Microwave-based breast cancer detection using a high-gain Vivaldi antenna and metasurface neural network approach for medical diagnostics

Abstract This paper presents a novel technique for detecting tumors in human breasts using a single high-gain antenna and a metasurface (MTS) layer. An artificial neural network (ANN) is employed to classify detected tumors as benign or malignant based on the permittivity of the tissue. The detection and classification process leverages the contrast in dielectric properties between normal and abnormal biological tissue, utilizing the actual permittivity as a distinguishing factor. This study highlights the effectiveness of the proposed technique in accurately detecting and localizing malignant tumors within human breasts. Electromagnetic analysis is conducted using voxel datasets derived from human models to validate the approach. Tumor localization is achieved with high precision based on the Specific Absorption Rate (SAR) magnitude. The study considers various fat layer thicknesses (10–100 mm) and tumor radii (2.5–10 mm), addressing scattering effects comparable to the wavelength of the applied microwave radiation. The proposed Vivaldi antenna operates at 3.5 GHz, achieving a gain of 15.5 dBi with a half-power beamwidth in the E-plane of ±12°. Results demonstrate minimal average errors and high-performance indices (PI) for fat thickness (0.1 %, 90 %), tumor size (0.06 %, 94 %), and tumor classification (0.11 %, 89 %). The experimental and simulation results exhibit strong agreement, confirming the feasibility and potential of the proposed antenna system for medical diagnostics and post-detection rehabilitation planning.