11:00
Session 7: Inverse Problems
Chair: Valia Guerra
11:00
20 mins
|
TOWARDS 2D AND 3D IMAGING OF MAGNETIC NANOPARTICLES USING EPR MEASUREMENTS
Annelies Coene, Guillaume Crevecoeur, Luc Dupré
Abstract: Electron Paramagnetic Resonance (EPR) is a promising technique for visualizing the distribution of magnetic nanoparticles (MNP) non-invasively. Currently EPR is only able to recover 1-dimensional (1D) MNP distributions. In this paper we extend 1D EPR towards 2D and 3D. We solve the associated inverse problem by Truncated Singular Value Decomposition (TSVD), Non-Negative Least Squares and a combination of both algorithms. Furthermore, we investigate the impact of noise on the reconstruction results.
|
11:20
20 mins
|
A RECONSTRUCTION OF DIELECTRIC OBJECTS BURIED UNDER A ROUGH SURFACE
Özgür Özdemir, Yasemin Altuncu
Abstract: We propose a computationally efficient method to reconstruct dielectric objects buried in a half-space with rough surface. Cauchy Data is exploited to formulate the total field inside the homogeneous neighborhood of the scatterers using homogeneous Green’s function instead of background Green’s function. Then Born-type linearization is employed to reconstruct the object from measured scattered field. The feasibility and efficiency of the method are demonstrated through numerical experiments.
|
11:40
20 mins
|
UNDERDETERMINED MAGNETOSTATIC INVERSE PROBLEM: BAYES THEOREM APPLICATION
Olivier Pinaud, Jean-Louis Coulomb, Olivier Chadebec, Laure-Line Rouve, Jean-Michel Guichon
Abstract: This paper deals with the use of the Bayesian approach to inverse an underdetermined problem. The modeled a priori information is statistically studied using either Unscented Transform (UT) method or Monte Carlo (MC) method to estimate the mean value and standard deviation. The whole approach is applied to an experimental setup illustrating an electric vehicle power circuitry. This demonstrates the strength of coupling modeling and measurements for complex systems identification.
|
12:00
20 mins
|
CURRENT DENSITY RECONSTRUCTION FOR INVERSE CALCULATIONS OF DEFECTS IN LORENTZ FORCE EVALUATION
Judith Mengelkamp, Konstantin Porzig, Matthias Carlstedt, Marek Ziolkowski, Hartmut Brauer, Jens Haueisen
Abstract: We apply the method of current density reconstruction to perform inverse calculations of defects in Lorentz Force Evaluation (LFE). Superior to previously introduced conductivity estimation, this approach is not limited to laminated materials. We apply the method to solid bodies performing the inversion in a fully three-dimensional source space. The results of defect reconstructions in solid and laminated specimen show comparable and acceptable errors. Thus, current density reconstruction is a promising method for LFE.
|
12:20
20 mins
|
EXPERIMENTAL SENSITIVITY ANALYSIS OF MAGNETORELAXOMETRIC IMAGING
Maik Liebl, Uwe Steinhoff, Frank Wiekhorst, Daniel Baumgarten, Jens Haueisen, Lutz Trahms
Abstract: In magnetorelaxometry (MRX) the amount of magnetic nanoparticles (MNP) in a sample can be quantified by measuring the amplitude of its time-delayed response to a sudden change of an external magnetizing field. For the quantitative imaging of MNP distributions the MRX method is expanded by a sequential application of multiple spatially distinct magnetizing fields to the sample. However, each magnetizing step prolongs the measurement duration. Here, we experimentally investigate the potential of reducing the number of magnetization steps. In a fixed measurement setup using a 304 channel magnetometer we varied crucial imaging parameters (distance of the sample to coils and sensors, number of coils and sensors and the excitation current). We found even a slight improvement of reconstruction quality by reducing the number of magnetization steps. Amplitude variation of a constant magnetizing current had no influence, while combining different magnetizing amplitudes improved the reconstruction.
|
|
