Design of Antipodal Vivaldi Antenna with CNN-based Classification for Non-Invasive Breast Cancer Detection

Breast cancer remains a major health issue around the world. This work presents a non-invasive method for detecting breast cancer using a specially designed dual-band Antipodal Vivaldi Antenna, combined with a deep learning classifier. The antenna, fabricated on flexible Rogers RT/duroid 5880 and operating at 7.9 GHz and 11.2 GHz, generates distinct electromagnetic signatures that separate healthy, benign, and malignant tissues. These signatures were used to train a Convolutional Neural Network, which achieved 91% classification accuracy. Specific Absorption Rate (SAR) analysis further confirmed safety and spatial localization of tumors. Together, this hybrid antenna–CNN approach provides a promising low-cost, radiation-free tool for real-time breast cancer screening.