Biomedical qualitative imaging by means of a two-step inverse scattering method

Abstract : X-rays imaging and magnetic resonance imaging (MRI) are currently reference methodologies for the non-invasive inspection procedures of biological bodies. In such a frame- work, microwave methodologies have recently emerged as suitable tools for biomedical diagnosis because of (a) the non-ionizing nature of microwave, thus the possibility to minimize the collateral effects; (b) the sensibility of microwave to the water content as well as to the contrast variations that exist between malignant and other tissues. In addition, microwave methodologies seem to be less expensive than MRI, as well. Thus, such approaches can play an important role. As far as microwave inverse scattering techniques are concerned, some qualitative imaging approaches have been recently developed in order to profitably exploit the amount of a-priori information on the problem at hand usually available. As an example, some approaches that approximate "contrast" regions (i.e., spatial regions characterized by dielectric differences with respect to "safe" reference distributions) with rectangular shapes have already been proposed. Despite the satisfactory results, such techniques cannot be reliably employed when an accurate knowledge on the shapes of these regions is needed. Yet, such approximate methods could be profitably used for providing a "first-step" concerned with a rough localization of the "contrast region" to be further improved by means of a successive refinement reconstruction carried out with suitable contour detection methods. This paper presents a two-step procedure aimed at improving the reconstruction of a pathology inside a biological body. More in detail, starting from the knowledge of the scattered field with and without the pathology, the approximate problem where the contrast region is assumed of simple shape (e.g., a rectangle) is reformulated in terms of an inverse scattering one and successively solved by means of the minimization of a suitably-defined cost function. After such a step, the region-of-interest (RoI) is estimated starting from the parameters describing the rough rectangular shape. Then, the second retrieval phase takes place with the application of shape-based optimization technique based on the numerical evolution of a level set function. The mathematical formulation of the proposed approach is presented and the effectiveness of the approach is discussed with reference to a set of representative numerical results in dealing with pathologies within biomedical bodies with noiseless as well as blurred data.
Complete list of metadatas
Contributor : Dominique Lesselier <>
Submitted on : Friday, November 4, 2011 - 2:41:12 PM
Last modification on : Friday, September 28, 2018 - 1:16:37 AM


  • HAL Id : hal-00638282, version 1



Manuel Benedetti, Massimo Donelli, Dominique Lesselier, Andrea Massa. Biomedical qualitative imaging by means of a two-step inverse scattering method. Progress in Electromagnetics Research Symposium, Mar 2008, Hangzhou, China. pp.548. ⟨hal-00638282⟩



Record views