In this talk, we shall discuss machine learning techniques for the characterization of biomedical images in MRI and CT in the context of two clinical applications. The first clinical application investigates methods for assessing diagnostic adequacy of liver images in MRI using traditional machine learning and deep neural networks. Proposed algorithms are compared across varying sample sizes to determine advantages of either approach. The second application investigates the relationship between lung structure using CT images and lung function using pulmonary function tests (PFTs) in the context of COPD. Two pathophysiological imaging features extracted from inspiratory and expiratory CT images were found to accurately predict PFT-based COPD diagnosis, suggesting imaging-based staging can be used as a screening tool for COPD.