Detection of Breast Cancer using Deep Learning Approach

Abstract

Breast cancer is a disease in which cells in the breast grow out of control. In Bangladesh the rate of breast cancer occurrence is estimated to be 22.5 per 100,000 females of all ages. In case of Bangladeshi women, aged between 15-44 years, breast cancer has the highest prevalence 19.3 per 100,000 compared to any other type of cancer. Breast cancer can occur in both men and women, but it's far more common in women. There are many researches who have worked on detection of breast cancer using deep learning techniques. But their works are dedicated to one particular dataset. In this paper I try to detect the breast cancer by using a deep learning techniques Convolutional Neural Network (CNN). I also use three pre-trained CNN models VGG-16, Inception V3 & ResNet50 to compare with my CNN model. In this paper I worked with two datasets. One BreaKHis dataset (Histopathology Image) & the another one (MRI Image) is taken from Kaggle. Aim of this paper is to detect the breast cancer by using deep learning model with maximum accuracy. The experimental results show that for histopathology image VGG-16 gives the highest accuracy of 99.32% and for the MRI image Inception V3 gives the highest accuracy of 95.36%. All experiments are executed within Jupyter Notebook, Google Colaboratory and PyCharm platform.