Breast Cancer Detection using Machine Learning

Abstract

Breast cancer is the most frequent cancer in women and the major source of death in women worldwide. Ultrasound imaging is the recommended tool for breast cancer diagnosis in hospitals since it is significantly safer than other imaging modalities. Ultrasound images, on the other hand, are distorted by non-Gaussian regions having additive noise. Medical technicians and clinicians currently diagnose breast cancer by manually reviewing ultrasound images, which is a time-consuming and costly method. This could be a major hurdle to early detection of breast cancer. As a result, early detection of breast cancer can help with not only prescribing medical procedures to prevent the cancer from spreading but also lowering the fatality rate. Automatic detection and diagnosis in ultrasonography is exceedingly difficult due to speckles (noise). In this research, a Convolutional Neural Network (CNN) model for debuler ultrasound pictures is suggested, followed by different CNN model for ultrasound image classification into being and virulent classifications. The proposed models are tested on a Mendeley Breast Ultrasound dataset. Experiments show that the suggested model achieves a classification accuracy of 99.89 percent and that the proposed model(s) surpass other methods proposed in previous publications.