Eduardo Costa da Silva

Modelagem da sensibilidade de amostras GMI por redes neurais

Over the past few years, several studies have been developed in order to quantitatively model the GMI effect (Giant Magnetoimpedance). However, these models adopt simplifications that significantly affect its theoretical-experimental performance and its generalization capability, and models that incorporate parameters that generate asymmetry - AGMI (asymmetric GMI) - such as the DC level of the excitation current of the GMI samples are still rare. This work aims to develop a new model, sufficiently general, which also incorporates the asymmetry induced by the DC level of the excitation current, capable of guiding the experimental procedures of characterization of the GMI samples. Thus, this paper proposes, presents and discusses the use of a computational model based on feedforward Multilayer Perceptron Neural Networks to model the impedance magnitude sensitivity and impedance phase sensitivity, of the GMI effect, as functions of the magnetic field, for Co70Fe5Si15B10 ferromagnetic amorphous alloys. The proposed model allows obtaining these sensitivities based on some of the main parameters that affect it: length of the samples, DC level and frequency of the excitation current and the external magnetic field.

Biomedical comparison of magnetometers for non-ferromagnetic metallic foreign body detection

The location and surgical removal of foreign bodies in patients is still challenging, especially for firearm projectiles, which are small and non-ferromagnetic. Conventional location techniques use ionizing radiation, posing health risks while the procedures often last several hours and end unsuccessfully. The use of high sensitivity magnetometers provides a noninvasive and innocuous alternative for metallic foreign body location. The developed technique consists of a primary AC magnetic field generator (a solenoid) inducing eddy currents in nonferromagnetic metallic foreign bodies, which results in an ultra-low secondary magnetic field that can be measured. This work compares the initially developed theoretical technique using Superconducting Quantum Interference Device (SQUID) magnetometers with the developed prototypes using lower cost alternatives, namely Giant Magnetoresistance (GMR) and Giant Magnetoimpedance (GMI). The comparison is based on biomedical device requirements for widespread clinical application. The proposed GMI location system is deemed the most qualified

Multi-parameter fuzzy design space for QbD approach applied in the development of biomedical devices

Since 2004, on the recommendation of the Food and Drug Administration (FDA), the concept of Quality by Design (QbD) has been widely applied by the pharmaceutical industry in drug development. The paradigm for quality assurance has been changed from inspections tests to focusing on increasing control within manufacturing processes. Recently, the methodology of QbD has gained space in the area of development of transducers for biomedical application. In order to advance on the adoption of this approach to the characteristics and needs of this new field, a methodology of representation of a multiparameter Design Space with fuzzy inference was developed.

Application of genetic algorithms to the solution of the biomagnetic inverse problem, using data acquired by a 16-Channel SQUID system

This paper proposes a new technique based on the employment of genetic algorithms to solve the inverse biomagnetic problem. The biomagnetic measurements analyzed in this paper were performed on isolated rabbit hearts and acquired by a 16-Channel SQUID system. The developed method focused on defining the single equivalent current dipole that best fits the experimental data. The genetic algorithm is used to determine the position, angle and magnitude of the current dipole, attempting to minimize the error between the experimental magnetic field map and the one obtained by using the estimated current dipole.

Enhanced Neuro-Genetic model aimed at optimizing the sensitivity of GMI sensors, subjected to linearity and span constraints

GMI magnetometers are magnetic transducers based on Giant Magnetoimpedance sensor elements, which are ferromagnetic samples that present a large variation in their impedance as a function of the external magnetic field. The sensitivity of magnetic transducers is directly dependent on the sensitivity of their sensor elements. Consequently, one can optimize the overall sensitivity of the transducer by maximizing the sensitivity of their sensor elements, which is essential for ultra-weak magnetic fields measurements. In the case of GMI samples, the sensitivity is affected by several parameters, such as: sample length, external magnetic field, DC level and frequency of the excitation current; bias magnetic field, etc. Currently, this dependency is yet to be well modeled quantitatively, in function of all parameters that affect it. Thus, the search for the optimum point is usually empirical. Besides maximizing the sensitivity, it is imperative to ensure that this sensitivity is held almost constant under the desired sensor span. In other words, instead of purely maximizing the sensitivity, it is important to take into account the linearity either. Then, following that goal, this work aims at developing a neuro-genetic model capable of fitting the sensitivity of GMI samples and defining the set of parameters that lead to the optimal sensitivity, under a predefined sensor span. The proposed computational model is based on a MLP Neural Networks, for modeling the sensitivity of the GMI samples, and a Genetic Algorithm, responsible for determining the combination of parameters that allow maximizing the sensitivity, considering the imposed restrictions.