Application of Machine Learning in Breast Cancer Diagnosis: A Review

The aim of this paper is to summarize recent literatures which focus on the application of artificial intelligence, specifically machine learning, for detecting breast cancer. Breast cancer is the second most invasive disease in women and can be treated efficiently if detected in the early stages. The process of diagnosing is inclusive of two standard procedures, namely image screening and classification of tumors using medical records by radiologists. Mammography is considered to be the most suitable technique for screening images in regular clinical practice. Computer-Aided Diagnosis (CAD) systems are developed to aid detection of the disease and save cost-time invested in the process. With the help of deep learning, which is a subset of machine learning, the procedure of classifying can be automated efficiently to avoid misdiagnosis due to distraction, fatigue, or lack of experience, thus increasing accuracy. Models and techniques are reviewed objectively in this paper to suggest that a convolutional neural network delivers satisfactory results in the classification of mammographic images. In addition, the paper also studies the limitations of CAD systems based on different methods, which are still not completely automated irrespective of the higher accuracy rates of the model. Finally, the scope of improvement is discussed to encourage research in the respective field with the aim to automate CAD systems completely. Thus, recent literature is reviewed to prevent heterogeneous evaluation of the study.