Analysing different modalities of biomedical data leads to better diagnostic and prognostic models of disease. For this reason, modern electronic health record applications try to incorporate every kind of information, leading to holistic records. Most openly available data today, however, come from specific examinations of a disease and provide results of a single type (unimodal). This thesis focuses on fusing clinical, imaging and non-imaging data (for example, biopsies) on the same disease but from different patients. After reviewing the state-of-the-art methods in the field, it will investigate fusion approaches and propose methods for processing such data to extract composite features that lead to better diagnostic and prognostic indicators.
