Detection of Breast Cancer from Ultrasound Imaging using Deep Learning Model

Abstract

The second most common cause of mortality for women is breast cancer. Female death rates can be decreased if breast cancer is found early. For early cancer detection, an automated system is needed because manual breast cancer diagnosis takes a long time. There is a 30% possibility that the disease can be treated with early identification, but late detection of advanced-stage malignancies makes therapy more challenging [1,2]. Using Deep learning, we created a model that can predict the likelihood of getting breast cancer. In this paper, deep learning models are used to provide a new framework for detecting breast cancer from ultrasound images. Images from the Breast Ultrasound Dataset are divided into three categories: normal, benign, and malignant. In order to increase the amount of the original dataset and improve Convolutional Neural Network (CNN) model learning, data augmentation is carried out. Uses of Model: VGG16, InceptionV3, Exception, DenseNet201. We used these 4 models in deep learning, among which the accuracy of inception is the best and the accuracy value is 88%.