ABSTRACT
Oncology research needs a huge quantity of images and test reports for diagnosing. In vitro methods like computed tomography and magnetic resonance imaging, invivo methods like a biopsy are adopted to diagnose cancer patients. This medical proof is the basis for cancer treatment. But this test proof is accompanied by more than one trouble in exploiting and finding information for predicting sickness dynamics. Due to this, almost 5% of outpatients acquire the incorrect diagnosis each year. One in every three of these misdiagnoses causes substantial patient loss. To combat this, researchers have focused significantly on classical mathematical analysis to discover statistical correlations in physiological and pathological data associated with a cancer diagnosis. Even though mathematical models provide a perfect summary of a realistic context, including its underlying system dynamics, they may not always allow for precise and measurable forecasts. Modern Machine Learning (ML) method has a high level of assurance that, with enough training, it can provide significant forecasting ability on the patient in real-time. ML methodology enables experts to examine enormous amounts of data in a structured manner and deduce underlying knowledge of complex systems. Modern ML methods are adopted to forecast the growth of tumors which is under study and helps to detect new malignancies. Machine learning-based approach in oncology created an impression among clinicians in selecting perfect, target treatment, and clinical outcome measures. To improve the outcome accuracy, the researcher should shed more light to upgrade the machine learning algorithm without compromising the clinician's needs. This chapter discusses the different strategies adopted by the researchers to enhance the result accuracy in classifying cancer in patients.
