The relationship between image biomarkers in structural MRI and knee osteoarthritis pain progression is investigated. A Bayesian Gaussian mixture model is selected to identify the distinct knee pain trajectories among subjects in the dataset from the Osteoarthritis Initiative. Deep learning is employed to predict the probability of an individual’s pain curve cluster membership using the 3D structural MRI. Utilizing the strength of the model-based approach, the pain curves are simulated from the GMM posterior probabilities and the weights learned to evaluate the 3D DenseNet’s performance.
This abstract and the presentation materials are available to members only; a login is required.