Deep Learning Algorithm for Automatic Breast Tumour Detection and Classification from Electromagnetic Scattering Data
K. Lalitha, Shreya Reddy K · Progress In Electromagnetics Research C · 2022
Breast cancer is, by far, the most diagnosed disease for the death of women worldwide.Researchers are working with an alternative technology to detect tumors before they reach a terrible stage because of the numerous limitations in the current imaging approach.This article suggests a promising technique by utilising non-ionizing microwave signal and artificial intelligence especially deep learning algorithms for early detection of breast cancer.This contribution will present a method to detect and classify the tumor category using backscatter signals obtained from antenna simulation in CST microwave studio software.The post-processed scattering parameters are utilized to create image through MATLAB programming environment.The high intensity in the image represents the precise position of tumor.The automatic classification of tumor is achieved by YOLOv 5 deep learning model from the created microwave images.A training dataset with fifty image samples are formed by preprocessing, and then augmentation is applied to create final dataset with 1000 samples.This approach can identify the location and type of early-stage tumor with size of 5 mm.