Cancer Detection by Machine Learning

Abstract— Cancer is a terminal condition, often caused by the aggregation of genetic defects and several disease modifications. Cancer cells are pathological regions that are often life-threatening in some portion of the human body. At the early stage, cancer, also known as the tumor, must be diagnosed easily and properly to determine what may potentially treat it. However, modalities are different, such as complicated history, inaccurate diagnostics and care, which are the major causes of death. The study aims to review, assess, classify, and discuss existing advances to diagnose human body cancer by breast, brain, lung, liver, skin cancer, and leukemia machine learning methods. The research demonstrates how machine learning with supervised, unmonitored, and deep learning approaches helps cancer diagnosis and the curing process. Various state-of-the-art methods are classified under the same class, and outcomes from precision, sensitivity, specificity, and false positives metrics. Metrics are compared on benchmark data sets. In conclusion, obstacles for potential future work are also outlined. Index Terms—Artificial Intelligence, Biomedical Image Analysis, Cancer diagnosis, Deep neural network, Machine Learning