Cancer is a lethal disease produced by the aggregation of hereditary disorders and a variety of pathological alterations. Cancerous cells are life-threatening abnormal regions that can grow in any portion of the human body. Cancer is also known as a tumour that must be diagnosed swiftly and accurately in the early stages in order to determine what treatment options are available. Despite the fact that each modality has its own set of problems, such as a difficult history, incorrect diagnoses, and therapy, which are all major causes of death. The goal of the study is to examine, review, evaluate, and discuss current breakthroughs in human body cancer detection utilising machine learning approaches for breast, brain, lung, liver, and skin cancers, as well as leukemia. The study shows how machine learning with supervised, unsupervised, and deep learning techniques can help in cancer diagnosis and cure. Several state-of-the-art approaches are grouped together, and findings from accuracy, sensitivity, specificity, and false-positive metrics are compared on benchmark datasets. Finally, potential future work is indicated by highlighting obstacles.