Knee osteoarthritis (OA), a leading contributor to physical disability and chronic pain, is associated with poor health outcomes and has a significant socio-economic impact. Bone marrow lesions (BMLs) often occur prior to irreversible cartilage degeneration, and earlier detection of them may allow identification of individuals at high risk of knee OA. The overall goal of the project is to develop accurate and fast machine methods to automatically quantify BMLs from magnetic resonance imaging (MRI) data.