International Journal of Research and Innovation in Applied Science (IJRIAS) | Volume VII, Issue VIII, August 2022 | ISSN 2454–6194
Machine Learning in Healthcare: Breast Cancer Detection Using Graph Convolutional Network
Sadah Anjum Shanto, Gourab Kanti Paul, Romzan Ali Mohon, Satyajit Sarker, Marjana Sariat Mahir, Syed Sanaul Haque
Department of Computer Science and Engineering, Bangladesh University of Business and Technology, Dhaka, Bangladesh
Abstract:
Machine learning is a vast field of research. The idea is to build a machine learning-based cad system for breast cancer detection using mammogram image data. At first, we use supervised classification techniques in our mammogram image data and then feed the classified data into the GCN model for detection. We investigated that the GCN model can give better accuracy than traditional machine learning models. Breast cancer is one of the most common cancers that women suffer the most. But breast cancer can be detected early. The vast amount of research shows that if breast cancer is successfully detected early, the patient life can be 99% saved early. A screening mammogram is the other most useful thing in the detection of breast cancer. According to researchers, with the help of mammograms breast cancer can be detected three years earlier before the start of cancer symptoms. Graph Convolutional Neural Network (GCN) is a new field of convolutional machine learning. Unlike CNN, GCN follows a non-Euclidian approach which can show better results in image classification. We aim to investigate the GCN model into breast cancer mammogram image data, that it can give better accuracy than traditional machine learning models. After evaluating our proposed GCN algorithm to four others, we discovered that GCN achieved the accuracy of 81 percent.
Keywords: Graph Convolutional Network Graph (GCN), Machine Learning, Breast Cancer, Mammogram, Graph Neural Network, Classification, Morphological Operation, Contrast limited adaptive histogram equalization
I. INTRODUCTION
For centuries, cancer has afflicted us. According to the International Agency for Research on Cancer (IARC), a part of the World Health Organization (WHO), there were 4.5 million deaths caused by cancer in 2016 [1]. Among the cancer types, Breast Cancer is the most common cancer type in women. It is the second leading cause of cancer death excluding lung cancer. Early detection of breast cancer can reduce its death rate. For the past few decades, Machine learning techniques have been used in various fields of healthcare worldwide. Afterward advances in picture handling and machine learning techniques allow building Computer-Aided Detection/Diagnosis (CAD / CAD) frameworks that can offer assistance specialists to be more productive, objective, and unfaltering within the assurance [2]. The mammogram is broadly utilized inside the early screening of breast cancer due to its by and large fetched and tall affectability to minor tumors. Within the real determination handle, be that as it may, the exactness can be contrarily influenced by numerous components, such as radiologist