With large number of radiomic features available with limited size datasets, especially in medical applications, feature selection and qualification become a prerequisite to improve performance in terms of accuracy, robustness, and reproducibility. The applications of radiomics include but not limited to diagnosis, treatment planning, prognosis in patient’s health care. Considering the diverse applications, two questions will be answered in this chapter: what types of radiomic features are available to be extracted and what kind of techniques can be used to perform the feature selection. Static features including morphology and statistical (first-order, second-order) features and dynamic features based on kinetic analysis of dynamic images will be discussed. In addition, “featureless” approach—based on deep learning with convolutional neural networks (CNN) will also be presented. With respect to feature selection methods, filter, wrapper, and embedded methods are covered, some other machine learning methods, such as principle component analysis (PCA), clustering, neural networks (NN), ensemble feature ranking methods, and graph-based methods will be presented as well.