Improving UWB Image Reconstruction for Breast Cancer Diagnosis by Doing an Iterative Analysis of Radar Signals

Torres-Quispe H.A.; Patino-Escarcina R.E. · Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (2022) · DOI: 10.1007/978-3-031-09037-0_36

Resumen

UWB (Ultra-Wideband) radar technology is based on the general principle that an antenna transmits an electromagnetic signal and the echo of the reflected signal is detected. This technology is used in different applications, such as medical monitoring or imaging applications for a more precise diagnosis as breast cancer. The diagnostic process of it consists of detecting any lesion or abnormality in the breast tissue, for this, imaging techniques of the breast are used. This UWB technology has given good results compared to traditional methods that lack effectiveness, producing false positives. The images obtained by microwave signals have electrical properties according to the different tissues. Comparing healthy tissues with malignant tissues, a contrast of 8 % in permittivity and 10 % in conductivity was found in research on the dielectric properties of breast tissue. Previous research proposed reconstruction algorithms for the detection of breast tumors based on a maximum likelihood expectation maximization (MLEM) algorithm. This work proposes an iterative algorithm for imaging based on traditional delay-and-sum (DAS) and delay-multiply-and-sum (DMAS) methods. The main characteristic of this work is the elaboration of an algorithm based on MLEM for a bistatic radar system, to reconstruct an enhanced image where possible tumors with a diameter of up to 1 cm are better distinguished. In a bistatic system, signal processing and reconstruction algorithm differs from a monostatic system, because 2 antennas are considered (emitter and receiver). The MLEM algorithm reduces statistical noise over conventional back-projection algorithms. Experiments were performed with data taken in the laboratory with simulated tumors inside a breast phantom. The results show the images where the tumors are highlighted with 4 iterations.

Instituciones: Universidad Católica San Pablo