Home » Key Scientific Articles » RF breast cancer detection employing a noncharacterized vivaldi antenna and a MUSIC-inspired algorithm

RF breast cancer detection employing a noncharacterized vivaldi antenna and a MUSIC-inspired algorithm

Significance Statement

 

Breast cancer is the most common cancer in women. Mass population screening campaigns use X-ray mammography but it is limited by weak contrast between benign/malignant tissues:  between 4% and 34% of all breast cancers are missed and nearly 70% of all breast lesions turn out to be benign. In addition, the technique requires breast compression which leaves adjacent chest wall areas unscreened. Furthermore, the exposure to ionizing radiation may reduce compliance for patients with screening recommendations. Microwave technology offers higher dielectric contrast between normal and diseased breast tissues which can supplement conventional diagnostics. A growing research community is exploiting the technique to image refined features of biological targets. False diagnosis can be reduced significantly to improve health system efficiencies. The microwave technology is based on low-power non-ionizing radiation which facilitates cost-competitive screenings aimed at detecting, localizing and categorizing tumors in breast tissues. While a contrast factor of up to 10 exists between malignant and adipose breast tissues, it can be as low as 0.1 between malignant and healthy fibroglandular tissues. This remains a research challenge for microwave breast imaging since breast tumors typically appear in the fibroglandular tissues. Microwave imaging techniques can be generally grouped into non-linear and linear inverse problem solving. In each case, the antenna is typically in close proximity with the breast which gives rise to difficult a priori characterization of the microwave source/sensor system. Multiple Signal Classification (MUSIC) algorithms have mitigated the requirement for antenna characterization in complex near-field scenarios. A proprietary Interferometric-MUSIC algorithm adapted for a multi-monostatic configuration was proposed as a multi-frequency variant of the well-known time-reversal MUSIC. The method offers improved focusing capability with a larger dynamic range between clutter and tumor levels when compared to the other non-coherent algorithms. The algorithm is suited to integration with complex antenna geometries without compensatory tailoring given that a priori antenna characterization is no longer required. In this paper, an antipodal planar Vivaldi antenna was selected for its compactness, directive behavior and broad impedance matching properties. The design was aided by a computationally efficient multiobjective evolutionary algorithm ParEGO that optimizes the spline-based contour of the arms for an 8-dB return loss bandwidth between 1 – 3 GHz. Combined with antipodal Vivaldi antennas, the Interferometric-MUSIC algorithm enables more refined mitigation of artefact-clutter in reconstructed images which allows for earlier-stage cancer diagnosis.

Figure Legend: Interferometric-MUSIC detection. a) Cross-section of a numerical MRI-based breast phantom (tumor marked in black); b) Reconstructed image.

 

RF breast cancer detection employing a noncharacterized vivaldi antenna and a MUSIC-inspired algorithm

 

 

 

 

 

 

Journal Reference

Giuseppe Ruvio, Raffaele Solimene, Antonietta D’Alterio, Max J. Ammann,Rocco Pierri. International Journal of RF and Microwave Computer-Aided Engineering, Volume 23, Issue 5, pages 598–609,  2013.

 Abstract

A novel microwave breast cancer detection system consisting of an Evolutionary Global Optimized Vivaldi antenna and an algorithm inspired by MUltiple SIgnal Classification (MUSIC) is presented. Its performance is assessed by using a simplified numerical breast phantom for a number of critical conditions including the presence of fibroglandular tissues.

Copyright © 2012 Wiley Periodicals, Inc.

Go To Journal