Breast Cancer Classification using Convolutional Neural Networks (CNNs)

Abstract

Abstract: In the present world, women are facing many health issues and medical problems. Breast cancer is one among them. It is a common cancer in women, and one of the major causes of death among women around the world. The Invasive Ductal Carcinoma (IDC) is the most widespread type of breast cancer with about 80% of all diagnosed cases.

The IDC is characterized by hard lumps with asymmetrical borders. The Invasive breast cancers spread from the origin into the adjoining breast tissue. On a mammogram, IDC typically appears like a mass with spikes radiating from the edges. Early accurate diagnosis plays an important role in choosing the right treatment plan and improving survival rate among the patients.

However, due to the small size and low contrast (of lumps?) compared to the background of images, it is challenging and time-consuming for radiologists to make an independent and accurate assessment. Hence, there is a necessity to develop helpful automated tools to overcome these obstacles in the diagnostic performance of breast cancer.

The proposed system is a breast cancer classifier on an IDC dataset that can accurately classify a histology image as benign or malignant using Artificial Intelligence. This design is implemented by using an image classification technique with the help of Deep Learning using six layered Convolutional Neural Network (CNN) architecture to identify the breast cancer. This design is tested by using different Machine Learning Algorithms like, Random Forest, Gradient Boosting, Extra Trees, and Logistic Regression for comparative analysis in terms of accuracy.

Index Terms: Convolutional Neural Network, Cancer, Machine Learning Algorithm, Breast Cancer, Deep Learning