afin de repousser l’apparition de solutions isolées. Deuxièmement, une méthode de calcul de modes normaux non-linéaires est présentée. Une condition de phase optimale et une régularisation de l’équation de mouvement sont proposées afin d’obtenir une méthode de continuation plus robuste au niveau des interactions modales. Ensuite, un problème quadratique aux valeurs propres modifié pour le calcul de stabilité et de points de bifurcation est présenté. Finalement, le calcul de modes normaux non-linéaires a été étendu aux systèmes non-conservatifs permettant la continuation des résonances d’énergie en déplacement et des résonances de phase. Troisièmement, la dynamique non-linéaire de réseaux de MEMS basé sur plusieurs micro-poutres résonantes est analysée à l’aide des méthodes proposées. Tout d'abord, un phénomène de synchronisation de points de bifurcations dû au couplage électrostatique dans les réseaux de MEMS est expliqué. Puis, la dynamique non-linéaire d'un réseau dissymétrisé par l'ajout d'une petite masse sur une micro-poutre est analysée. Enfin, des mécanismes de détection de masse exploitant ces phénomènes non-linéaires sont présentés. / One of the goals of this thesis is to enhance the comprehension of nonlinear dynamics, especially MEMS nonlinear dynamics, by proposing new methods for parametric analysis and for nonlinear normal modes computation. In a first part, methods for the detection, the localization and the tracking of bifurcation points with respect to a single parameter are recalled. Then, a new method for parametric analysis, based on recursive continuation of extremum, is presented. This method is then applied to a Nonlinear Tuned Vibration Absorber in order to push isolated solutions at higher amplitude of forcing. Secondly, a method is presented for the computation of nonlinear normal modes. An optimal phase condition and a relaxation of the equation of motion are proposed to obtain a continuation method able to handle modal interactions. Then, a quadratic eigenvalue problem is shifted to compute the stability and bifurcation points. Finally, nonlinear normal modes are extended to non-conservatives systems permitting the continuation of phase and energy resonances. Thirdly, the nonlinear dynamics of MEMS array, based on multiple resonant micro-beams, is analyzed with the help of the proposed methods. A frequency synchronization of bifurcation points due to the electrostatic coupling is discovered. Then, the nonlinear dynamics of a MEMS array after symmetry breaking event induced by the addition of a small mass onto one of the beam of the array is analyzed. Finally, mass detection mechanisms exploiting the discovered phenomena are presented.