Detection of Early-Stage Breast Cancer using Electromagnetic Sensor with Aid of the Convolution Neural Network

Abstract

Objectives: This study aimed to develop an end-to-end system for the diagnosis of breast cancer using a novel combination of a monopole electromagnetic sensor and Convolutional Neural Network (CNN). Methods: The research involved the design and simulation of an electromagnetic sensor, utilizing a denim gene substrate, to capture dielectric changes within breast tissue across a broad spectrum (1GHz to 10GHz). The recorded data was processed by a pre-trained CNN to identify irregularities in the breast's internal structure. Findings: Through extensive simulations, the electromagnetic sensor displayed a remarkable sensitivity to changes in the dielectric properties of breast tissue. The CNN analysis accurately identified the presence of cancer cells and estimated tumor size with an impressive 98% accuracy and a 1% tolerance margin. This method significantly outperformed existing models in both accuracy and efficiency, reducing the need for costly imaging techniques. Novelty: This research offers a non-invasive, cost-effective solution for early-stage breast cancer detection. Unlike traditional imaging techniques, this approach provides accurate diagnostics without the need for extensive equipment or high-cost procedures.